MiniMax Group Inc. (MiniMax集团股份有限公司) IPO Prospectus

# MiniMax Group Inc.

(A company controlled through weighted voting rights and incorporated in the Cayman Islands with limited liability)

Stock Code: 0100

GLOBAL OFFERING Joint Sponsors, Overall Coordinators, Joint Global Coordinators, Joint Bookrunners and Joint Lead Managers (in alphabetical order)

Overall Coordinators, Joint Global Coordinators, Joint Bookrunners and Joint Lead Managers (in alphabetical order)

IMPORTANT IMPORTANT: If you are in any doubt about any of the contents of this Prospectus, you should seek independent professional advice.

MiniMax Group Inc. (A company controlled through weighted voting rights and incorporated in the Cayman Islands with limited liability)

GLOBAL OFFERING

| Item | Details | |---|---| | Number of Offer Shares under the Global Offering | 25,389,220 Offer Shares (subject to the Offer Size Adjustment Option and the Over-allotment Option) | | Number of Hong Kong Offer Shares | 1,269,480 Offer Shares (subject to reallocation) | | Number of International Offer Shares | 24,119,740 Offer Shares (subject to reallocation, the Offer Size Adjustment Option and the Over-allotment Option) | | Maximum Offer Price | HK$165.00 per Offer Share, plus brokerage of 1%, SFC transaction levy of 0.0027%, Stock Exchange trading fee of 0.00565% and AFRC transaction levy of 0.00015% (payable in full on application in Hong Kong dollars and subject to refund) | | Nominal value | US$0.0001 per Offer Share | | Stock code | 0100 |

Joint Sponsors, Overall Coordinators, Joint Global Coordinators, Joint Bookrunners and Joint Lead Managers (in alphabetical order)

Overall Coordinators, Joint Global Coordinators, Joint Bookrunners and Joint Lead Managers (in alphabetical order)

Joint Bookrunners and Joint Lead Managers (in alphabetical order)

Hong Kong Exchanges and Clearing Limited、香港联合交易所有限公司及香港中央结算有限公司对本招股章程的内容概不负责，对其准确性或完整性亦不发表任何声明，并明确表示，概不对因本招股章程全部或任何部分内容而产生的任何损失承担任何责任。

本招股章程的副本（连同「附录五——送呈香港公司注册处处长存档及供查阅的文件」一节所列明的文件）已按照香港法例第32章《公司（清盘及杂项条文）条例》第342C条的规定，在香港公司注册处处长处登记注册。香港证券及期货事务监察委员会及香港公司注册处处长对本招股章程或上述任何其他文件的内容概不负责。

最终发售价预期由整体协调人（为其本身及代表包销商）与本公司于定价日（预期为2026年1月7日（星期三）或前后）协议厘定。发售价将不超过每股发售股份165.00港元，目前预期不低于每股发售股份151.00港元，另有公告除外。如因任何原因，整体协调人（为其本身及代表包销商）与本公司未能于2026年1月7日（星期三）中午12时前就最终发售价达成协议，全球发售将不会进行并将告失效。

发售股份未曾亦不会根据美国《证券法》或美国任何州证券法律进行登记，且不得在美国境内提呈、出售、质押或转让，惟发售股份可（a）在美国境内仅向合资格机构买家出售，且须依赖规则144A或美国《证券法》登记规定的其他豁免，或进行不受该等登记规定所规限的交易；或（b）在美国境外，依赖S规例进行离岸交易方式提呈或出售或交付。

香港发售股份的申请人于申请时（视乎申请渠道而定）可能须就每股香港发售股份缴付165.00港元的发售价，连同1%经纪佣金、0.0027%证监会交易征费、0.00565%联交所交易费及0.00015%会计及财务汇报局交易征费。

在作出投资决定前，潜在投资者应仔细考虑本招股章程所载的全部资料，包括「风险因素」一节所载的风险因素。

香港包销商根据香港包销协议所承担的责任，可由整体协调人（为其本身及代表香港包销商）在上市日期上午8时前在若干情况出现时予以终止。详情请参阅「包销——包销安排及费用——香港公开发售——终止条件」。

本公司为专业技术公司（定义见上市规则第18C章）。专业技术公司的证券存在较高投资风险，包括股价波动风险及因难以对该等公司进行估值而导致估值偏高的风险。投资者在作出投资决定前，应充分了解专业技术公司的投资风险及本公司披露的风险。此外，本公司为预商业化公司（定义见上市规则第18C章）。预商业化公司为无法符合上市规则第18C.03(4)条所载收益规定的专业技术公司，因此若其于上市后无法获得足够的外部资金及╱或无法产生足够收益以维持营运，则面临更高的公司失败风险。

本公司将于上市后透过加权投票权架构实施控制。潜在投资者应注意投资于具有加权投票权架构的公司所涉及的潜在风险，尤其是加权投票权受益人（其利益未必与本公司全体股东的利益一致）将能够对股东决议案的结果施加重大影响，而不论其他股东如何投票。有关加权投票权架构相关风险的进一步资料，请参阅「风险因素——与加权投票权架构相关的风险」。潜在投资者仅应在经过审慎及仔细考虑后，方可作出投资本公司的决定。

注意

我们已就香港公开发售采用全电子申请程序。我们将不会就香港公开发售向公众提供本招股章程的印刷版本。

本招股章程载于联交所网站www.hkexnews.hk及本公司网站https://www.minimaxi.com。如需本招股章程的印刷版本，阁下可从上述网址下载并打印。

2025年12月31日

IMPORTANT

IMPORTANT NOTICE TO INVESTORS OF HONG KONG OFFER SHARES FULLY ELECTRONIC APPLICATION PROCESS

The Company has adopted a fully electronic application process for the Hong Kong Public Offering.

This prospectus is available at the website of the Stock Exchange at www.hkexnews.hk under the "HKEXnews > New Listings > New Listing Information" section, and our website at https://www.minimaxi.com.

The Company will not provide any physical channels to accept any application for the Hong Kong Offer Shares by the public. The contents of the electronic version of this prospectus are identical to the prospectus as registered with the Registrar of Companies in Hong Kong pursuant to section 342C of the Companies (Winding Up and Miscellaneous Provisions) Ordinance.

To apply for the Hong Kong Offer Shares, you may:

(1) apply online through the HK eIPO White Form service at www.hkeipo.hk; or

(2) apply electronically through the HKSCC EIPO channel and cause HKSCC Nominees to apply on your behalf by instructing your broker or custodian who is a HKSCC Participant to give electronic application instructions via HKSCC's FINI system to apply for the Hong Kong Offer Shares on your behalf.

If you are an intermediary, broker or agent, please remind your customers, clients or principals, as applicable, that this prospectus is available online at the website addresses stated above. Please refer to the section headed "How to Apply for Hong Kong Offer Shares" in this prospectus for further details of the procedures through which you can apply for the Hong Kong Offer Shares.

Your application through the HK eIPO White Form service or the HKSCC EIPO channel must be for a minimum of 20 Hong Kong Offer Shares and in one of the numbers set out in the table.

If you are applying through the HK eIPO White Form service, you may refer to the table below for the amount payable for the number of Hong Kong Offer Shares you have selected. You must pay the respective maximum amount payable on application in full upon application for Hong Kong Offer Shares.

– ii –

IMPORTANT

If you are applying through the HKSCC EIPO channel, you are required to pre-fund your application based on the amount specified by your broker or custodian, as determined based on the applicable laws and regulations in Hong Kong.

| No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | |---|---|---|---|---|---|---|---| | 20 | 3,333.28 | 400 | 66,665.61 | 6,000 | 999,984.16 | 80,000 | 13,333,122.00 | | 40 | 6,666.56 | 500 | 83,332.01 | 7,000 | 1,166,648.18 | 90,000 | 14,999,762.26 | | 60 | 9,999.84 | 600 | 99,998.41 | 8,000 | 1,333,312.20 | 100,000 | 16,666,402.50 | | 80 | 13,333.13 | 700 | 116,664.82 | 9,000 | 1,499,976.23 | 200,000 | 33,332,805.00 | | 100 | 16,666.40 | 800 | 133,331.22 | 10,000 | 1,666,640.26 | 300,000 | 49,999,207.50 | | 120 | 19,999.68 | 900 | 149,997.62 | 20,000 | 3,333,280.50 | 400,000 | 66,665,610.00 | | 140 | 23,332.96 | 1,000 | 166,664.03 | 30,000 | 4,999,920.76 | 500,000 | 83,332,012.50 | | 160 | 26,666.24 | 2,000 | 333,328.06 | 40,000 | 6,666,561.00 | 634,740(1) | 105,788,323.23 | | 180 | 29,999.52 | 3,000 | 499,992.08 | 50,000 | 8,333,201.26 | | | | 200 | 33,332.80 | 4,000 | 666,656.10 | 60,000 | 9,999,841.50 | | | | 300 | 49,999.21 | 5,000 | 833,320.13 | 70,000 | 11,666,481.76 | | |

(1) Maximum number of Hong Kong Offer Shares you may apply for and this is 50% of the Hong Kong Offer Shares initially offered.

(2) The amount payable is inclusive of brokerage, SFC transaction levy, the Stock Exchange trading fee and AFRC transaction levy. If your application is successful, brokerage will be paid to the Exchange Participants (as defined in the Listing Rules) or to the HK eIPO White Form Service Provider (for applications made through the application channel of the HK eIPO White Form service) while the SFC transaction levy, the Stock Exchange trading fee and the AFRC transaction levy will be paid to the SFC, the Stock Exchange and the AFRC, respectively.

– iii –

EXPECTED TIMETABLE(1)

If there is any change in the following expected timetable, we will issue an announcement to be published on the websites of the Company at https://www.minimaxi.com and the Stock Exchange at www.hkexnews.hk.

| Event | Date(1) | |---|---| | Hong Kong Public Offering commences | 9:00 a.m. on Wednesday, December 31, 2025 | | Latest time for completing electronic applications under HK eIPO White Form service through the designated website at www.hkeipo.hk(2) | 11:30 a.m. on Tuesday, January 6, 2026 | | Application lists open(3) | 11:45 a.m. on Tuesday, January 6, 2026 | | Latest time for (a) completing payment for HK eIPO White Form applications by effecting internet banking transfer(s) or PPS payment transfer(s) and (b) giving electronic application instructions to HKSCC(4) | 12:00 noon on Tuesday, January 6, 2026 |

If you are instructing your broker or custodian who is a HKSCC Participant to submit an EIPO application on your behalf through HKSCC's FINI system in accordance with your instruction to apply for the Hong Kong Offer Shares, you are advised to contact your broker or custodian for the earliest and latest time for giving such instructions, as this may vary by broker or custodian.

| Event | Date(1) | |---|---| | Application lists close(3) | 12:00 noon on Tuesday, January 6, 2026 | | Expected Price Determination Date(5) | on or before 12:00 noon, Wednesday, January 7, 2026 |

(1)

Announcement of the Offer Price, the level of indication of interest in the International Offering, the level of applications in the Hong Kong Public Offering and the basis of allocation of the Hong Kong Offer Shares under the Hong Kong Public Offering to be published on the website of Hong Kong Stock Exchange at www.hkexnews.hk and the Company's website at https://www.minimaxi.com (6) on or before (10) . . . . . . . . . . . . . . . . . . . . 11:00 p.m. on Thursday, January 8, 2026

– iv –

EXPECTED TIMETABLE(1) (2)

Results of allocations in the Hong Kong Public Offering (with successful applicants' identification document or business registration numbers, where appropriate) to be available through a variety of channels, including: •

in the announcement to be published on the website of the Hong Kong Stock Exchange at www.hkexnews.hk and on the Company's website at https://www.minimaxi.com, at or before . . . . . . . . . . . . 11:00 p.m. on Thursday, January 8, 2026

•

from the "Allotment Results" page in the designated results of allocations website at www.hkeipo.hk/IPOResult (or www.tricor.com.hk/ipo/result) from . . . . . . . . . . . . . . . . . . . . . 11:00 p.m. on Thursday, January 8, 2026 to 12:00 midnight on Wednesday, January 14, 2026

•

from the allocation results telephone enquiry line by calling +852 3691 8488 between 9:00 a.m. and 6:00 p.m. from . . . . . . . . . . . . . Friday, January 9, 2026 to Wednesday, January 14, 2026 (except Saturday, Sunday and public holiday in Hong Kong)

Share certificates in respect of wholly or partially successful applications to be dispatched or deposited into CCASS on or before (7) . . . . . . . . . . . . . . . . . Thursday, January 8, 2026

HK eIPO White Form e-Auto Refund payment instructions/refund checks in respect of (i) wholly or partially successful applications if the final Offer Price is less than the price payable on application (if applicable) and (ii) wholly or partially unsuccessful applications under the Hong Kong Public Offering to be dispatched on or before (8)(9) . . . . . . . . . . . Friday, January 9, 2026

Dealings in the Class A Ordinary Shares on the Hong Kong Stock Exchange expected to commence at 9:00 a.m. on . . . . . . . . . . . . . Friday, January 9, 2026

– v –

EXPECTED TIMETABLE(1)

Notes: (1)

All times refer to Hong Kong local time, except as otherwise stated.

(2)

You will not be permitted to submit your application through the designated website at www.hkeipo.hk after 11:30 a.m. on the last day for submitting applications. If you have already submitted your application and obtained an application reference number from the designated website at or before 11:30 a.m., you will be permitted to continue the application process (by completing payment of application monies) until 12:00 noon on the last day for submitting applications, when the application lists close.

(3)

If there is/are a tropical cyclone warning signal number 8 or above, a "black" rainstorm warning and/or Extreme Conditions in force in Hong Kong at any time between 9:00 a.m. and 12:00 noon on Tuesday, January 6, 2026, the application lists will not open or close on that day. Please see "How to Apply for Hong Kong Offer Shares — E. Severe Weather Arrangements".

(4)

Applicants who apply for Hong Kong Offer Shares through HKSCC EIPO channel or by instructing your broker or custodian to apply on your behalf via HKSCC EIPO Channel should see "How to Apply for Hong Kong Offer Shares — A. Application for Hong Kong Offer Shares — 2. Application Channels".

(5)

The Price Determination Date is expected to be on or before Wednesday, January 7, 2026 and, in any event, not later than 12:00 noon on Wednesday, January 7, 2026. If, for any reason, the Offer Price is not agreed between the Overall Coordinators (for themselves and on behalf of the Underwriters) and us by 12:00 noon on Wednesday, January 7, 2026, the Global Offering will not proceed and will lapse.

(6)

None of the website or any of the information contained on the website forms part of this prospectus.

(7)

The Share certificates will only become valid evidence of title at 8:00 a.m. on the Listing Date provided that the Global Offering has become unconditional and the right of termination described in "Underwriting — Underwriting arrangements and expenses — Hong Kong Public Offering — Grounds for termination" has not been exercised. Investors who trade Class A Ordinary Shares on the basis of publicly available allocation details or prior to the receipt of Share certificates or the Share certificates becoming valid do so entirely at their own risk.

(8)

HK eIPO White Form e-Auto Refund payment instructions/refund checks will be issued in respect of wholly or partially unsuccessful applications pursuant to the Hong Kong Public Offering and also in respect of wholly or partially successful applications in the event that the final Offer Price is less than the price payable per Offer Share on application. Part of the applicant's identification document number, or, if the application is made by joint applicants, part of the identification document number of the first-named applicant, provided by the applicant(s) may be printed on the refund check, if any. Such data would also be transferred to a third party for refund purposes. Banks may require verification of an applicant's identification document number before encashment of the refund check. Inaccurate completion of an applicant's identification document number may invalidate or delay encashment of the refund check.

(9)

EXPECTED TIMETABLE(1)

The above expected timetable is a summary only. You should see "Structure of the Global Offering" and "How to Apply for Hong Kong Offer Shares" for details of the structure of the Global Offering, including the conditions of the Global Offering, and the procedures for application for the Hong Kong Offer Shares.

If the Global Offering does not become unconditional or is terminated in accordance with its terms, the Global Offering will not proceed. In such a case, the Company will make an announcement as soon as practicable thereafter.

– vii –

CONTENTS

IMPORTANT NOTICE TO INVESTORS

This Prospectus is issued by us solely in connection with the Hong Kong Public Offering and does not constitute an offer to sell or a solicitation of an offer to buy any security other than the Hong Kong Offer Shares offered by this Prospectus pursuant to the Hong Kong Public Offering. This Prospectus may not be used for the purpose of, and does not constitute, an offer or a solicitation of an offer to subscribe for or buy, any security in any other jurisdiction or in any other circumstances. No action has been taken to permit a public offering of the Offer Shares or the distribution of this Prospectus in any jurisdiction other than Hong Kong. The distribution of this Prospectus and the offering and sale of the Offer Shares in other jurisdictions are subject to restrictions and may not be made except as permitted under the applicable securities laws of such jurisdictions pursuant to registration with or authorization by the relevant securities regulatory authorities or an exemption therefrom.

You should rely only on the information contained in this Prospectus to make your investment decision. We have not authorized anyone to provide you with information that is different from what is contained in this Prospectus. Any information or representation not made in this Prospectus must not be relied on by you as having been authorized by us, the Joint Sponsors, the Overall Coordinators, the Joint Global Coordinators, the Joint Bookrunners, the Joint Lead Managers, the Capital Market Intermediaries, the Underwriters, any of our or their respective directors, officers or representatives, or any other person or party involved in the Global Offering.

| | Page | |---|---| | EXPECTED TIMETABLE | iv | | CONTENTS | viii | | SUMMARY | 1 | | DEFINITIONS | 34 | | GLOSSARY OF TECHNICAL TERMS | 49 | | FORWARD-LOOKING STATEMENTS | 58 |

Applicants who have applied for Hong Kong Offer Shares through HKSCC EIPO channel should refer to the section headed "How to Apply for Hong Kong Offer Shares — D. Despatch/Collection of Share Certificates and Refund of Application Monies" for details.

Applicants who have applied through the HK eIPO White Form service and paid their application monies through single bank accounts may have refund monies (if any) dispatched to the bank account in the form of HK eIPO White Form e-Auto Refund payment instructions. Applicants who have applied through the HK eIPO White Form service and paid their application monies through multiple bank accounts may have refund monies (if any) dispatched to the address as specified in their application instructions in the form of refund checks in favor of the applicant (or, in the case of joint applications, the first-named applicant) by ordinary post at their own risk.

Any uncollected Share certificates will be dispatched by ordinary post, at the applicants' risk, to the addresses specified in the relevant applications.

Further information is set out in the section headed "How to Apply for Hong Kong Offer Shares — D. Despatch/Collection of Share Certificates and Refund of Application Monies".

– vi –

RISK FACTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

60

WAIVERS AND EXEMPTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

116

INFORMATION ABOUT THIS PROSPECTUS AND THE GLOBAL OFFERING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

126

– viii –

CONTENTS DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING . . . . .

131

CORPORATE INFORMATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

141

INDUSTRY OVERVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

144

REGULATORY OVERVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

168

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE . . . . . . . . . .

201

BUSINESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

236

DIRECTORS AND SENIOR MANAGEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . .

350

RELATIONSHIP WITH OUR CONTROLLING SHAREHOLDERS . . . . . . . . . .

366

CONNECTED TRANSACTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

370

SUBSTANTIAL SHAREHOLDERS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

376

CORNERSTONE INVESTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

379

FUTURE PLANS AND USE OF PROCEEDS. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

466

UNDERWRITING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

474

STRUCTURE OF THE GLOBAL OFFERING . . . . . . . . . . . . . . . . . . . . . . . . . . . .

491

HOW TO APPLY FOR HONG KONG OFFER SHARES . . . . . . . . . . . . . . . . . . .

509

APPENDIX I

ACCOUNTANT'S REPORT . . . . . . . . . . . . . . . . . . . . . . . .

I-1

APPENDIX II

UNAUDITED PRO FORMA FINANCIAL INFORMATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

II-1

APPENDIX III

SUMMARY OF THE CONSTITUTION OF OUR COMPANY AND CAYMAN ISLANDS COMPANY LAW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

III-1

APPENDIX IV

STATUTORY AND GENERAL INFORMATION . . . . . . . .

IV-1

APPENDIX V

DOCUMENTS DELIVERED TO THE REGISTRAR OF COMPANIES IN HONG KONG AND AVAILABLE ON DISPLAY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

V-1

– ix –

SUMMARY

This summary aims to give you an overview of the information contained in this Prospectus. As it is a summary, it does not contain all the information that may be important to you. You should read the whole Prospectus before you decide to invest in the Offer Shares. In particular, we are a specialist technology company seeking to list on the Main Board of the Hong Kong Stock Exchange under Chapter 18C of the Listing Rules because we are unable to meet the requirements under Rule 8.05 (1), (2) or (3) of the Listing Rules. There are unique challenges, risks and uncertainties associated with investing in companies such as ours. In addition, we have incurred operating loss since our inception, and we may incur adjusted net loss (non-IFRS measure) and operating loss for the foreseeable future. We had negative net cash flow from operating activities during the Track Record Period. We did not declare or pay any dividends during the Track Record Period and may not pay any dividends in the foreseeable future. Your investment decision should be made in light of these considerations.

There are risks associated with any investment. Some of the particular risks in investing in the Offer Shares are set out in the section headed "Risk Factors" in this Prospectus. You should read that section carefully in full before you decide to invest in the Offer Shares.

OVERVIEW

MiniMax is a global AI foundation model company. Founded by a group of forward-thinking engineers, we are committed to driving AI innovation towards performing the full range of human intellectual tasks, from learning and reasoning to planning and generalizing knowledge across diverse domains.

The foundation model market is expanding at an unprecedented pace, rapidly reshaping human society. The global foundation model market is projected to exceed US$300 billion by 2030. IDC estimates that AI will cumulatively contribute US$19.9 trillion to the global economy through 2030 and drive 3.5% of global GDP in 2030. We believe we have established a solid foundation to capture this market potential and have already made meaningful progress.

Our Journey

Our journey has been guided by a clear vision since inception centered on two key areas: developing advanced foundation models and creating AI-native products that enhance productivity and enrich life. Recognizing that real-world human interaction is inherently multi-modal, we stand out as one of the few foundation model developers who are committed to developing multi-modal models from day one. We take a cost-efficient approach in pursuing AI advancement, delivering high performance while ensuring our technological breakthroughs remain accessible and affordable to users globally. We adopted the Mixture-of-Experts (MoE) architecture and hybrid attention mechanism at an early stage, which significantly reduced computation resources while maintaining globally recognized performance.

–1–

SUMMARY

| | | | | | | | |---|---|---|---|---|---|---| | Large Language Model | abab 1 | | abab 5.5 | abab 6.0 (MoE) | Text-01 | | | | | | | | | M1 | | | | | | | | M2 | | Video Generation Model | | | | | Hailuo-01 | Hailuo-02 | | Audio Model | | Speech-01 | | Speech-02 | | | | | | | Music-01 | Music-02 | | | | Product Launching | | Talkie/Xingye | MiniMax (with Agent) | OPEN Platform | MiniMax Audio | | | | 2022 | 2023 | | 2024 | | 2025 |

We have been consistently iterating our models to higher intelligence levels. Today, our proprietary foundation model suite, led by MiniMax-M2, Hailuo-02, and Speech-02, has long context processing capacity and can understand, generate, and integrate a wide range of modalities, including text, video, and audio. These models power our major AI-native products — including MiniMax, Hailuo AI, MiniMax Audio, Talkie/Xingye, and our enterprise and developer-facing Open Platform, delivering intelligent and dynamic experiences to users globally.

As of September 30, 2025, our AI-native products had cumulatively served over 200 million individual users across over 200 countries and regions, and more than 100 thousand enterprises and developers across over 100 countries and regions.

Scalability

We believe scalability is pivotal to our long-term goals. To build one of the most scalable AI businesses globally, we focus on three core competencies — original research, a sustainable business model, and organizational efficiency. These pillars support both continuous model advancement and product commercialization at scale. Together, the three core competencies enable an elevated level of intelligence for everyone — powering productivity and enriching life. See "Business — Scalability."

–2–

SUMMARY

OUR MODELS AND PRODUCT OFFERINGS

Our Foundation Model Suite

We have leveraged our R&D capabilities to build a comprehensive suite of foundation models, and maintain competitiveness across various modalities. Our foundation model suite includes large language models, video generation models, and models for speech and music generation.

Large Language Model: MiniMax M Series

The MiniMax M Series, comprising MiniMax-M1 and MiniMax-M2, represents our flagship family of large language models. MiniMax-M1, launched in June 2025, is an open-source, large-scale hybrid-attention reasoning model. It adopts a hybrid MoE architecture combined with a lightning attention mechanism, enabling long-context processing with a context window of up to 1 million tokens and supporting the development of more capable AI agents.

MiniMax-M2, our latest large language model, is engineered for elite performance in coding and agentic tasks. Leveraging a carefully engineered, data-efficient MoE architecture and activation-parameter design, MiniMax-M2 delivers higher-performance capabilities at substantially faster inference speeds compared with MiniMax-M1, while maintaining an optimized profile across model intelligence, responsiveness and cost-efficiency.

Video Generation Model: Hailuo-02

The Hailuo-02 series model generates high-quality video content from a variety form of information inputs. Commercialized at scale with competitive results on global benchmarks upon its release, Hailuo-02 offers cinematic video quality, advanced prompt adherence, smooth motion, and style diversity. With user-friendly interface and ability to do aesthetic refinement, it helps content creators and advertisers produce compelling videos out of simple prompts.

Speech Generation Model: Speech-02

The Speech-02 model series is designed to generate natural, high-quality speech from text input. Widely recognized as a top performing speech model globally upon its release in April 2025, our Speech-02 model delivers hyper-realistic, personalized voice synthesis across multiple languages.

Our AI-Native Product Offerings

Leveraging our multi-modal foundation model suite, we deliver AI-native products and services that unleash the power of AI to benefit both individual users, developers and enterprise customers around the world. The evolution of our AI-native products is rooted in advancements

–3–

SUMMARY

in its underlying foundation models. Through continuous iterations and upgrades of foundation models and the development of new ones, we are able to design and create AI-native products with enhanced productivity and user experience.

MiniMax: Intelligent Agent Application

MiniMax is our intelligent AI agent application, which is designed to autonomously perform a wide range of tasks through natural language instructions. Supported by our foundation models, MiniMax Agent can plan, reason, and execute complex actions such as coding, research, document drafting, and presentation creation within a unified workspace.

Hailuo AI: Flagship Visual Generation Platform

Hailuo AI fully integrates our Hailuo-02 model that has quickly become one of the world's most popular AI image and video creation platforms through organic user adoption. It is offered in both web and app forms, and is designed for real-time, high-quality image and video generation.

MiniMax Audio: Advanced Audio Generation Tool

MiniMax Audio is designed to provide users with high-fidelity audio generation capabilities. Accessible via web platform, MiniMax Audio integrates the Company's Speech-02 model to support interactive audio synthesis and generate natural, high-quality speech from text input.

Talkie/Xingye: Multi-modal Entertainment Platform

Talkie (for international markets)/Xingye (星野) (for Chinese domestic market) is a globally recognized AI-native multi-modal entertainment platform. Users of Talkie/Xingye can engage with emotionally responsive AI themes or virtual characters powered by the Company's proprietary AI-models.

MiniMax Open Platform

Our Open Platform offers scalable, configurable AI services to enterprise customers and developers across more than 100 countries and regions as of September 30, 2025. Through public APIs and services, enterprise and developer customers can access the Company's foundation models and integrate such text, video and audio model capabilities into their own products and services. Our Open Platform supports rapid business deployment in key industry sectors such as smart devices, healthcare, cultural tourism, finance, and internet services — making it one of the world's largest open platforms for enterprises and developers in terms of average daily token volume, signifying widespread adoption.

–4–

SUMMARY

OUR BUSINESS AND REVENUE MODEL

Our revenue is derived from two primary sources — (i) AI-native products and (ii) Open Platform and other AI-based enterprise services. Each revenue stream reflects a distinct monetization pathway aligned with our product and platform strategies. The following table sets forth the breakdown of our revenue by nature, in absolute amounts and as a percentage of our total revenue, for the periods indicated.

| | For the year ended December 31, | | | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | | | AI-native products | — | — | 758 | 21.9 | 21,805 | 71.4 | 13,529 | 69.5 | 38,020 | 71.1 | | Open Platform and other AI-based enterprise services | — | — | 2,702 | 78.1 | 8,718 | 28.6 | 5,925 | 30.5 | 15,417 | 28.9 | | Total revenue | — | — | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 |

AI-native products. We generate revenue from individual users through subscription-based access to our monetized AI-native consumer applications, such as MiniMax, MiniMax Audio, Hailuo AI, and Talkie/Xingye. Subscriptions provide users with premium functionality across multi-modal generation, intelligent interaction, and personalized content. Revenue is recognised ratably over the subscription period, as we satisfy a stand-ready performance obligation to provide continuous access to content and services throughout the term. Users have the option to pre-purchase additional credits to recharge their accounts and buy these virtual items. For consumable virtual items, revenue is recognised when the virtual items are consumed. For non-consumable virtual items, revenue is recognised over the estimated average acting period of the paying users. In addition, we generate online marketing service revenue by providing marketing services to mediation platforms on certain of our AI-native applications. Revenue is recognised at a point in time, when a user views or clicks on an advertisement, thereby fulfilling our performance obligation. These services enable mediation platforms to engage with end users in a contextually relevant and measurable manner. As our user base and engagement levels expand, this revenue stream is expected to continue contributing to our overall monetization.

Open Platform and other AI-based enterprise services. We provide enterprise customers and developers with access to our usage-based Open Platform and other AI-based enterprise services. Revenue from API usage is recognised at a point in time when the customers call APIs with tokens, which are billed under a certain agreed fee schedule or usage-based structure. Revenue from other AI-based enterprise services, which mainly consists of arrangements customized to enterprise requirements and licensed deliverables, is typically recognised at a point in time,

–5–

SUMMARY

when control is transferred or acceptance is confirmed. In relation to our AI-based enterprise services, for customised arrangements, we work with enterprise customers to set up dedicated inference resource pools tailored to their needs, helping ensure stable and predictable model inference performance. For licensed deliverables, we license our foundation models to enable customers to deploy and operate such models in their own systems. These services support enterprise use cases across sectors such as smart devices, healthcare, cultural tourism, finance, and internet services.

The tables below set forth breakdowns of revenue by product and further by monetization method:

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | (in thousands, except for percentages) |

AI-native products | MiniMax | – | – | – | – | – | – | – | – | 756 | 1.4 | | Hailuo AI | – | – | – | – | 2,347 | 7.7 | – | – | 17,464 | 32.6 | | MiniMax Audio | – | – | – | – | – | – | – | – | 1,050 | 2.0 | | Talkie/Xingye | – | – | 758 | 21.9 | 19,458 | 63.7 | 13,529 | 69.5 | 18,750 | 35.1 | | Open Platform and other AI-based enterprise services | – | – | 2,702 | 78.1 | 8,718 | 28.6 | 5,925 | 30.5 | 15,417 | 28.9 | | Total revenue | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 |

–6–

SUMMARY

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | (in thousands, except for percentages) |

AI-native products MiniMax | — In-app top-up | – | – | – | – | – | – | – | – | 204.0 | 0.4 | | — Subscriptions | – | – | – | – | – | – | – | – | 552.0 | 1.0 | Hailuo AI | — In-app top-up | – | – | – | – | 527 | 1.7 | – | – | 3,317 | 6.2 | | — Subscriptions | – | – | – | – | 1,820 | 6.0 | – | – | 14,147 | 26.4 | MiniMax Audio | — In-app top-up | – | – | – | – | – | – | – | – | 196 | 0.4 | | — Subscriptions | – | – | – | – | – | – | – | – | 854 | 1.6 | Talkie/Xingye | — In-app top-up | – | – | 164 | 4.8 | 897 | 3.0 | 712 | 3.7 | 958 | 1.8 | | — Subscriptions | – | – | 594 | 17.1 | 3,960 | 12.9 | 2,917 | 14.9 | 6,604 | 12.4 | | — Online marketing service | – | – | – | – | 14,601 | 47.8 | 9,900 | 50.9 | 11,188 | 20.9 | | Open Platform and other AI-based enterprise services | – | – | 2,702 | 78.1 | 8,718 | 28.6 | 5,925 | 30.5 | 15,417 | 28.9 | | Total revenue | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 |

We expect our cost of sales as well as our research and development expenses to continue to increase in absolute terms for the year ended December 31, 2026, reflecting our strategic focus on advancing foundational AI model capabilities. See "Financial Information."

Key Operating Data

Our suite of AI-native products has attracted a broad user base, with average MAU rising more than six times from 3.1 million in 2023 to 19.1 million in 2024 and further to 27.6 million in the nine months ended September 30, 2025. Cumulative users of our AI-native products increased to more than 212 million by September 30, 2025. The growing number of users of our consumer-facing products provides valuable feedback, enabling rapid product iteration and improvement.

Our number of paying users for AI-native products expanded from around 119,700 in 2023 to around 650,300 in 2024, and further to approximately 1,771,600 in the nine months ended September 30, 2025.

Complementing our growing individual paying user base, we have also cultivated a portfolio of enterprise customers and developers. Enterprise customers and developers access our core AI models via our Open Platform, which supports growing business needs across key industry sectors. Our Open Platform demonstrates solid monetization capabilities for the

–7–

SUMMARY

我们提供的基础模型。我们持续观察到在开放平台上付费客户的快速增长。我们在开放平台上的付费用户数量（定义为在特定时期内个人消费API调用金额不低于50美元的用户）从2023年的约100人增长至2024年的约700人，并进一步增长至截至2025年9月30日止九个月的约2,500人。请参阅"业务——关键运营数据"。

下表列示了业绩记录期各期内我们的用户及客户数量¹²：

| | 截至12月31日 | | | 截至9月30日 | | |---|---|---|---|---|---| | | 2022年 | 2023年 | 2024年 | 2024年 | 2025年 | | （千用户） | | | | | | | AI原生产品 | | | | | | | MiniMax | – | 686 | 13,541 | 10,969 | 19,057 | | Hailuo AI | – | – | 5,735 | 36 | 42,348 | | MiniMax Audio | – | – | 47 | – | 3,742 | | Talkie/星野 | – | 10,445 | 96,055 | 65,566 | 147,100 | | 开放平台 | – | 13 | 42 | 34 | 132 | | **合计** | **–** | **11,144** | **115,420** | **76,605** | **212,379** |

注： 1. 用户数量包括我们基于网页端AI原生产品的所有注册用户以及基于应用程序AI原生产品的所有已激活设备。由于部分用户可能拥有多个账户，我们无法保证每位用户均为独立个体。

2. 开放平台的客户数量包括所有已注册并在我们开放平台上进行过API调用的客户。已注册但未进行API调用的客户不包括在内。我们的开放平台设计为面向广泛开发者（包括个人开发者和企业客户）的技术访问平台。开放平台以开发者账户为基础进行管理。在注册及后续使用过程中，我们仅要求用户提供基本联系信息（如电子邮件地址、手机号码及账户昵称），不要求上传营业执照或身份证明文件，亦不将此类信息作为强制性分类标准。因此，我们目前无法可靠区分开放平台客户是个人开发者还是企业客户。

–8–

SUMMARY

下表列示了业绩记录期各期内我们的付费用户数量¹：

| | 截至12月31日止年度 | | | 截至9月30日止九个月 | | |---|---|---|---|---|---| | | 2022年 | 2023年 | 2024年 | 2024年 | 2025年 | | （千用户） | | | | | | | AI原生产品 | | | | | | | MiniMax | – | – | – | – | 10.3 | | Hailuo AI | – | – | 64.8 | – | 311.1 | | MiniMax Audio | – | – | – | – | 59.8 | | Talkie/星野 | – | 119.7 | 585.5 | 489.1 | 1,390.4 | | 开放平台 | – | 0.1 | 0.7 | 0.4 | 2.5 | | **合计** | **–** | **119.8** | **651.0** | **489.5** | **1,774.1** |

注： 1. AI原生产品的付费用户定义为在特定时期内至少完成一笔货币交易的用户。开放平台的付费用户定义为在特定时期内个人消费API调用金额不低于50美元的用户。

下表列示了业绩记录期各期内的平均月活跃用户数（"MAU"）¹²：

| | 截至12月31日止年度 | | | 截至9月30日止九个月 | | |---|---|---|---|---|---| | | 2022年 | 2023年 | 2024年 | 2024年 | 2025年 | | （千用户） | | | | | | | AI原生产品 | | | | | | | MiniMax | – | 239 | 2,195 | 2,166 | 1,429 | | Hailuo AI | – | – | 2,172 | 36 | 5,648 | | MiniMax Audio | – | – | 47 | – | 494 | | Talkie/星野 | – | 2,905 | 14,692 | 12,399 | 20,051 | | 开放平台 | – | 4 | 5 | 4 | 16 | | **合计** | **–** | **3,148** | **19,111** | **14,605** | **27,638** |

注： 1. 月活跃用户数包括在特定月份内在我们AI原生应用程序上至少执行过一次操作的所有独立设备，以及至少登录过我们网页平台一次的所有注册用户账户，包括付费用户和非付费用户。由于部分用户可能拥有多个账户，我们无法保证每位用户均为独立个体。

2. 开放平台的平均月活跃客户数包括在特定月份内在我们开放平台上进行过API调用的所有注册客户，包括付费客户和非付费客户。已注册但未进行API调用的客户不包括在内。

–9–

SUMMARY

下表列示了业绩记录期各期内的新增用户数量¹²：

| | 截至12月31日止年度 | | | 截至9月30日止九个月 | | |---|---|---|---|---|---| | | 2022年 | 2023年 | 2024年 | 2024年 | 2025年 | | （千用户） | | | | | | | AI原生产品 | | | | | | | MiniMax | – | 686 | 12,855 | 10,283 | 5,516 | | Hailuo AI | – | – | 5,735 | 36 | 36,613 | | MiniMax Audio | – | – | 47 | – | 3,695 | | Talkie/星野 | – | 10,445 | 85,610 | 55,121 | 51,045 | | 开放平台 | – | 13 | 29 | 21 | 90 | | **合计** | **–** | **11,144** | **104,276** | **65,461** | **96,959** |

注： 1. 新增用户包括我们基于网页端AI原生产品的所有新注册用户以及基于应用程序AI原生产品的所有新激活设备。

2. 开放平台的新增客户包括所有新注册并进行过API调用的客户。已注册但未进行API调用的客户不包括在内。

下表列示了业绩记录期各期内每位付费客户的平均消费金额¹：

| | 截至12月31日止年度 | | | 截至9月30日止九个月 | | |---|---|---|---|---|---| | | 2022年 | 2023年 | 2024年 | 2024年 | 2025年 | | （美元） | | | | | | | AI原生产品 | | | | | | | MiniMax | – | – | – | – | 73 | | Hailuo AI | – | – | 36 | – | 56 | | MiniMax Audio | – | – | – | – | 18 | | Talkie/星野 | – | 6 | 8 | 7 | 5 | | 开放平台 | – | 27,020 | 12,454 | 14,813 | 6,167 |

注： 1.

SUMMARY

CUSTOMERS AND SUPPLIERS

Our customers comprise a broad and diverse base across various sectors and geographies. In 2023, 2024 and nine months ended September 30, 2025, revenue from our five largest customers in each year/period during the Track Record Period amounted to US$2.1 million, US$13.4 million and US$11.6 million, representing 60.5%, 44.1% and 21.7% of our total revenue for the respective periods. In 2023, 2024 and nine months ended September 30, 2025, revenue derived from our largest customer in each year/period during the Track Record Period amounted to US$1.3 million, US$9.4 million and US$7.8 million, representing 37.2%, 30.9% and 14.7% of our total revenue for the respective periods.

We maintain stable and long-standing relationships with a select group of suppliers, principally in the areas of cloud infrastructure services. In 2022, 2023, 2024 and the nine months ended September 30, 2025, purchases from our five largest suppliers in each year/period during the Track Record Period amounted to US$4.5 million, US$49.8 million, US$149.0 million and US$143.6 million, representing 63.9%, 63.0%, 57.3% and 62.5% of our total purchases for the respective periods. In 2022, 2023, 2024 and the nine months ended September 30, 2025, purchases derived from our largest supplier in each year/period during the Track Record Period amounted to US$1.6 million, US$23.0 million, US$72.8 million and US$54.9 million, representing 22.8%, 29.1%, 28.0% and 23.9% of our total purchases for the respective periods. See "Business — Ecosystem of Partners."

LEGAL PROCEEDINGS

On September 16, 2025, a group of major U.S. movie studio companies, including Disney, Universal and Warner Bros. Discovery (the "Plaintiffs"), filed a civil complaint (the "Complaint") in the United States District Court for the Central District of California, against our Group in relation to Hailuo AI, our visual generation platform. The Plaintiffs allege (i) direct infringement, on the basis that the Company itself, through Hailuo AI, created and displayed videos and images depicting a number of well-known film and animation characters owned by the Plaintiffs, and (ii) secondary infringement, including contributory and vicarious infringement, on the basis that the Company knew or should have known that users could create content depicting the Plaintiffs' characters, and because the Company is allegedly benefiting from that use. In their prayer for relief, the Plaintiffs primarily seek, among other things, monetary relief in the form of actual or statutory damages, injunctive relief, attorneys' fees and other equitable remedies.

These claims are commercial disputes in nature, and having considered advice from our U.S. litigation advisor, our Directors believe that they are without merit in all material respects and that there is insufficient evidence to support them. Based on the Joint Sponsors' due diligence conducted, there was no reasonable basis for the Joint Sponsors to disagree with the Directors' view that the claims are without merit in all material respects. The Company categorically denies the allegations of direct infringement, as Hailuo AI only produces outputs in response to user prompts and therefore lacks the volitional conduct required for direct liability. In respect of the secondary infringement claims, Hailuo AI is a general-purpose

– 12 –

摘要

为合法用途而创建的创意工具，且本公司并无任何与被指控侵权行为直接相关的经济利益，因此不满足构成帮助侵权或替代侵权的必要要件。

原告在其诉状中主张，其有权就每项被侵权作品获得最高15万美元的法定赔偿，该金额为美国版权法规定的每件作品最高可判赔金额，且仅在认定侵权行为属于故意侵权时方可判处。诉状附件列明了本案涉及的约500项影视作品注册登记。因此，假设原告胜诉并全面支持其诉求，按照原告所主张的最坏情形，货币赔偿请求将达7,500万美元的法定赔偿金（计算方式为：被指控作品数量乘以被主张的最高法定赔偿金额），并附带禁令救济。海螺AI输出的用户创作内容中，绝大多数与原告的角色无关。经参考本公司美国诉讼顾问的意见，本公司董事认为原告在责任认定及最高法定赔偿方面均全面胜诉的可能性极为微小。尽管原告主张其所列明的约500项注册登记中每一项均有资格被计为一件"作品"，但经参考本公司美国诉讼顾问的意见，本公司董事认为，依法可纳入法定赔偿计算的"作品"数量很可能大幅低于诉状所引用的约500项注册登记数量，原因在于：美国法律并非按注册登记数量来评估法定赔偿，且许多被指控角色在多项注册登记中反复出现，并无新的可受保护的表达内容；因此，该等角色不应被视为独立的"作品"。经进一步参考本公司美国诉讼顾问的意见，本公司董事认为，每件作品15万美元的法定赔偿金极少在不存在故意且反复侵权的情况下被判处——董事认为本案事实不支持该标准——且统计研究表明，此类最高赔偿金仅在属于故意侵权的例外情形下方会被判处。根据联席保荐人所实施的独立尽职调查步骤，包括与本公司美国诉讼顾问进行访谈和讨论，联席保荐人有合理依据相信上述董事所表达的观点如实反映了本公司美国诉讼顾问的意见。

基于上述情况，经参考本公司美国诉讼顾问的意见，董事认为上述索赔不会对本公司的业务、经营业绩或财务状况产生重大不利影响，理由如下：(i) 即便在极为微小的可能性下原告全面胜诉并获得所主张的最高法定赔偿，约7,500万美元的最高被指控赔偿金也仅约占本公司可用财务资源（包括现金及现金等价物、可用金融资产流动部分及预期IPO所得款项）的4.9%，且考虑到美国民事诉讼通常历时较长以及本公司预期持续增长，该比例预计将随时间推移而下降；(ii) 禁令对本公司业务造成重大干扰的可能性较低，因为申请临时禁令需就包括实体胜诉可能性、无禁令将遭受不可弥补损害及紧迫性等因素作出充分论证，而上述任何因素均未获证实，且法院在类似科技案件中通常倾向于采取针对特定输出内容的有限救济措施，而非宽泛的全平台限制措施；

— 13 —

SUMMARY

作出任何此类运营影响在最终判决中不太可能发生；以及（三）原告的知识产权对海螺AI的商业可行性并不重要，海螺AI是一个为合法用途而创建的通用创意平台，董事认为此类知识产权并非用户参与度、收入产生或增长的重要驱动因素，因此排除此类内容预计不会对我们的业务、财务表现或前景产生重大影响。根据联席保荐人所采取的独立尽职调查步骤，包括（其中包括）：（a）与本公司美国诉讼顾问就潜在最坏情形、批准禁令救济的可能性及其在此方面的法律分析进行讨论；（b）审阅本集团的财务资料；以及（c）审查海螺AI于2025年6月至11月的月活跃用户数据及海螺AI输出内容的系统检查结果，联席保荐人同意董事的上述观点。详情请参阅"业务——法律程序及合规——版权侵权诉讼"、"风险因素——与我们业务及行业相关的风险——我们用于训练基础模型的内容或数据以及我们基础模型所生成的内容可能受到第三方知识产权侵权索赔，这可能对我们的业务、财务状况及经营业绩产生重大不利影响"以及"风险因素——与我们业务及行业相关的风险——我们可能成为第三方就我们侵犯其知识产权权利而提起诉讼的被告"。

我们的市场机遇及竞争

全球基于模型的基础模型市场仍处于商业化早期阶段。随着技术持续成熟及用户付费意愿稳步提升，全球基于模型的基础模型市场预计将从2024年的107亿美元（US$10.7 billion）迅速增长至2029年的2,065亿美元（US$206.5 billion），复合年增长率为80.7%。

我们通过对长上下文建模及可扩展多模态架构设计的技术专注来实现差异化，这使我们能够构建能够处理跨文本、视觉及音频的复杂、多维交互的模型。我们是全球第十大基础模型公司，市场份额为0.3%。此外，我们是全球第四大纯基础模型公司，按2024年基于模型的收入计算。

详情请参阅"行业概览"。

— 14 —

SUMMARY

RISK FACTORS

We are a Specialist Technology Company seeking Listing on the Main Board of the Stock Exchange under Chapter 18C of the Listing Rules. Our business and the Global Offering involve certain risks as set out in "Risk Factors" in this document. You should read that section in its entirety carefully before you decide to invest in our Shares. We believe the most significant risks we face include but are not limited to the following:

• We have recorded net losses, net liabilities and operating cash outflow during the Track Record Period and recorded net current liabilities as of September 30, 2025, and we may not be able to achieve or subsequently maintain profitability.

• We operate in a rapidly evolving and increasingly competitive global foundation model industry. Our business is subject to constant technological advancements and industry transformation. If we fail to continuously innovate and adapt to evolving customer needs, our competitive position would be impacted and our business, financial condition and results of operations may be materially and adversely affected.

• The content or data that we use to train our foundation models and the content generated by our foundation models could be subject to third-party intellectual property infringement claims which may materially and adversely affect our business, financial condition and results of operations.

• Any actual or perceived flaws or inappropriate usage of foundation model technologies committed by us or other third parties intentionally or inadvertently, could materially and adversely impact our reputation, business, financial condition, results of operations and the broader acceptance of foundation model products by society at-large.

• The competitiveness of our foundation models and offerings depends on our continuous and significant investment in research and development, and we intend to continue investing significantly in research and development. Such investment may negatively impact our profitability and operating cash flow in the short term and may not generate the results we expect to achieve.

See "Risk Factors" for details.

INTELLECTUAL PROPERTY RIGHTS

Intellectual property lies at the heart of our research, product development and commercial success. We safeguard our proprietary technologies through a layered strategy that combines (i) statutory protection under patent, copyright, trademark, trade-secret and unfair-competition laws in the PRC and other jurisdictions, and (ii) contractual safeguards such as confidentiality undertakings, invention-assignment covenants and license agreements. All

– 15 –

SUMMARY

employment and key commercial contracts expressly delineate ownership of, and obligations to protect, intellectual property created or used in the course of our business. During the Track Record Period, our core technologies were patented. Such patents are typically valid for 10 to 20 years.

As of the Latest Practicable Date, we had 75 patents registered with the National Intellectual Property Administration of the PRC, 144 trademarks registered in the PRC, 73 copyrights registered with the National Copyright Administration of the PRC, 28 registered strategic domain names in the PRC and 87 trademarks registered internationally. See "Appendix IV — Statutory and General Information — B. Further Information about our Business — 2. Intellectual Property Rights of our Group" for a schedule of material intellectual property rights.

During the Track Record Period and up to the Latest Practicable Date, we were not involved in any IP litigation, arbitration or administrative proceedings, nor have we received any claim alleging infringement of third-party rights that would have a material adverse effect on our business, results of operations, or financial condition. We will continue to monitor the landscape and, where necessary, defend or enforce our rights vigorously.

Notwithstanding the foregoing measures, we cannot rule out the possibility of future challenges to our IP or allegations of infringement against us. Enforcement actions may involve significant cost and management distraction. For a discussion of these and other related risks, please refer to "Risk Factors — We may not be able to adequately protect or enforce our intellectual property rights throughout the world, and our efforts to do so may be costly" and "Risk Factors — We may become subject to litigation brought by third parties claiming infringement by us of their intellectual property rights." See "Business — Intellectual Property."

WEIGHTED VOTING RIGHTS STRUCTURE

Our Company has a weighted voting rights structure. The Company satisfies the presumption for the "Innovative Company Requirements" and the "external validation" requirements under Chapter 2.2 of the Guide for New Listing Applicants published by the Stock Exchange pursuant to the Joint Announcement on Launch of Technology Enterprises Channel published on 6 May 2025 as it meets all relevant requirements under Chapter 18C of the Rules and Chapter 2.5 of the Guide for New Listing Applicants. Under our weighted voting rights structure, our share capital comprises Class A Ordinary Shares and Class B Ordinary Shares. Each Class B Ordinary Share entitles the holder to exercise ten votes, and each Class A Ordinary Share entitles the holder to exercise one vote, respectively, on any matters subject to the vote at general meetings of the Company, subject to Rule 8A.24 of the Listing Rules that requires the Reserved Matters to be voted on a one vote per share basis.

– 16 –

SUMMARY

The table below sets out the beneficial interests entitled to and voting rights to be held by the WVR Beneficiaries upon the completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised):

| | Number of Class B Ordinary Shares held | Number of Class A Ordinary Shares interested in (3) | |---|---|---| | Dr. Yan (2) | 74,102,534 | | | Ms. Yun (2) | 7,000,000 | |

– 17 –

摘要

我们的控股股东

全球发售完成后（假设发售规模调整权及超额配股权均未获行使），合共5,000,000股A类普通股及74,102,534股B类普通股将由MiniMax Awakening、MiniMax Limited、Alpha EXP及MiniMax Matrix持有，相当于（i）我们已发行股本于股东大会（就保留事项决议除外）中约72.05%的投票权，及（ii）我们已发行股本于股东大会就保留事项决议中约25.90%的投票权。MiniMax Awakening及MiniMax Limited由严博士通过Local Linearity全资拥有。MiniMax Matrix亦为严博士通过Local Linearity控制的实体。Alpha EXP由Scaling EXP Limited持有99%，Local Linearity持有1%。Scaling EXP Limited由Trident Trust Company (Hong Kong) Limited全资拥有，该公司作为Alpha EXP信托的受托人。Alpha EXP信托为严博士（作为委托人）为其自身利益而设立的信托。据此，严博士、Local Linearity Inc.、MiniMax Awakening、MiniMax Limited、Alpha EXP、Scaling EXP Limited及MiniMax Matrix于上市后将共同构成本公司的一组控股股东。

上市前投资

我们已进行数轮上市前投资。有关我们主要上市前投资者的背景及上市前投资主要条款的详情，请参阅「历史、重组及公司架构——上市前投资」。

上市规则第18C.14条规定的禁售要求

严博士、云女士及其紧密联系人以及我们的探路者策略性投资者，须遵守上市规则第18C.14条所规定的禁售要求。详情请参阅「历史、重组及公司架构——禁售期」一节。在本公司提出申请且联交所通知本公司于上市后将不再被视为上市前商业公司后，禁售期将于以下较晚日期届满：（i）倘本公司以商业公司身份申请上市，该等禁售期本应届满的日期；及（2）根据上市规则第18C.24条规定就取消上市前商业公司认定刊发公告后第30日。

---

3,355,030 1,644,970

于已发行股本中 所持实益权益的 大约百分比

25.36% 2.83%

所控制投票权的 大约百分比(1)

72.05% 6.76%

注释： (1) 基于每股A类普通股赋予股东每股一票的投票权，每股B类普通股赋予股东每股十票的投票权。

(2) 有关我们加权投票权受益人股权架构的详情，请参阅「历史、重组及公司架构」一节。

(3) 严博士及云女士分别于MiniMax Matrix中持有67.1%及32.9%的权益。

本公司采纳加权投票权架构，旨在使加权投票权受益人能够对本公司行使投票控制权。此举将使本公司受益于加权投票权受益人持续的远见与领导力，从而以长远前景与战略为目标对本公司实施控制。考虑到加权投票权受益人对本集团的贡献，此等安排符合本公司及其全体股东的最佳利益。

谨提醒有意投资者注意投资于具有加权投票权架构公司的潜在风险，尤其是加权投票权受益人的利益未必始终与全体股东的利益一致，且加权投票权受益人将能够行使其较高投票权，以影响本公司事务及股东决议的结果，无论其他股东如何投票。

有意投资者应在经过审慎及仔细考虑后，方可作出投资本公司的决定。有关本公司所采纳加权投票权架构相关风险的更多资料，请参阅「风险因素——与加权投票权架构相关的风险」。

除B类普通股所附带的加权投票权外，两类股份所附带的权利完全相同。有关A类普通股及B类普通股的权利、优先权、特权及限制的更多资料，请参阅本招股章程附录三「本公司章程及开曼群岛公司法概要——2 章程细则」以了解进一步详情。

– 17 –

SUMMARY OF HISTORICAL FINANCIAL INFORMATION The following tables set forth summary financial data from our consolidated financial information for the Track Record Period, extracted from the Accountants' Report set out in Appendix I. You should read this summary in conjunction with our consolidated financial information included in the Accountants' Report set out in Appendix I, including the accompanying notes, and the information set forth in "Financial Information."

– 18 –

SUMMARY Summary of Consolidated Statements of Profit or Loss The following table sets forth a summary of our consolidated statements of profit or loss, in absolute amounts and as a percentage of our total revenue, for the periods indicated. See "Financial Information" for details.

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | | | |---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | | | | | | | (unaudited) | | | | | | (in thousands, except for percentages) | | Revenue | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 | | Cost of sales | – | – | (4,314) | (124.7) | (26,785) | (87.8) | (18,944) | (97.4) | (40,961) | (76.7) | | Gross (loss)/profit | – | – | (854) | (24.7) | 3,738 | 12.2 | 510 | 2.6 | 12,476 | 23.3 | | Other income and gains, net | 1,155 | – | 8,942 | 258.4 | 36,151 | 118.4 | 25,278 | 129.9 | 31,232 | 58.4 | | Selling and distribution expenses | (587) | – | (22,827) | (659.7) | (86,995) | (285.0) | (53,389) | (274.4) | (39,325) | (73.6) | | Administrative expenses | (3,213) | – | (7,615) | (220.1) | (14,384) | (47.1) | (9,610) | (49.4) | (22,074) | (41.3) | | Research and development expenses | (10,560) | – | (70,002) | (2,023.2) | (188,979) | (619.1) | (138,684) | (712.9) | (180,312) | (337.4) | | Fair value loss on financial liabilities | (60,509) | – | (176,826) | (5,110.6) | (214,172) | (701.7) | (128,063) | (658.3) | (313,477) | (586.6) | | Finance costs | (14) | – | (61) | (1.8) | (509) | (1.7) | (316) | (1.6) | (511) | (1.0) | | Impairment losses on financial assets, net | – | – | (3) | (0.1) | (88) | (0.3) | (68) | (0.3) | (22) | (0.0) | | Loss before tax | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) | | Income tax expense | – | – | – | – | – | – | – | – | – | – | | Loss for the year/period | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) | | Attributable to: | | | | | | | | | | | | Owners of the parent | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) | | Non-controlling interests | – | – | – | – | – | – | – | – | – | – | | Loss and total comprehensive income for the year | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) |

Loss per share attributable to ordinary equity holders of the parent Basic and diluted – For loss for the year/period (US$): (0.74) | (2.56) | (4.28) | (2.80) | (4.71)

– 19 –

SUMMARY Non-IFRS Financial Measure We use adjusted net loss (non-IFRS measure), which is a non-IFRS financial measure, in evaluating our operating results and for financial and operational decision-making purposes. We believe that adjusted net loss (non-IFRS measure) helps identify underlying trends in our business that could otherwise be distorted by the effect of certain expenses that we include in our net loss. We believe that adjusted net loss (non-IFRS measure) provides useful information about our results of operations, enhances the overall understanding of our past performance and future prospects and allows for greater visibility with respect to key metrics used by our management in its financial and operational decision-making.

Adjusted net loss (non-IFRS measure) should not be considered in isolation or construed as an alternative to net loss or any other measure of performance or as an indicator of our operating performance. Investors are encouraged to review adjusted net loss (non-IFRS measure) and the reconciliation to its most directly comparable IFRS measure. Adjusted net loss (non-IFRS measure) presented here may not be comparable to similarly titled measures presented by other companies. Other companies may calculate similarly titled measures differently, limiting their usefulness as comparative measures to our data. We encourage investors and others to review our financial information in its entirety and not rely on a single financial measure.

We define our adjusted net loss (non-IFRS measure) as net loss adjusted by adding back (i) share-based payment expenses that are included in cost of sales, general administrative, research and development, and sales and marketing expenses, relates to the share-based awards that we grant to participants of our share incentive schemes and is a non-cash expense, (ii) fair value losses on financial liabilities, comprising fair value changes of convertible redeemable preferred shares which will be re-designated from liabilities to equity as a result of the automatic conversion into ordinary shares upon Listing, and convertible bonds, which have subsequently been repaid in full as of the Latest Practicable Date, and (iii) listing expenses.

The following table presents our non-IFRS financial measure for the years ended December 31, 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025. See "Financial Information — Non-IFRS Financial Measure" for details.

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 US$ | 2023 US$ | 2024 US$ | 2024 US$ | 2025 US$ | | | | | | (unaudited) | | | | (in thousands) | | Net loss for the year/period | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | Add: | | | | | | | Share-based payment expenses | 1,069 | 3,346 | 6,823 | 6,100 | 8,581 | | Fair value loss on financial liabilities | 60,509 | 176,826 | 214,172 | 128,063 | 313,477 | | Listing expenses | – | – | – | – | 3,675 | | Adjusted net loss for the year/period (non-IFRS measure) | (12,150) | (89,074) | (244,243) | (170,179) | (186,280) |

– 20 –

SUMMARY

We recorded US$73.7 million, US$269.2 million, US$465.2 million, US$304.3 million and US$512.0 million in loss for the year/period in 2022, 2023, 2024 and for the nine months ended September 30, 2024 and 2025, respectively, due to significant initial investment in foundation model R&D and AI infrastructure and fair value loss on financial liabilities.

Summary of Consolidated Statements of Financial Position

The table below sets forth selected information from our consolidated statements of financial position as of the dates indicated, which has been extracted from our consolidated financial statements included in Appendix I to this Prospectus.

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) |

NON-CURRENT ASSETS

| Property, plant and equipment | 231 | 458 | 709 | 1,093 | | Right-of-use assets | 458 | 3,313 | 3,077 | 2,746 | | Prepayments, other receivables and other assets | – | 435 | 561 | 731 | | Financial assets at fair value through profit or loss | – | – | 95,331 | 70,228 | | Financial assets at fair value through other comprehensive income | – | – | – | – | | Restricted cash | – | 39 | 4,836 | 6,440 | | | – | 39 | 38 | 41 | | Total non-current assets | 689 | 4,496 | 104,936 | 81,320 |

CURRENT ASSETS

| Trade receivables | – | 1,338 | 6,982 | 8,063 | | Prepayments, other receivables and other assets | 569 | 4,378 | 13,470 | 11,811 | | Financial assets at amortised costs | – | – | 147,444 | – | | Financial assets at fair value through profit or loss | 65,791 | 15,802 | 295,220 | 644,154 | | Time deposits | – | 91,698 | 26,327 | – | | Restricted cash | 2,221 | – | 27,293 | 25,097 | | Cash and cash equivalents | 4,691 | 206,295 | 288,912 | 362,647 |

– 21 –

SUMMARY

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) |

| Total current assets | 73,272 | 319,511 | 805,648 | 1,051,772 |

CURRENT LIABILITIES

| Interest-bearing bank borrowings | – | – | 19,455 | 19,102 | | Trade and bills payables | 2,394 | 17,242 | 51,212 | 70,219 | | Other payables, accruals and other liabilities | 2,326 | 14,741 | 51,512 | 17,322 | | Contract liabilities | – | 559 | 1,553 | 4,657 | | Lease liabilities | 349 | 1,248 | 1,964 | 1,694 | | Convertible redeemable preferred shares | 145,175 | 629,001 | 1,581,949 | 2,321,193 | | Total current liabilities | 150,244 | 662,791 | 1,707,645 | 2,434,187 |

| NET CURRENT LIABILITIES | (76,972) | (343,280) | (901,997) | (1,382,415) |

| TOTAL ASSETS LESS CURRENT LIABILITIES | (76,283) | (338,784) | (797,061) | (1,301,095) |

NON-CURRENT LIABILITIES

| Lease liabilities | 91 | 1,912 | 1,059 | 937 | | Other non-current liabilities | – | 1,218 | 1,200 | 1,467 | | Total non-current liabilities | 91 | 3,130 | 2,259 | 2,404 | | Net liabilities | (76,374) | (341,914) | (799,320) | (1,303,499) |

Our net current liabilities increased from US$77.0 million as of December 31, 2022 to US$343.3 million as of December 31, 2023, primarily due to (i) an increase in convertible redeemable preferred shares from US$145.2 million as of December 31, 2022 to US$629.0 million as of December 31, 2023, (ii) a decrease in financial assets at fair value through profit or loss from US$65.8 million as of December 31, 2022 to US$15.8 million as of December 31, 2023, (iii) an increase in trade and bills payables from US$2.4 million as of December 31, 2022 to US$17.2 million as of December 31, 2023, and (iv) an increase in other payables, accruals and other liabilities from US$2.3 million as of December 31, 2022 to US$14.7 million as of December 31, 2023, partially offset by (i) an increase in cash and cash equivalents from US$4.7 million as of December 31, 2022 to US$206.3 million as of December 31, 2023, and (ii) the recognition of time deposits of US$91.7 million.

– 22 –

SUMMARY

Our net current liabilities increased from US$343.3 million as of December 31, 2023 to US$902.0 million as of December 31, 2024, primarily due to (i) an increase in convertible redeemable preferred shares from US$629.0 million as of December 31, 2023 to US$1,581.9 million as of December 31, 2024, (ii) an increase in trade and bills payables from US$17.2 million as of December 31, 2023 to US$51.2 million as of December 31, 2024, and (iii) an increase in other payables, accruals and other liabilities from US$14.7 million as of December 31, 2023 to US$51.5 million as of December 31, 2024. This was partially offset by (i) an increase in financial assets at fair value through profit or loss from US$15.8 million as of December 31, 2023 to US$295.2 million as of December 31, 2024, and (ii) an increase in financial assets at amortized costs from nil as of December 31, 2023 to US$147.4 million as of December 31, 2024.

Our net current liabilities increased from US$902.0 million as of December 31, 2024 to US$1,382.4 million as of September 30, 2025, primarily due to (i) an increase in convertible redeemable preferred shares from US$1,581.9 million as of December 31, 2024 to US$2,321.2 million as of September 30, 2025, and (ii) a decrease in financial assets at amortized cost from US$147.4 million as of December 31, 2024 to nil as of September 30, 2025, and (iii) an increase in trade and bills payables from US$51.2 million as of December 31, 2024 to US$70.2 million as of September 30, 2025. This was partially offset by (i) an increase in financial assets at fair value through profit or loss from US$295.2 million as of December 31, 2024 to US$644.2 million as of September 30, 2025, and (ii) a decrease in other payables, accruals and other liabilities from US$51.5 million as of December 31, 2024 to US$17.3 million as of September 30, 2025.

We had net liabilities of US$76.4 million, US$341.9 million, US$799.3 million and US$1,303.5 million as of December 31, 2022, 2023, 2024 and September 30, 2025, respectively.

As of September 30, 2025, we had net liabilities of US$1,303.5 million primarily because of convertible redeemable preferred shares totaling US$2,321.2 million. Nevertheless, as these convertible redeemable preferred shares will be re-designated from financial liabilities to equity as a result of the automatic conversion into ordinary shares upon Listing, our net liabilities position will turn into a net assets position.

Our net liabilities increased from US$76.4 million as of December 31, 2022 to US$341.9 million as of December 31, 2023, primarily due to an increase of US$269.2 million in the accumulated losses for the period, partially offset by (i) an increase of US$3.3 million in the recognition of share-based payment expenses, and (ii) an increase of US$0.4 million in the exchange differences on translation of foreign operations.

Our net liabilities increased from US$341.9 million as of December 31, 2023 to US$799.3 million as of December 31, 2024, primarily due to an increase of US$465.2 million in the accumulated losses for the period, partially offset by (i) an increase of US$6.8 million in the

– 23 –

SUMMARY

recognition of share-based payment expenses, (ii) an increase of US$0.4 million in the exchange differences on translation of foreign operations, and (iii) an increase of US$0.7 million of change in fair value of equity investments at fair value through other comprehensive income, net of tax.

Our net liabilities increased from US$799.3 million as of December 31, 2024 to US$1,303.5 million as of September 30, 2025, primarily due to an increase of US$512.0 million in the accumulated losses for the period, partially offset by (i) an increase of US$8.6 million in the recognition of share-based payment expenses, and (ii) an increase of US$1.6 million of change in fair value of equity investments at fair value through other comprehensive income, net of tax.

Summary of Consolidated Statements of Cash Flows

The following table sets forth our cash flows for the periods indicated.

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | | | | | Net cash flows used in operating activities | (11,019) | (64,455) | (258,483) | (195,596) | (209,396) | | Net cash flows used in investing activities | (35,156) | (40,320) | (431,300) | (630,463) | (126,231) | | Net cash flows generated from financing activities | 49,786 | 306,243 | 771,092 | 718,827 | 407,913 | | Net increase/(decrease) in cash and cash equivalents | 3,611 | 201,468 | 81,309 | (107,232) | 72,286 | | Cash and cash equivalents at the beginning of the year/period | 994 | 4,691 | 206,295 | 206,295 | 288,912 | | Effect of foreign exchange differences, net | 86 | 136 | 1,308 | 500 | 1,449 | | Cash and cash equivalents at the end of the year/period | 4,691 | 206,295 | 288,912 | 99,563 | 362,647 |

During the Track Record Period, we recorded net cash flows used in operating activities, primarily reflecting our position as an R&D-intensive Specialist Technology Company in the early stages of commercialization. In particular, our operating cash outflows were mainly attributable to (i) significant research and development expenses, including staff costs for our

– 24 –

SUMMARY

研发及AI基础设施团队以及用于训练和运营我们基础模型所产生的大量第三方云服务及其他计算相关成本，(ii) 我们处于商业化早期阶段，其间我们的AI原生产品和开放平台分别于2023年及2024年才开始产生收入，导致与我们的运营成本结构相比收入基础相对较小，以及 (iii) 其他运营现金流出，包括与员工雇用、营销推广以获取和留存用户以及一般行政费用相关的现金运营成本。上述因素共同导致在整个往绩记录期内产生净运营现金流出。

我们的现金消耗是指以下各项的合计金额：(i) 经营活动所用净现金，(ii) 资本支出，以及 (iii) 租赁付款。我们的历史现金消耗在2022年、2023年、2024年及截至2025年9月30日止九个月分别为1,150万美元、6,590万美元、26,070万美元及21,130万美元。往绩记录期内，随着我们扩大研发活动规模，经营活动所用净现金增加，我们的现金消耗持续上升。未来，我们旨在通过以下重点领域持续提升盈利能力并改善净运营现金流出状况：(i) 把握基础模型行业的快速增长机遇，(ii) 持续提升基础模型的智能水平，(iii) 提高我们AI技术的可负担性，(iv) 拓展我们AI原生产品套件的变现渠道，以及 (v) 优化组织效率与可扩展性。请参阅"业务——我们专业技术产品商业化路径"，了解我们的详细策略。

就截至2025年12月31日止年度而言，我们预计每月现金消耗为2,810万美元。于2025年9月30日，我们的现金结余为104,620万美元（US$1,046.2 million），包括现金及现金等价物36,260万美元（US$362.6 million）、按公允价值计入损益的金融资产流动部分64,420万美元（US$644.2 million）及未动用银行融资额度3,940万美元（US$39.4 million），此等均代表可用于资助我们运营的流动资金。假设未来预期平均每月现金消耗为2,810万美元（约为截至2024年12月31日止十二个月平均每月现金消耗的1.3倍），我们估计在无首次公开招股所得款项的情况下，我们的现金结余足以支撑我们运营约37个月，持续至约2028年10月。在估计首次公开招股所得款项净额46,870万美元（US$468.7 million）的情况下（假设按发售价范围低端每股港币151.0元的发售价发行25,389,220股发售股份，且发售规模调整权及超额配股权均未获行使，并扣除估计首次公开招股费用），加上首次公开招股所得款项后，我们的现金足以支撑我们运营约54个月，持续至约2030年3月。

详情请参阅"财务资料——现金消耗"。

– 25 –

SUMMARY

商业化及业务可持续性

我们专业技术产品的商业化

自成立以来，我们的商业战略围绕两大核心领域：开发先进的基础模型，以及创建能够提升生产力和生活品质的AI原生产品。我们所有的基础模型及AI原生产品均按照《上市规则》第18C章的定义设计为专业技术产品。根据CIC的意见，我们确认所有专业技术产品均属于《上市规则》下人工智能的可接受行业范畴，且往绩记录期内产生的所有收入均来自上述产品的销售。我们进一步确认，所有基础模型及AI原生产品均为自主研发。有关我们主要知识产权归属的描述，请参阅"业务——知识产权"。

• 基础模型：我们已构建多个跨模态的综合基础模型。我们的基础模型套件包括大语言模型、视频生成模型，以及语音和音乐生成模型。

• AI原生产品：依托我们的多模态基础模型套件，我们为全球个人用户和企业用户提供AI原生产品和服务，借助AI的力量为其赋能。我们AI原生产品的演进根植于基础模型的持续进步。通过对现有基础模型的持续升级及新基础模型的开发，我们能够设计和打造具有更佳用户体验的AI原生产品。

SUMMARY

KEY FINANCIAL RATIOS

The following table sets forth certain of our key financial ratios for the periods indicated:

| | For the year ended December 31, | | | For the nine months ended September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | **Key Income Statement Ratio** | | | | | | Revenue growth | N/A | N/A | 782.2% | 174.7% | | Gross margin | N/A | (24.7%) | 12.2% | 23.3% | | Net loss margin | N/A | (7,781.7%) | (1,524.2%) | (958.2%) | | Adjusted net loss margin (non-IFRS measure) | N/A | (2,574.4%) | (800.2%) | (348.6%) | | Research and development expenses growth rate | N/A | 562.9% | 170.0% | 30.0% | | **Key Balance Sheet Ratio** | | | | | | Current ratio | 0.49 | 0.48 | 0.47 | 0.43 |

Note: 1. Current ratio is calculated based on total current assets divided by total current liabilities.

We are still at a nascent stage in terms of monetization and commercialization as historically we have been largely focused on developing our foundation AI models. During the Track Record Period, we have scaled our product offerings and therefore have experienced rapid revenue growth, reflecting our ability to advance proprietary foundation models while rapidly scaling the usage of our AI-native products across individual users, developers, and enterprise customers. Our revenue increased from US$3.5 million in 2023 to US$30.5 million in 2024, as a result of growth momentum from both our developer and enterprise-facing Open Platform and multi-modal AI-native consumer-facing products. For the nine months ended September 30, 2025, our revenue further increased to US$53.4 million, compared to US$19.5 million in the nine months ended September 30, 2024. These gains were driven by the enhancement of the intelligent level of our foundation models, the expansion of our AI-native product suite, increased adoption by individual users, developers, and enterprise customers, and diversified monetization channels across subscriptions, in-app top-up, enterprise API usage, and online marketing services.

As we scaled up operations, we significantly improved our gross profit margin, from negative 24.7% in 2023 to 12.2% in 2024, and further to 23.3% in the nine months ended September 30, 2025. These improvements were primarily driven by advancement in the intelligence level of our models, improved model and system efficiency, optimization of infrastructure allocation, and increased scale of revenue relative to compute intensity, in line with our strategy to enhance efficiency of our AI infrastructure. In particular, gross margin of AI-native products significantly improved during the Track Record Period due to improvements in user engagement and monetization and introduction of new monetized features, reflecting our ongoing commercial emphasis on enhancing monetization from core AI-native product offerings.

We recorded US$73.7 million, US$269.2 million, US$465.2 million, US$304.3 million and US$512.0 million in loss for the year/period in 2022, 2023, 2024, and for the nine months ended September 30, 2024 and 2025, respectively, due to significant initial investment in foundation model R&D and AI infrastructure. Excluding share-based payment expenses, fair value changes in financial instruments and listing expenses, our adjusted net loss (non-IFRS measure) narrowed meaningfully as a percentage of revenue, from over 2,500% in 2023 to 800.2% in 2024 and further to 348.6% in the nine months ended September 30, 2025. Our R&D expenses as a percentage of revenue declined from over 2,000% in 2023 to 619.1% in 2024 and further to 337.4% in the nine months ended September 30, 2025. We expect continued net losses in the foreseeable future as we remain largely an R&D focused company operating in the AI research space.

We recorded net losses in 2022, 2023, 2024, and the nine months ended September 30, 2025. We recorded adjusted net loss (non-IFRS measure) and net operating cash outflow during the Track Record Period. We currently expect such positions may continue until we achieve a greater scale. We anticipate a significant increase in net loss for the year ended December 31, 2025, primarily due to the expected R&D expenses as we continue to elevate the intelligence level of our foundation models and fair value loss on financial liabilities, as the valuation of our company is expected to increase in 2025. In the future, we aim to maintain business sustainability and achieve long-term commercialization through the following focus areas: (i) leveraging the rapid growth of the foundation model industry, (ii) continuing to enhance foundation model intelligence levels, (iii) enhancing the affordability of our AI technologies, (iv) broadening monetization of our AI-native product suite, and (v) optimizing organizational efficiency and scalability.

We expect to achieve the revenue requirement for a Commercial Company pursuant to Chapter 18C of the Listing Rules in the twelve months ended December 31, 2025, primarily driven by commercial expansion of our AI-native products and increasing adoption of our Open Platform. To demonstrate our rapid growth during the Track Record Period and monetization potential, for our AI-native user products, our number of paying users expanded from around 119,700 in 2023 to 650,300 in 2024 and from approximately 489,100 in the nine months ended September 30, 2024 to approximately 1,771,600 during the same period in 2025. For our Open Platform, paying users, defined as a user who has individually consumed no less than US$50 worth of API calls (or its equivalent in other currencies), increased from approximately 400 in the nine months ended September 30, 2024 to approximately 2,500 during the same period in 2025.

Based on the foregoing, our Directors believe, and the Joint Sponsors concur, that our business is sustainable. See "Business — Commercialization and Business Sustainability."

See "Financial Information — Key Financial Ratios" for more information.

APPLICATION FOR LISTING ON THE STOCK EXCHANGE

We have applied to the Stock Exchange for the granting of the listing of, and permission to deal in (a) the Class A Ordinary Shares in issue (including the Class A Ordinary Shares on conversion of the Preferred Shares and the Class B Ordinary Shares issued before Listing) and to be issued pursuant to the Global Offering (including any Class A Ordinary Shares which may be issued pursuant to the exercise of the Offer Size Adjustment Option and the Over-allotment Option), and (b) the Class A Ordinary Shares which may be issued under the Post-IPO Share Incentive Plan. We satisfy the requirements under Rule 18C.03 of the Listing Rules as a Pre-Commercial Company (as defined in the Listing Rules) and Rule 8A.06(1) of the Listing Rules, with reference to our expected market capitalization at the time of Listing, which exceeds HK$40 billion based on the low-end of the indicative Offer Price range.

– 28 –

SUMMARY

GLOBAL OFFERING STATISTICS

All statistics in the following table are based on the assumption that the Offer Size Adjustment Option and the Over-allotment Option are not exercised.

| | Based on an Offer Price of HK$151.00 per Share | Based on an Offer Price of HK$165.00 per Share | |---|---|---| | Market capitalization of our Shares (1) | HK$46,122.54 million | HK$50,398.80 million | | Unaudited pro forma adjusted consolidated net tangible assets per Share (2) | HK$37.96 | HK$39.08 |

(1) The calculation of the market capitalization of our Shares is based on the assumption that 305,447,288 Shares will be in issue and outstanding immediately following the completion of the Global Offering.

(2) The unaudited pro forma adjusted consolidated net tangible assets per Share is calculated after making the adjustments referred to in "Appendix II — Unaudited Pro Forma Financial Information" and on the basis that 305,447,288 shares were in issue assuming that the Global Offering and reclassification of financial liabilities arising from the convertible redeemable preferred shares and ordinary shares into equity had been completed on September 30, 2025, without taking account of the exercise of the Offer Size Adjustment Option and the Over-allotment Option. The unaudited pro forma adjusted consolidated net tangible assets per Share amounts in USD are converted into Hong Kong dollars at USD1.00 = HKD7.7805 prevailing on the Latest Practicable Date.

Listing Expenses

Our listing expenses mainly include (i) underwriting-related expenses, such as underwriting fees and commissions, and (ii) non-underwriting-related expenses, comprising professional fees paid to our legal advisors and reporting accountants for their services rendered in relation to the Listing and the Global Offering, and other fees and expenses. Assuming full payment of the discretionary incentive fee, the estimated total listing expenses (based on the mid-point of the Offer Price range and assuming that the Offer Size Adjustment Option and the Over-allotment Option are not exercised) for the Global Offering are approximately HK$193.2 million, accounting for approximately 4.8% of our gross proceeds. Among such estimated total listing expenses, we expect to pay underwriting-related expenses of HK$133.0 million, professional fees for our legal advisors and reporting accountants of HK$40.3 million and other fees and expenses of HK$19.9 million. During the Track Record Period, the listing expenses charged to our consolidated statements of profit or loss were US$3.7 million (HK$28.6 million) and the issuance costs which were recognized as prepayments and are expected to be deducted from equity upon the Listing, were US$0.4 million (HK$3.3 million). After the Track Record Period approximately HK$26.7 million is expected to be charged to our consolidated statements of profit or loss, and approximately HK$134.7 million is expected to be accounted for as a deduction from equity upon the Listing.

– 29 –

SUMMARY

Future Plans and Use of Proceeds

We estimate that we will receive net proceeds from the Global Offering of approximately HK$3,818.3 million, after deducting underwriting commissions, fees and estimated expenses payable by us in connection with the Global Offering, assuming the Offer Size Adjustment Option or Over-allotment Option is not exercised and an Offer Price of HK$158.00 per Offer Share, being the midpoint of the indicative Offer Price range stated in this Prospectus.

We intend to use the net proceeds of the Global Offering for the following purposes:

• Approximately 90%, or HK$3,436.4 million of the net proceeds will be used for our research and development over the next five years including the development of our foundation models and our AI-native products. Specifically, (i) approximately 70.0%, or HK$2,672.8 million, of the net proceeds over the next five years to the research and development of our foundation models; and (ii) approximately 20.0%, or HK$763.7 million, of the net proceeds over the next five years to the development, refinement and global scaling of our AI-native products.

• Approximately 10.0%, or HK$381.8 million, of the net proceeds will be allocated to working capital and general corporate purposes.

SUMMARY

RECENT DEVELOPMENTS

As of the Latest Practicable Date, we had not experienced any material impact from tariffs, export controls, or other trade-related measures imposed by governmental authorities in the jurisdictions in which we operate. Although certain countries, including the United States, have in recent periods introduced or adjusted tariffs and related policies on goods and technologies, such measures have not had a material adverse effect on our operations, financial condition or prospects to date. We continue to monitor relevant developments and assess their potential implications for our business.

Our Directors confirm that, up to the date of this Prospectus, there has been no material adverse change in our financial or trading position or prospects since September 30, 2025, being the end date of the periods reported in the Accountant's Report set out in Appendix I, and there is no event since September 30, 2025 that would materially affect the information shown in the Accountant's Report set out in Appendix I.

We anticipate a significant increase in net loss for the year ended December 31, 2025, primarily due to the expected R&D expenses as we continue to elevate the intelligence level of our foundation models and fair value loss on financial liabilities, as the valuation of our company is expected to increase in 2025.

RECENT REGULATORY DEVELOPMENT

Outbound Investment Rules

Effective on January 2, 2025, the final rule issued by Treasury to implement the executive order of August 9, 2023 (the "Final Rule") imposes investment prohibition and notification requirements on U.S. Persons for a wide range of investments in entities associated with China (including Hong Kong and Macau) that are engaged in activities relating to three sectors: (i) semiconductors and microelectronics, (ii) quantum information technologies, and (iii) AI systems. U.S. persons subject to the Final Rule are prohibited from making, or required to report, certain investments in covered foreign persons, which are defined as "covered transactions," and include acquisitions of equity interests (including contingent equity interests), certain debt financing, joint ventures, and certain investments as a limited partner in a non-U.S. person pooled investment fund. Since our principal place of business is in China and we engage in the development of certain AI models, we are likely to be deemed as a "covered foreign person" as described in the Final Rule. Based on information we provided to our international sanctions advisor, it appears likely that some U.S. persons that purchase our Shares in the Global Offering or are the parents of non-U.S. person subsidiaries that purchase our Shares in the Global Offering would be required to file notifications regarding their or their subsidiaries' purchases with Treasury no later than 30 days after such purchases of the Shares. See "Risk Factors — Risks Related to Our Business and Industry — We are subject to the risks associated with international trade policies, geopolitics and trade protection measures. Changes in international relationships, trade and investment policies, trade protection and investment restriction measures may adversely impact our business, financial condition and results of operations."

— 31 —

SUMMARY 出口管制法规 近年来，美国通过由美国商务部工业和安全局（"BIS"）管理的《出口管理条例》（"EAR"），不断扩大对中国的出口管制限制。近年来，美国已将越来越多的实体（包括多家中国实体）列入实体清单及其他受限或被禁止的当事方名单。除向上述名单增加新的实体外，BIS还就与名单上的实体开展业务制定了复杂且严格的规则。例如，2025年9月29日，BIS发布了一项立即生效的临时最终规则，将实体清单和军事最终用户名单的限制范围延伸至由名单上的股东直接或间接持有50%或以上股权的实体。美国政府已表示，关联方规则的实施将至少推迟一年（即推迟至2026年10月）。上述近期举措与美国出口管制制度共同规范美国产品、软件及技术的出口、再出口和转让，包括在美国境外制造但含有超过最低限度受控美国内容的特定物项，或属于特定美国软件或技术的境外直接产品的物项。

关税法规 我们亦在密切关注关税政策的潜在变化，并评估此类政策变化对我们业务运营和财务表现的潜在影响。例如，近期美国拟对从多个国家（包括中国）进口的大量商品征收多轮关税，中国亦采取报复性关税措施予以回应。根据我们国际法律顾问的意见，美国进口关税仅适用于向美国出口实物商品的情形。基于此，鉴于我们不向美国出口实物商品，我们的董事认为，美国关税不太可能对我们的业务运营和财务表现产生重大不利影响。由于相关政策正在快速演变，评估这些关税措施未来潜在影响可能存在一定难度。详情请参阅"风险因素——与我们业务及行业相关的风险——我们面临与国际贸易政策、地缘政治及贸易保护措施相关的风险。国际关系、贸易与投资政策、贸易保护及投资限制措施的变化可能对我们的业务、财务状况及经营业绩产生不利影响。"

AI聊天机器人法规 美国多个州（包括加利福尼亚州、纽约州、缅因州和犹他州）近期已颁布专门规范AI驱动聊天机器人的法律。这些法律规定了新的运营要求，例如明确且反复的用户披露义务，并要求实施安全协议以防止有害内容的产生，尤其是与自我伤害相关的内容。例如，加利福尼亚州的相关法律将于2026年1月生效，该法律对未成年人设有专项保护条款，并规定了允许主张法定损害赔偿的私人诉权。

– 32 –

SUMMARY 我们正在审查上述法规，以确保我们的Talkie应用程序持续合规。根据（i）对Talkie现有功能特性的评估，以及（ii）对现行州级聊天机器人立法及法规的研究，并依据我们美国数据法律顾问的意见：（a）Talkie的现有功能和设计已实质性符合目前在犹他州、纽约州和缅因州生效的聊天机器人法律；（b）Talkie不受美国其他州已颁布的聊天机器人法律约束，因为此类法律所规范的是Talkie不提供的功能，例如提供医疗服务。

在我们美国数据法律顾问的法律意见指导下，我们目前正在推进为符合即将生效的加利福尼亚州聊天机器人法律所需的更新工作。这些更新主要涉及审查用户界面、补充特定强制性披露内容以及纳入所要求的安全功能。上述更新与我们针对Talkie的常规产品开发周期相符，预计将在合理的时间框架内以合理的成本完成。我们目前正在设计和推进上述更新，并预计在2025年底前完成，早于加利福尼亚州聊天机器人法律预计于2026年1月生效的日期。根据我们的计划时间表及迄今取得的进展，依据我们美国数据法律顾问的意见，Talkie预计将于该法律2026年1月生效后符合加利福尼亚州聊天机器人法律的相关要求。

鉴于上述我们美国数据法律顾问的意见，尽管遵守上述州级法规可能导致一定的额外运营成本，我们预计此类法规不会对我们的业务、经营业绩或财务状况产生任何重大不利影响。有关我们AI安全与对齐措施、防范不当或有害输出及用户滥用的保障机制，以及我们为确保负责任的AI开发所持续作出的努力的更多详情，请参阅"业务——研究与开发"一节。

– 33 –

释义

在本招股说明书中，除文义另有所指外，下列词语具有以下所列含义。若干其他词语的释义载于本招股说明书"技术术语词汇表"一节。

"会计师报告"　　　　　　本公司的会计师报告，其全文载于本招股说明书附录一

"关联方"　　　　　　　　就任何特定人士而言，指直接或间接控制该特定人士、受该特定人士控制或与该特定人士处于直接或间接共同控制下的任何其他人士

"AFRC"

Accounting and Financial Reporting Council (會計及財務匯報局)

"Alpha EXP"

Alpha EXP Limited, a business company incorporated in the BVI on November 23, 2021, and one of our Controlling Shareholders

"Articles" or "Articles of Association"

the amended and restated articles of association of our Company with effect upon the Listing Date (as amended from time to time), a summary of which is set out in Appendix III to this Prospectus

"associate(s)"

has the meaning ascribed thereto under the Listing Rules

"Audit Committee"

the audit committee of the Board

"Beijing Jizhi"

Beijing Xiyu Jizhi Technology Co., Ltd. (北京稀宇極智科技有限公司) (formerly known as Mingri Zhimeng (Beijing) Technology Co., Ltd. (名日之夢(北京)科技有限公司)), a limited liability company established in the PRC on November 18, 2021 and a wholly-owned subsidiary of the Company

"Board", "Board of Directors" or "our Board"

the board of Directors of the Company

"Business Day"

a day on which banks in Hong Kong are generally open for normal business to the public and which is not a Saturday, Sunday or public holiday in Hong Kong

"BVI"

the British Virgin Islands

– 34 –

DEFINITIONS

"Capital Market Intermediaries" or "capital market intermediary(ies)" or "CMI(s)"

the capital market intermediaries participating in the Global Offering and has the meaning ascribed thereto under the Listing Rules

"CCASS"

the Central Clearing and Settlement System established and operated by HKSCC

"China" or "the PRC"

the People's Republic of China, unless the context requires otherwise, excluding, for the purposes of this Prospectus only, the regions of Hong Kong, Macau and Taiwan of the People's Republic of China

"CIC"

China Insights Industry Consultancy Limited (灼識行業諮詢有限公司), an independent professional market research and consulting company

"Circular 37"

the Notice of the SAFE on Issues Concerning Foreign Exchange Administration of the Overseas Investment and Financing and the Round-Tripping Investment Made by Domestic Residents through Special-Purpose Companies (《國家外匯管理局關於境內居民通過特殊目的公司境外投融資及返程投資外匯管理有關問題的通知》)

"Class A Ordinary Shares"

Class A ordinary shares in the share capital of the Company with a par value of US$0.0001 each, conferring a holder of a Class A ordinary share one vote per share on all matters subject to the vote at general meetings of the Company

"Class B Ordinary Shares"

Class B ordinary shares in the share capital of the Company with a par value of US$0.0001 each, conferring weighted voting rights in the Company such that a holder of a Class B ordinary share is entitled to ten votes per share on all matters subject to the vote at general meetings of the Company, subject to the requirements under Rule 8A.24 of the Hong Kong Listing Rules that the Reserved Matters shall be voted on a one vote per share basis

"close associate(s)"

has the meaning ascribed thereto under the Listing Rules

– 35 –

DEFINITIONS

"Companies (Winding Up and Miscellaneous Provisions) Ordinance"

the Companies (Winding Up and Miscellaneous Provisions) Ordinance (Chapter 32 of the Laws of Hong Kong), as amended, supplemented or otherwise modified from time to time

"Companies Act" or "Cayman Companies Act"

the Companies Act (As Revised) of the Cayman Islands, as amended or supplemented or otherwise modified from time to time

"Companies Ordinance"

the Companies Ordinance (Chapter 622 of the Laws of Hong Kong), as amended, supplemented or otherwise modified from time to time

"Company", "our Company", or "the Company"

MiniMax Group Inc., an exempted company with limited liability incorporated under the laws of the Cayman Islands on June 30, 2021

"Compliance Adviser"

Somerley Capital Limited

"connected person(s)"

has the meaning ascribed thereto under the Listing Rules

"connected transaction(s)"

has the meaning ascribed thereto under the Listing Rules

"Controlling Shareholders"

has the meaning ascribed thereto under the Listing Rules and unless the context requires otherwise, refers to Dr. Yan, Local Linearity, Alpha EXP, MiniMax Matrix, MiniMax Limited, Scaling EXP Limited, and MiniMax Awakening, being the members constituting the group of our Controlling Shareholders

"core connected person(s)"

has the meaning ascribed thereto under the Listing Rules

"Corporate Governance Code"

the Corporate Governance Code set out in Appendix C1 to the Listing Rules

"Corporate Governance Committee"

the corporate governance committee of the Board

"CSRC"

China Securities Regulatory Commission (中國證券監督管理委員會)

"Director(s)"

the director(s) of our Company

– 36 –

DEFINITIONS

"Dr. Yan"

Dr. Yan Junjie (閆俊傑), the founder, the chairman of the Board, chief executive officer, chief technology officer and one of our Controlling Shareholders

"EIT"

enterprise income tax

"EIT Law"

the PRC Enterprise Income Tax Law (《中華人民共和國企業所得稅法》)

"ESG"

Environmental, Social and Governance

"Exchange Participant"

a person (a) who, in accordance with the Hong Kong Listing Rules, may trade on or through the Hong Kong Stock Exchange; and (b) whose name is entered in a list, register or roll kept by the Hong Kong Stock Exchange as a person who may trade on or through the Hong Kong Stock Exchange

"Extreme Conditions"

extreme conditions caused by a super typhoon as announced by the government of Hong Kong

"FINI"

"Fast Interface for New Issuance", the online platform operated by HKSCC that is mandatory for admission to trading and, where applicable, the collection and processing of specified information on subscription in and settlement for the Listing

"Floating Sky"

Floating Sky Limited (formerly known as Sapiens Origin Limited), a business company incorporated in the BVI on November 23, 2021

"Global Offering"

the Hong Kong Public Offering and the International Offering

"Group", "our Group", "our", "we" or "us"

the international underwriting agreement dated December 30, 2025, relating to the International Offering, entered into by, among others, our Company, the Controlling Shareholders and the Overall Coordinators, as further described in the section headed "Underwriting — Underwriting Arrangements and Expenses — International Offering — International Underwriting Agreement"

"JCEF"

Jingchen Equity Fund (景辰股权投资基金合伙企业 (有限合 伙)), incorporated in Hangzhou, China, under the laws of the PRC as a limited partnership, which is one of the PreIPO Investors

"JD Group"

JD.com, Inc. (京东集团股份有限公司), a company listed on the Nasdaq Stock Exchange and the Hong Kong Stock Exchange with ticker symbols "JD" and "9618.HK" respectively, together with its subsidiaries

"JD Health"

JD Health International Inc. (京东健康股份有限公司), a company listed on the Hong Kong Stock Exchange with ticker symbol "6618.HK", and a subsidiary of JD Group

"JDIH"

JD Industrial Holdings Ltd. (京东工业股份有限公司), a company incorporated in the Cayman Islands, and a subsidiary of JD Group

"JDIH ESOP Platform"

Jingdong Xingchen Industrial Investment (Suzhou) Partnership (Limited Partnership) (京东星辰工业投资(苏 州)合伙企业(有限合伙)), incorporated in Suzhou, China, under the laws of the PRC as a limited partnership, which is one of the Pre-IPO Investors

"JDIH RMB Fund"

Jingdong Industrial (Suzhou) Investment Fund Partnership (Limited Partnership) (京东工业(苏州)投资基 金合伙企业(有限合伙)), incorporated in Suzhou, China, under the laws of the PRC as a limited partnership, which is one of the Pre-IPO Investors

"JDIH Shares"

the shares in the share capital of JDIH

– 41 –

DEFINITIONS

"JD Industrial"

JD Industrial Technology Co., Ltd. (京东工业技术有限公 司), a company incorporated under the laws of the PRC, and a wholly-owned subsidiary of JDIH

"Jinan Jingdong"

Jinan Jingdong Jinchen Industrial Internet Technology Co., Ltd. (济南京东金辰工业互联网科技有限公司), a company incorporated under the laws of the PRC, and an indirect subsidiary of our Company

"Kunlun Wanwei"

Kunlun Tech Co., Ltd. (昆仑万维科技股份有限公司), a company listed on the Shenzhen Stock Exchange with ticker symbol "300418"

"Latest Practicable Date"

December 19, 2025, being the latest practicable date prior to the printing of this prospectus for the purpose of ascertaining certain information contained herein

"Listing"

the listing of the Shares on the Main Board of the Stock Exchange

"Listing Committee"

the Listing Committee of the Stock Exchange

"Listing Date"

the date on which the Shares are listed and on which dealings in the Shares first commence on the Stock Exchange

"Listing Rules"

the Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited, as amended, supplemented or otherwise modified from time to time

"M&E"

the mechanical and electrical systems

"Main Board"

the main board of the Stock Exchange

"MIIT"

the Ministry of Industry and Information Technology of China (中华人民共和国工业和信息化部)

– 42 –

DEFINITIONS

"MOC"

the Ministry of Commerce of the PRC (中华人民共和国 商务部)

"MOF"

the Ministry of Finance of the PRC (中华人民共和国财政 部)

"MOFCOM"

the Ministry of Commerce of the PRC (中华人民共和国 商务部)

"MPS"

the Ministry of Public Security of the PRC (中华人民共 和国公安部)

"NDRC"

the National Development and Reform Commission of the PRC (中华人民共和国国家发展和改革委员会)

"Nomination Committee"

the nomination committee of the Board

"Non-Voting Shares"

non-voting preference shares in the share capital of our Company

"Offer Price"

the final offer price per Offer Share (exclusive of brokerage of 1.0%, SFC transaction levy of 0.0027%, AFRC transaction levy of 0.00015% and Stock Exchange trading fee of 0.00565%) to be determined in the manner described in "Structure of the Global Offering — Pricing and Allocation"

"Offer Price Range"

the indicative range of the Offer Price being HK$37.58 to HK$41.58 per Offer Share

"Offer Share(s)"

the Hong Kong Offer Shares and the International Offer Shares

"Offer Size Adjustment Option"

the option to be granted by our Company to the Overall Coordinators (for themselves and on behalf of the International Underwriters), exercisable by the Overall Coordinators (for themselves and on behalf of the International Underwriters) at any time within 30 days from the last day for lodging applications under the Hong Kong Public Offering

– 43 –

DEFINITIONS

to require our Company to allot and issue up to an aggregate of 3,817,380 additional Class A Ordinary Shares at the Offer Price, representing up to 15% of the initial number of International Offer Shares under the International Offering, if the Over-allotment Option is not exercised, or up to 3,817,380 Class A Ordinary Shares (representing 15% of the initial number of International Offer Shares offered under the International Offering), reduced by the number of Class A Ordinary Shares actually repurchased by the Stabilizing Manager under the Over-allotment Option, if the Over-allotment Option is exercised

"Overall Coordinators"

Goldman Sachs (Asia) L.L.C., J.P. Morgan Securities (Far East) Limited and BofA Securities (Asia Pacific) Limited

"Over-allotment Option"

the option to be granted by our Company to the Stabilizing Manager, exercisable by the Stabilizing Manager at any time during the Stabilization Period, to require our Company to allot and issue up to an aggregate of 3,817,380 additional Class A Ordinary Shares at the Offer Price, representing up to 15% of the initial number of International Offer Shares offered under the International Offering, solely to cover over-allocations in the International Offering, if any

"PBOC"

the People's Bank of China (中国人民银行)

"PRC" or "China"

the People's Republic of China, and for the purposes of this prospectus, excluding Hong Kong, Macau and Taiwan

"PRC Legal Advisors"

Commerce & Finance Law Offices (通商律师事务所), our legal advisors on PRC laws

"Pre-IPO Investors"

the investors who subscribed for the shares in our Company or JDIH before the Global Offering, as further described in the section headed "History, Reorganization and Corporate Structure — Pre-IPO Investments"

– 44 –

DEFINITIONS

"Prospectus"

this prospectus

"QIBs"

"qualified institutional buyers" as defined in Rule 144A

"RMB" or "Renminbi"

Renminbi, the lawful currency of the PRC

"Rule 144A"

Rule 144A under the US Securities Act

"SAFE"

the State Administration of Foreign Exchange of the PRC (国家外汇管理局)

"SAT"

the State Administration of Taxation of the PRC (国家税 务总局)

"SCNPC"

the Standing Committee of the National People's Congress of the PRC (全国人民代表大会常务委员会)

"SFC"

the Securities and Futures Commission of Hong Kong

"SFO"

the Securities and Futures Ordinance (Chapter 571 of the Laws of Hong Kong), as amended, supplemented or otherwise modified from time to time

"Share(s)"

ordinary shares in the share capital of our Company, including the Class A Ordinary Shares and the Class B Ordinary Shares

"Shareholder(s)"

holder(s) of our Shares

"Share Registrar"

Maples Fund Services (Cayman) Islands Limited

"Stabilization Period"

the period commencing on the Listing Date and ending on the 30th day after the last day for the lodging of applications under the Hong Kong Public Offering

"Stabilizing Manager"

Goldman Sachs (Asia) L.L.C.

– 45 –

DEFINITIONS

"State Council"

the State Council of the PRC (中华人民共和国国务院)

"Substantial Shareholder(s)"

has the meaning given to it under the Listing Rules

"Supervisory Board"

the supervisory board of JDIH, as further described in the section headed "Directors, Supervisory Board, Senior Management and Employees"

"Track Record Period"

the period comprising the years ended December 31, 2022, 2023 and 2024 and the six months ended June 30, 2025

"US Securities Act"

the United States Securities Act of 1933 (as amended)

"US$" or "U.S. dollars"

United States dollars, the lawful currency of the United States

"VIE" or "Consolidated Affiliated Entity"

Yitong Wupin (Tianjin) Technology Co., Ltd. (易通物品 (天津)科技有限公司), a limited liability company incorporated under the laws of the PRC

"VIE Agreements" or "Contractual Arrangements"

the contractual arrangements entered into among our Company's PRC subsidiary(ies), the VIE and the shareholder(s) of the VIE, by virtue of which our Company is able to exercise effective control over the VIE and is entitled to consolidate the VIE's financial results into our consolidated financial statements

"WFOE" or "Wholly Foreign- Owned Enterprise"

Jingdong Industrial Technology (Chengdu) Co., Ltd. (京东工业科技(成都)有限公司), a wholly foreign-owned enterprise established under the laws of the PRC, and a wholly-owned subsidiary of JDIH

– 46 –

the post-IPO share incentive plan adopted by our Company on December 29, 2025 and described in the section headed "Statutory and General Information — Post-IPO Share Incentive Plan" in Appendix IV to this Prospectus

"Pre-IPO Share Incentive Plan"

the pre-IPO share incentive plan adopted by our Company as described in the section headed "Statutory and General Information — Pre-IPO Share Incentive Plan" in Appendix IV to this Prospectus

"Prospectus"

this prospectus issued by our Company in connection with the Global Offering

"PRC" or "China"

the People's Republic of China, which for the purpose of this Prospectus, excludes Hong Kong, Macau and Taiwan

"PRC Legal Advisors"

Han Kun Law Offices (韓坤律師事務所), our legal advisors as to PRC law in connection with the Global Offering

"Pre-IPO Investors"

investors who subscribed for Shares in our Company prior to the Global Offering

– 44 –

DEFINITIONS

"Regulation S"

Regulation S under the U.S. Securities Act

"Remuneration Committee"

the remuneration committee of the Board

"Reserved Shares"

the 152,334 Shares initially reserved for issuance upon exercise of options to be granted under the Post-IPO Share Incentive Plan

"RMB" or "Renminbi"

Renminbi, the lawful currency of the PRC

"SAFE"

the State Administration of Foreign Exchange of the PRC (中华人民共和国国家外汇管理局)

"SAT"

the State Administration of Taxation of the PRC (中华 人民共和国国家税务总局)

"SCNPC"

the Standing Committee of the National People's Congress of the PRC (中华人民共和国全国人民代表大 会常务委员会)

"SFC"

the Securities and Futures Commission of Hong Kong (香港证券及期货事务监察委员会)

"SFO"

the Securities and Futures Ordinance (Chapter 571 of the Laws of Hong Kong) (香港法律第571章《证券及期货 条例》), as amended, supplemented or otherwise modified from time to time

"Share(s)"

ordinary share(s) in the capital of our Company with a par value of US$0.00001 each

"Share Registrar"

Computershare Hong Kong Investor Services Limited

"Shareholder(s)"

holder(s) of our Share(s)

"Stabilizing Manager"

Morgan Stanley Asia Limited

"Stock Exchange" or "Hong Kong Stock Exchange"

The Stock Exchange of Hong Kong Limited (香港联合交 易所有限公司)

– 45 –

DEFINITIONS

"subsidiary(ies)"

has the meaning ascribed to it under the Listing Rules

"Substantial Shareholder(s)"

has the meaning ascribed to it under the Listing Rules

"Supervisor"

a supervisor of our PRC Operating Entity or our PRC Subsidiary, as applicable

"Track Record Period"

the period comprising the three financial years ended December 31, 2022, 2023 and 2024, and the six months ended June 30, 2025

"U.S." or "United States"

the United States of America, its territories, its possessions and all areas subject to its jurisdiction

"U.S. Securities Act"

the United States Securities Act of 1933, as amended

"USD" or "US$"

United States dollars, the lawful currency of the United States

"VIE Agreements"

the contractual arrangements entered into among our wholly foreign-owned entity, the PRC Operating Entity and/or the registered shareholder(s) of the PRC Operating Entity, details of which are set out in the section headed "Corporate Structure — VIE Structure and Contractual Arrangements" in this Prospectus

"VIE Structure"

the variable interest entity structure adopted by our Company as described in the section headed "Corporate Structure — VIE Structure and Contractual Arrangements" in this Prospectus

"WFOE" or "Wholly ForeignOwned Entity"

MiniMax Intelligence Technology (Shanghai) Co., Ltd. (名之极(上海)智能科技有限公司), a wholly foreignowned enterprise established in China, being a whollyowned subsidiary of MiniMax HongKong

– 46 –

DEFINITIONS

"White Form eIPO"

the application for Hong Kong Offer Shares to be issued in the applicant's own name by submitting applications online through the White Form eIPO Service Provider

"White Form eIPO Service Provider"

Computershare Hong Kong Investor Services Limited

Unless otherwise specified, for ease of reference, this Prospectus uses "MiniMax (名之极 (上海)科技有限公司)" to refer to the consolidated entity of MiniMax AI Ltd. (our Company) and its subsidiaries and affiliated entities, as the context requires.

Unless otherwise specified, this Prospectus uses "MiniMax Group (MiniMax集团)" to refer collectively to MiniMax (名之极(上海)科技有限公司) and its subsidiaries and affiliated entities (other than our Company and the entities controlled by our Company through the VIE Structure).

– 47 –

the U.S. Securities Act of 1933, as amended

"Underwriter(s)"

the Hong Kong Underwriters and the International Underwriters

"Underwriting Agreements"

the Hong Kong Underwriting Agreement and the International Underwriting Agreement

"United States" or "U.S."

the United States of America, its territories, its possessions and all areas subject to its jurisdiction

"USD", "US$" or "U.S. dollar(s)"

the lawful currency of the United States

"VIE"

variable interest entity

"VIE Agreements"

the contractual arrangements entered into by our Group members, the relevant PRC entities and their registered shareholders in connection with the VIE structure of our Group, as detailed in "Our Holding Structure and Corporate Governance"

"VIE Structure"

the variable interest entity structure of our Group, as detailed in "Our Holding Structure and Corporate Governance"

– 47 –

DEFINITIONS

"Weighted Voting Rights" or "WVR"

weighted voting rights attached to Class B Ordinary Shares, which entitle each Class B Ordinary Share to 10 votes on all resolutions tabled at general meetings of the Company, other than the Reserved Matters (as defined in this Prospectus), pursuant to which each Share is entitled to one vote, including a resolution to approve any of the Reserved Matters

"WVR Beneficiary(ies)"

the holder(s) of Class B Ordinary Shares who are entitled to the Weighted Voting Rights, details of which are set out in the section headed "Our Holding Structure and Corporate Governance — Weighted Voting Rights — WVR Beneficiaries" in this Prospectus

"%" or "percent"

per cent

Certain amounts and percentage figures included in this Prospectus have been subject to rounding adjustments. Accordingly, figures shown as totals in certain tables may not be an arithmetic aggregation of the figures preceding them.

– 48 –

GLOSSARY OF TECHNICAL TERMS

This glossary contains explanations of certain technical terms used in this Prospectus in connection with our Group and our business. The terms and their meanings may not correspond to standard industry meanings or usage of these terms.

"AI"

artificial intelligence

"AI Agent"

an AI system capable of autonomously perceiving its environment, planning, and executing tasks

"AI native application"

an application built from the ground up with AI models as its core

"AIGC"

AI-generated content

"API"

Application Programming Interface; a set of protocols that enables software applications to communicate with each other

"ASR"

Automatic Speech Recognition; technology that converts spoken language into text

"benchmark"

a standardized test or set of tests used to evaluate and compare the performance of AI models

"character card"

a structured profile or template that defines the personality, backstory, and behavioral guidelines of an AI character or persona

– 49 –

GLOSSARY OF TECHNICAL TERMS

"chatbot"

a software application designed to simulate conversation with human users, especially over the internet

"context window"

the maximum amount of text or information (measured in tokens) that an AI model can process and consider at one time

"DAU"

Daily Active Users; the number of unique users who engage with a platform or product on a given day

"deep research"

an advanced AI capability that enables a model to autonomously conduct comprehensive, multi-step research on complex topics

"diffusion model"

a generative AI model that creates outputs by iteratively refining from random noise toward a target distribution

"emotion AI"

a branch of AI focused on recognizing, interpreting, and simulating human emotions; includes affective computing

"end-to-end training"

a machine learning approach where a model is trained directly from raw inputs to outputs, optimizing all components together rather than separately

"fine-tuning"

a technique where a pre-trained model is further trained on a specific dataset to specialize its capabilities for particular tasks

– 50 –

GLOSSARY OF TECHNICAL TERMS

"flow matching"

a generative modeling technique that trains models by learning to transform one probability distribution into another through continuous flows

"foundation model"

a large-scale AI model trained on broad data that can be adapted to a wide range of downstream tasks

"FPS"

frames per second; a measure of video or animation smoothness

"GAN"

Generative Adversarial Network; a machine learning framework consisting of a generator and discriminator trained in opposition to produce realistic synthetic data

"GPU"

Graphics Processing Unit; specialized processors originally designed for rendering graphics but now widely used for training and running AI models

"in-context learning"

an AI model's ability to adapt to new tasks based on examples or instructions provided within the prompt, without updating model parameters

"inference"

the process of using a trained AI model to generate outputs or make predictions from new inputs

"instruction fine-tuning"

a fine-tuning method where models are trained on datasets of instruction-response pairs to improve their ability to follow user instructions

– 51 –

GLOSSARY OF TECHNICAL TERMS

"LLM"

Large Language Model; a type of AI model trained on vast amounts of text data that can understand and generate human language

"long-context"

referring to an AI model's capability to process and generate coherent responses over extended lengths of text or conversation, often thousands to millions of tokens

"MAU"

Monthly Active Users; the number of unique users who engage with a platform or product within a given month

"MiniMax-01"

MiniMax's flagship foundation model series, with its most capable variant being MiniMax-Text-01 for text understanding and generation and MiniMax-Voice-01 for voice understanding and generation

"MiniMax-M1"

MiniMax's latest open-source reasoning model released in May 2025

"MiniMax-Text-01"

MiniMax's text foundation model, with a context window of up to 1 million tokens

"MiniMax-Voice-01"

MiniMax's voice foundation model that deeply integrates voice and language model capabilities into a single unified model, enabling seamless understanding and generation across both modalities

"MoE"

Mixture of Experts; a neural network architecture that routes inputs to specialized subnetworks (experts) to improve efficiency and scalability

– 52 –

GLOSSARY OF TECHNICAL TERMS

"multimodal"

referring to AI models or systems capable of processing and generating multiple types of data, such as text, images, audio, and video

"NLP"

Natural Language Processing; a field of AI focused on enabling machines to understand and generate human language

"open-source model"

an AI model whose weights and/or architecture are publicly available, allowing anyone to use, modify, and distribute it

"parameters"

numerical values within an AI model that are learned during training and define the model's behavior; often used as a measure of model scale

"prompt"

the input text or instructions given to an AI model to guide its output

"RAG"

Retrieval-Augmented Generation; a technique that combines AI generation with real-time retrieval of relevant external information to improve output accuracy and factuality

"reasoning model"

an AI model specifically designed or trained to perform complex logical reasoning and multi-step problem-solving

"reinforcement learning"

a machine learning paradigm where an agent learns to make decisions by receiving rewards or penalties based on its actions

– 53 –

GLOSSARY OF TECHNICAL TERMS

"RLHF"

Reinforcement Learning from Human Feedback; a technique that trains AI models using human evaluations of their outputs to better align them with human preferences

"roleplay"

an interactive experience where users engage with AI characters in fictional or real-world scenarios

"RPM"

requests per minute; a metric measuring API throughput

"scaling law"

empirical relationships in AI research showing that model performance improves predictably with increases in model size, dataset size, and compute

"social entertainment"

digital entertainment experiences centered on social interaction, including chatting, roleplaying, and collaborative storytelling with AI characters

"speech synthesis"

the artificial generation of human speech from text input, also known as Text-to-Speech (TTS)

"system prompt"

a special type of prompt given to an AI model to define its behavior, persona, or constraints before user interaction begins

"T2V"

Text-to-Video; an AI capability that generates video content from text descriptions

"token"

the basic unit of text processed by a language model; tokens can represent words, subwords, or characters

– 54 –

GLOSSARY OF TECHNICAL TERMS

"TPU"

Tensor Processing Unit; custom AI accelerator chips developed by Google, designed to accelerate machine learning workloads

"training"

the process of exposing an AI model to large amounts of data and adjusting its parameters to minimize prediction errors and improve performance

"Transformer"

a neural network architecture that uses self-attention mechanisms to process sequential data; the foundational architecture behind most modern LLMs

"TTS"

Text-to-Speech; technology that converts written text into spoken audio output

"video generation model"

an AI model capable of producing video content, either from text, images, or other inputs

"video understanding"

an AI capability that enables models to analyze, interpret, and respond to video content

"vocal cloning"

a technology that enables AI to replicate a specific person's voice characteristics from a short audio sample

"voice interaction"

a mode of engaging with AI systems using spoken language for both input and output

– 55 –

United States Securities Act of 1933, as amended, supplemented or otherwise modified from time to time

"Underwriters"

the Hong Kong Underwriters and the International Underwriters

"Underwriting Agreements"

the Hong Kong Underwriting Agreement and/or the International Underwriting Agreement, as the context may require

"United States", "USA" or "U.S."

the United States of America, its territories and possessions, any State of the United States, and the District of Columbia

"USD", "US$" or "U.S. dollars"

United States dollar, the lawful currency of the United States

"VAT"

value-added tax

– 47 –

DEFINITIONS "WVR Beneficiary(ies)"

has the meaning ascribed to it under the Hong Kong Listing Rules and unless the context otherwise requires, refers to each of Dr. Yan and Ms. Yun, being the holder of the Class B Ordinary Shares upon Listing

"WVR structure"

has the meaning ascribed to it under the Hong Kong Listing Rules

"%"

per cent

– 48 –

技术术语词汇表

本技术术语词汇表包含对本招股说明书中所用若干技术术语的释义。因此，此类术语及其含义未必与该等术语的标准行业含义或用法相符。

"AI"

人工智能，计算机科学的一个分支，旨在开发能够执行通常需要人类智能才能完成的任务的系统，例如感知、学习、推理和决策

"AIGC"

人工智能生成内容，指由人工智能技术自动创建（无需人类直接创作）的文本、视觉及音频等内容

"AI agent" （AI智能体）

能够自主代表用户执行特定任务以实现既定目标的智能系统

"AI Infrastructure" （AI基础设施）

训练和推理AI模型所需的硬件、软件、数据系统或云资源的集成体系

"AI music synthesis model" （AI音乐合成模型）

一种通过学习现有音乐数据中的规律来生成或创作音乐的人工智能系统，能够创作原创旋律、和声及节奏

"AI-native" （AI原生）

指从底层架构开始即以人工智能为核心组件构建的产品、服务或系统，而非将AI作为附加功能进行集成。AI原生解决方案旨在于其基础架构和功能中充分利用AI技术

"AI-powered Multi-modal Entertainment Platform" （AI驱动多模态娱乐平台）

利用人工智能使智能体能够处理多种类型输入并作出响应的平台，从而实现更自然、更多样化的交互

"API"

一套允许不同软件系统相互通信和交互的规则与工具

"API Platform" （API平台）

提供标准化接口的系统，允许开发者将公司的服务或数据接入其自有应用程序

– 49 –

技术术语词汇表

"app"或"application" （应用程序）

专为智能手机及其他移动设备运行而设计的应用软件

"Asia-Pacific" （亚太地区）

由东亚、南亚、东南亚及大洋洲国家组成的地理区域，通常包括澳大利亚和新西兰

"attention" （注意力机制）

一种使基础模型在处理信息时能够衡量输入数据不同部分重要性的复杂机制

"Audio Generation Tool" （音频生成工具）

利用AI根据用户输入或预定义参数创建合成音频内容（包括语音、音乐或音效）的软件

"auto-regressive model" （自回归模型）

一种利用时间序列的历史值来预测该时间序列未来值的统计模型

"B2B"或"2B"或"ToB"

"企业对企业"，指企业之间的商业交易或关系，而非企业与个人消费者之间的交易或关系

"B2C"或"2C"或"ToC"

"企业对消费者"，指企业与个人消费者之间的商业交易或关系

"benchmark" （基准测试）

用于衡量和比较语言模型性能的标准化评估框架

"CAGR"

复合年均增长率

"Chat Completions API" （对话补全API）

一种应用程序编程接口，允许开发者通过以对话形式生成AI驱动的响应（通常基于大型语言模型）来构建交互式聊天体验

"chain-of-thought"或"CoT" （思维链）

一种推理技术，模型通过生成中间推理步骤或解释来解决复杂问题

"Diffusion Model" （扩散模型）

机器学习中一种生成模型，擅长通过逐步向数据点添加噪声并学习逆转该过程来生成高质量数据（尤其是图像和文本）

– 50 –

技术术语词汇表

"DiT"

带变换器的扩散模型，一种以变换器架构为骨干网络用于图像生成的扩散模型

"ELO"

一种用于评估竞技游戏或活动中参与者相对技能水平的评分系统，分数越高表示表现越强

"ESG"

环境、社会及治理

"fine-tuning" （微调）

将预训练基础模型调整以更高精度和相关性执行特定任务或专注于特定领域的过程

"Flow-VAE"

一种结合变分自编码器与归一化流的混合AI模型，通过更有效地捕捉复杂数据分布来提升生成数据的质量和灵活性

"foundation model" （基础模型）

基于广泛多样数据集开发的大规模预训练模型，旨在作为通用模型用于解决各类任务

"fps"

每秒帧数，衡量视频或动画中每秒显示或生成的单帧（图像）数量。帧率越高，动作越流畅

"freemium"

GLOSSARY OF TECHNICAL TERMS

"model call"

a single instance of using an AI model to process an input and generate an output, typically initiated by a user action or automated request

"model training"

the process of feeding large amounts of data to an AI model and adjusting its internal parameters to minimize prediction errors, resulting in a model capable of performing specific tasks

"multimodal" or "multimodal AI"

AI systems that can process and generate multiple types of data, such as text, images, audio, and video, enabling more comprehensive and versatile interactions

"MiniMax app"

our AI-native application that provides users with access to our foundation models and AI capabilities through a conversational interface

"NLP"

Natural Language Processing, a branch of artificial intelligence that focuses on enabling computers to understand, interpret, and generate human language

"open-source model"

an AI model whose underlying code, architecture, and often weights are made publicly available, allowing developers to use, modify, and distribute the model freely

"parameters"

numerical values within an AI model that are learned during training and determine how the model processes inputs to generate outputs; models with more parameters generally have greater capacity to learn complex patterns

"PRC"

the People's Republic of China

"prompt"

an input or instruction provided to an AI model to guide its output generation, such as a question, command, or descriptive text

"reasoning model"

an AI model specifically designed or trained to perform multi-step logical reasoning, problem-solving, and complex inference tasks

"reinforcement learning"

a machine learning technique where an AI model learns by receiving feedback in the form of rewards or penalties based on its actions, enabling it to improve performance over time

"reinforcement learning from human feedback" or "RLHF"

a training technique that uses human evaluations of model outputs as feedback signals to guide reinforcement learning, helping align AI model behavior with human preferences and values

"RMB" or "Renminbi"

the lawful currency of the PRC

"robot" or "AI robot"

an AI-powered conversational agent or virtual character that can engage in natural language interactions with users

"role-play"

a type of interactive experience where users engage with AI characters or personas in fictional or simulated scenarios

"Softmax Attention"

the traditional attention mechanism used in transformer models, which computes attention scores using the Softmax function; while highly accurate, it has quadratic computational complexity relative to sequence length, making it resource-intensive for long sequences

"speech-to-text"

a technology that converts spoken audio into written text

"T2I"

text-to-image, a technology or model that generates images from textual descriptions

"T2V"

text-to-video, a technology or model that generates video content from textual descriptions

"text-to-speech" or "TTS"

a technology that converts written text into spoken audio output

"token"

a basic unit of text (such as a word or part of a word) used by AI language models to process and generate text; model usage and pricing are often measured in terms of the number of tokens processed

"transformer"

a neural network architecture that uses attention mechanisms to process sequential data, forming the foundation of most modern large language models

"user-generated content" or "UGC"

content created and shared by users of a platform, rather than by the platform operator itself

"video-to-video" or "V2V"

a technology or model that transforms or modifies existing video content to generate new video sequences

"virtual human"

a computer-generated character with human-like appearance and behavior, capable of interacting with users through speech, expressions, and gestures

"voice cloning"

a technology that creates a synthetic replica of a specific person's voice, enabling AI to generate speech that sounds like that individual

"web scraping"

the automated process of extracting data from websites

– 54 –

GLOSSARY OF TECHNICAL TERMS

"MoE" or "Mixture-of-Experts architecture"

混合专家架构，一种AI模型架构，使用一组子网络或"专家"，其中每次输入仅激活其中的一个子集。这使模型能够根据任务性质分配计算资源，从而高效扩展，在保持效率的同时提升性能

"Multi-Modal Model Suite"

多模态模型套件，一套旨在以集成方式处理和理解多种类型数据输入（如文本、音频和视频）的AI模型集合，能够实现更全面、更接近人类的感知与交互

"Open Platform"

开放平台，一种公开可访问的接口，通过提供标准化编程访问，允许第三方开发者与公司的系统或服务集成，从而基于该平台的能力开发应用程序或服务

"open-source"

开源，将软件源代码公开的做法，允许任何人在通常遵循开源许可证的条件下查看、使用、修改和分发该代码

"p"

像素，表示数字图像或显示器中单个点的单位，通常用于衡量屏幕分辨率或图像尺寸

"parameters"

参数，基础模型在训练过程中学习和调整的内部数值变量或配置。这些参数有效地封装了模型从其所训练的大量数据集中提取的知识、模式和关系

"Q&A"

问答

"R&D"

研究与开发

"reasoning model"

推理模型，一种旨在模拟逻辑思维的AI模型，能够基于输入数据进行推断、做出决策或解决问题，通常跨越多个步骤或情境

– 55 –

GLOSSARY OF TECHNICAL TERMS

"reinforcement learning from human feedback" or "RLHF"

基于人类反馈的强化学习，通过人类示范和偏好数据对模型进行强化训练，使模型能够遵循指令并生成详尽、相关的回应

"Scaling Law"

规模定律，一种描述模型性能随模型规模、数据集规模和计算资源增加而系统性提升的经验性规律

"SDK"

软件开发工具包，开发者用于为特定平台、系统或服务构建应用程序的工具、库、文档和代码示例的集合

"semantic space"

语义空间，一种将词语或概念的含义表示为多维空间中点的方式，其中各维度代表语义特征或情境。这是自然语言处理及相关领域中使用的一种计算方法，用于根据词语和概念的含义及情境对其关系进行建模

"SLA"

服务级别协议，服务提供商与客户之间的正式合同，规定了预期的服务水平，包括正常运行时间、响应时间和支持质量等性能指标，以及未达到标准时的补救措施

"softmax attention"

一种算法，通过使用softmax函数分配权重，帮助AI模型聚焦于输入中最相关的部分

"text-to-speech" or "TTS"

文本转语音，一种将书面文字转换为语音音频的技术，使机器能够以自然、清晰的声音朗读文本

"test-time compute"

推理时计算，指AI模型在完成训练后生成回应或执行任务时所使用的计算能力

"token"

词元，作为模型输入处理和输出生成的基本元素的文本单位

– 56 –

GLOSSARY OF TECHNICAL TERMS

"Transformer"

一种复杂的神经网络架构，能够通过利用注意力机制对关键输入元素进行优先级排序，从而高效处理文本等序列数据

"URL"

统一资源定位符，用于在互联网上定位和访问网页等资源的地址

"Video Generation API"

视频生成应用程序接口，允许开发者以编程方式创建或控制AI模型生成视频内容的接口，通常以文本或音频等作为输入

"video generation model"

视频生成模型，一种旨在通过随时间推移生成图像序列来创建合成视频内容的AI模型，通常基于文字描述、图像或其他输入数据。该模型通过学习动态、视觉一致性和时间连贯性，生成逼真或风格化的视频输出

– 57 –

FORWARD-LOOKING STATEMENTS

前瞻性陈述

本招股章程载有前瞻性陈述。并非历史事实的陈述，包括但不限于有关我们未来意向、信念、预期或预测的陈述，均属前瞻性陈述。在本招股章程中使用"目标"、"预期"、"相信"、"可能"、"预计"、"展望"、"意图"、"应当"、"计划"、"寻求"、"应该"、"将会"、"希望"、"愿景"、"立志"、"目的"、"时间表"等词语及其否定形式和其他类似表述，以及与我们或我们的管理层相关的此类词语，旨在识别前瞻性陈述。该等陈述反映了我们管理层目前对未来事件、运营、流动性及资本资源的看法，其中部分内容可能无法实现或可能发生变化。该等陈述受若干风险、不确定性及假设的影响，包括本招股章程所述风险因素，其中部分因素超出我们的控制范围，可能导致我们的实际业绩、表现或成就，或行业业绩，与前瞻性陈述所明示或暗示的任何未来业绩、表现或成就存在重大差异。

敬请注意，依赖任何前瞻性陈述涉及已知及未知的风险与不确定性。我们面临的可能影响前瞻性陈述准确性的风险与不确定性包括但不限于以下各项：

• 我们的运营及业务前景；

• 我们与供应商及客户维持关系的能力，以及影响供应商及客户的行动和发展；

• 我们所运营或计划运营的行业及市场的未来发展、趋势及状况；

• 我们所运营市场的整体经济、政治及商业状况；

• 我们所运营行业及市场的监管环境变化；

• 我们维持市场领先地位的能力；

• 竞争对手的行动及发展；

• 第三方按合同条款及规范履行义务的能力；

• 我们留任高级管理人员及关键人员并招募合格员工的能力；

— 58 —

前瞻性陈述

• 我们的业务策略及实现该等策略的计划；

• 我们质量控制体系的有效性；

• 利率、外汇汇率、股票价格、交易量、商品价格及整体市场趋势的变化或波动，包括与中国及我们运营所在行业和市场相关的变化或波动；以及

• 资本市场发展。

就其性质而言，与上述及其他风险相关的若干披露仅为估计，若上述不确定性或风险中的一项或多项成为现实，实际结果可能与估计、预期或预测的结果以及历史业绩存在重大差异。具体而言（但不限于此），销售额可能下降、成本可能增加、资本成本可能上升、资本投资可能延迟，且预期的绩效改善可能无法完全实现。

在适用法律、规则及法规规定的范围内，我们无义务亦不承诺更新或修订本招股章程中的前瞻性陈述，无论是由于新信息、未来事件或其他原因。由于上述及其他风险、不确定性及假设，本招股章程所讨论的前瞻性事件及情况可能不会按我们预期的方式发生，或根本不会发生。因此，您不应过度依赖任何前瞻性信息。本招股章程中的所有前瞻性陈述均受本节警示性陈述以及"风险因素"一节所述风险与不确定性的约束。

在本招股章程中，有关我们或我们董事意向的陈述或提述均于本招股章程日期作出。此类信息可能随未来发展情况而发生变化。

— 59 —

RISK FACTORS

An investment in our Class A Ordinary Shares involves significant risks. You should carefully consider all of the information in this Prospectus, including the risks and uncertainties described below, before deciding to invest in our Class A Ordinary Shares. Particularly, we are a Pre-Commercial Company seeking to list on the Main Board of the Stock Exchange under Chapter 18C and Chapter 8A of the Listing Rules. Our operations and the global foundation model industry in which we operate involve certain risks and uncertainties, some of which are beyond our control and may cause you to lose all your investments in our Class A Ordinary Shares.

The following is a description of what we consider to be our material risks. Any of the following risks could have a material adverse effect on our business, financial condition, results of operations and growth prospects. In any such event, the trading price of our Class A Ordinary Shares could decline, and you may lose all or part of your investment. Additional risks and uncertainties not presently known to us, or not expressed or implied below, or that we deem immaterial, could also harm our business, financial condition and results of operations.

These factors are contingencies that may or may not occur, and we are not in a position to express a view on the likelihood of any such contingency occurring. The information given is as of the Latest Practicable Date unless otherwise stated, will not be updated after the date hereof, and is subject to the cautionary statements in the section headed "Forward-Looking Statements" in this Prospectus.

RISKS RELATED TO THE COMMERCIALIZATION OF OUR PRODUCTS

We have recorded net losses, net liabilities and operating cash outflow during the Track Record Period and recorded net current liabilities as of September 30, 2025, and we may not be able to achieve or subsequently maintain profitability.

We have recorded net losses and net liabilities during the Track Record Period, and have experienced, and expect to continue to experience, cash outflow from operating activities. Our operating expenses, comprising selling and distribution expenses, administrative expenses and research and development expenses, increased from US$14.4 million in 2022 to US$100.4 million in 2023 and further to US$290.4 million in 2024, primarily due to increasing investment in third-party cloud services related to training. We recorded net losses of US$73.7 million, US$269.2 million, US$465.2 million, US$304.3 million and US$512.0 million in 2022, 2023, 2024 and nine months ended September 30, 2024 and 2025, respectively. We may continue to record net losses in the short term as we are still in the expansion stage of our business and operations in the rapidly evolving global foundation model industry, and are continuously investing in research and development activities to support our long-term growth. We anticipate that our operating expenses, net losses, net liabilities and cash outflow from operations will continue to increase in the foreseeable future as we continue to expand our

– 60 –

RISK FACTORS

business and operations, invest in our foundation model research and development, and carry out sales and marketing activities. If we fail to manage such increases, our business operations, results of operations, financial position and profitability would be materially adversely affected.

In addition, we may not be able to achieve or subsequently maintain profitability in the near future. We believe that our future revenue growth will depend on, among other factors, our ability to develop new technologies, enhance user experience, establish effective commercialization strategies, compete effectively and successfully and develop new products. Accordingly, you should not rely on the revenues of any prior period as an indication of our future performance. Furthermore, we expect to incur substantial costs and expenses as a result of being a public company. If we are unable to generate adequate revenues and manage our expenses, we may continue to incur significant losses and may not be able to achieve or subsequently maintain profitability.

The content or data that we use to train our foundation models and the content generated by our foundation models could be subject to third-party intellectual property infringement claims which may materially and adversely affect our business, financial condition and results of operations.

The model training process begins with the curation and organisation of massive volumes of available data sourced from a wide range of domains. It is possible that our training data includes third-party text, video, image, audio or other content without all required third-party consent. The legal frameworks governing the use of such data for training foundation models are still developing and subject to change. If our use of training data is found by competent jurisdictional authorities to infringe on third-party IP rights, we could be subject to costly and time-consuming litigation, be forced to pay substantial damages or licensing fees, or be required to remove the data and retrain our models, which would be a difficult and expensive undertaking.

Additionally, there is a risk that our foundation models may be capable of generating outputs that are substantially similar to existing copyrighted works, trademarks, or other protected intellectual properties. The legal standards for determining infringement by AI-generated content are not yet settled. If the output of our foundation models is deemed by competent jurisdictional authorities to infringe on a third party's intellectual property rights, we, and the users of our AI-native products, could be held liable. Such claims could lead to legal action against us, demands for indemnification from our customers, and could damage our reputation and the perceived reliability of our products.

On September 16, 2025, a group of major U.S. movie studio companies, including Disney, Universal and Warner Bros. Discovery (the "Plaintiffs"), filed a civil complaint (the "Complaint") in the United States District Court for the Central District of California, against our Group in relation to Hailuo AI, our visual generation platform. The Plaintiffs allege (i) direct infringement, on the basis that the Company itself, through Hailuo AI, created and displayed videos and images depicting a number of well-known film and animation characters

– 61 –

RISK FACTORS

owned by the Plaintiffs, and (ii) secondary infringement, under contributory and vicarious infringement doctrines, on the basis that the Company knew or should have known that users could create content depicting the Plaintiffs' characters, and because the Company is allegedly benefiting from that use. In their prayer for relief, the Plaintiffs primarily seek, among other things, monetary relief in the form of actual or statutory damages, injunctive relief, attorneys' fees and other equitable remedies.

The Plaintiffs have alleged in their Complaint that they are entitled to statutory damages of up to US$150,000 per infringed work, which is the maximum amount awardable per work under U.S. copyright law and is only awarded if infringement is found to be willful. The attachments to the Complaint identify approximately 500 registrations for motion pictures and television programs that are at issue in this case. Therefore, assuming the plaintiffs prevail and fully succeed in their claims, the worst-case scenario, as alleged by the plaintiffs, would be a monetary claim of US$75 million and injunctive relief. In the alternative, the plaintiffs have also alleged they are entitled to actual damages and disgorgement of profits. For details, see "Business — Legal Proceedings and Compliance — Copyright Infringement Lawsuit". In the event that the Plaintiffs prevail in the Lawsuit, we could be subject to significant monetary damages, damage to our brand image, among others, which could negatively affect our business, financial condition and results of operations. Additionally, we cannot preclude the possibility that the Plaintiffs may raise additional claims against the Company, in which case the Company may be liable for further damages, legal costs, and reputational harm, which could further affect our business operations and financial positions. Furthermore, we cannot preclude the possibility that other enterprises may raise similar claims against the Company, exposing us to additional legal risks and potential liabilities. Regardless of the merits or outcome of the Lawsuit, other third-party intellectual property owners may seek to assert claims against us in connection with the same or different content. The existence of ongoing litigation, together with the evolving legal standards, judicial practices and licensing arrangements in our industry, may result in additional third-party intellectual property owners pursuing similar claims, which could expose us to further legal proceedings and other adverse consequences.

We are aware of recent licensing arrangements among certain industry participants, including licensing arrangements between IP owners and generative AI technology companies. These licensing arrangements may reflect a broader trend toward increased collaboration between traditional IP rights holders and AI companies. While we will closely monitor industry developments and potential commercial opportunities, we may not be able to identify suitable IP rights holders for cooperation in a timely manner, or to effectively capture the benefits of such industry trends, which could adversely affect our business, results of operations and growth prospects. In addition, such cooperation may lead to more widespread use of licensing and other commercial arrangements relating to copyrighted content and generative AI technologies, and regulatory frameworks and market practices may evolve toward greater clarity and standardization. While such developments may reduce certain areas of uncertainty, they may also result in a more active IP enforcement environment. For example, holders of

– 62 –

RISK FACTORS

exclusive licenses, or future exclusive licensees of copyrighted works, may seek to assert or enforce their licensed rights against third parties, including through infringement claims, which could expose us to additional legal risks.

We operate in a rapidly evolving and increasingly competitive global foundation model industry. Our business is subject to constant technological advancements and industry transformation. If we fail to continuously innovate and adapt to evolving customer needs, our competitive position would be impacted and our business, financial condition and results of operations may be materially and adversely affected.

We primarily compete in the global foundation model industry, which is characterized by intense competition and constant changes, including rapid technological evolution, frequent introductions of new products, continual shifts in customer demands and periodic emergence of new industry standards and practices. Our success depends, in part, on our ability to respond to such competition and changes in a cost-effective and timely manner. We need to develop expertise across different industry sectors, adapt our products for different industry verticals and constantly anticipate the emergence of new technologies and assess their market acceptance. We also need to invest significant resources, including financial resources, in research and development to lead technological advances in order to keep our products competitive in the market.

In particular, the foundation model industry is experiencing rapid innovation across multiple dimensions. From a technological perspective, advances include (i) new model architectures such as Mixture-of-Experts and other sparsely activated networks that improve the balance between reasoning performance and computational efficiency; (ii) extended context windows and multi-step reasoning mechanisms that enable models to understand longer documents, perform complex analytical tasks, and act as autonomous agents; (iii) multi-modal foundation models that can process and generate text, images, audio and video simultaneously, allowing for more natural and context-rich human-AI interaction; and (iv) improved pre-train and post-train techniques, which enhance accuracy and allow models to be tailored to specific business uses.

At the same time, customer needs are also evolving. Users and enterprise clients are increasingly seeking: (i) powerful foundation models that integrate knowledge across multiple domains and can be efficiently augmented with their proprietary context to address sector-specific use cases; (ii) lower latency and higher cost efficiency that enable large-scale, real-time use in both consumer and enterprise applications; (iii) stronger privacy, security and data-governance features to satisfy regulatory obligations and internal controls; and (iv) AI systems with persistent memory, tool integration and autonomous task capabilities, enabling proactive, agent-like performance instead of simple one-off responses. These trends are accelerating industry transformation and driving leading participants to invest continually in model algorithm, computing infrastructure, alignment and safety features built around general-use models.

– 63 –

风险因素

基础模型市场正以前所未有的速度扩张，迅速重塑人类社会。根据CIC的数据，全球基础模型市场预计到2030年将超过3,000亿美元。IDC估计，到2030年，人工智能将为全球经济累计贡献19.9万亿美元，并在2030年推动全球GDP增长3.5%。尽管这一增长带来了重大机遇，但也吸引了越来越多的竞争者，其中许多竞争者拥有比我们更雄厚的技术、资金和数据资源。这些竞争者可能能够在模型训练、基础设施、人才引进和市场推广方面投入更多资源，从而可能限制我们获取或维持市场份额的能力。随着人工智能成为数字基础设施的核心组成部分，竞争可能进一步加剧，使我们更难以使我们的产品脱颖而出、维持定价权并实现可持续盈利。

如果我们无法提供满足或超越用户不断演变的性能期望的产品，或未能跟上创新周期的步伐，我们的产品可能失去竞争力，导致定价权下降或错失商业机会。因此，我们的业务、财务状况及经营业绩可能受到重大不利影响。

我们或其他第三方有意或无意实施的基础模型技术的任何实际或被认为存在的缺陷或不当使用，可能对我们的声誉、业务、财务状况、经营业绩以及社会整体对基础模型产品的广泛接受程度产生重大不利影响。

基础模型技术尚处于发展早期阶段，仍在持续演进。与许多创新一样，基础模型技术存在风险和挑战，例如第三方可能将其用于不当目的或带有偏见的应用，从而损害公众信心，或引发某些个人以侵犯隐私权或人格权等合法权益为由提起的诉讼或其他法律程序。

我们的用户可能使用我们的技术生成内容，或在网上发布其使用我们产品的体验。部分用户生成内容或通讯可能包含违法、淫秽或煽动性内容，可能对其他用户产生负面影响。此类内容或通讯可能依据适用法律法规被认定为违法，政府主管部门可能要求我们停止或限制某些已经导致或可能导致上述情况的功能、特性或产品，或终止与相关平台服务提供商的合同。我们可能因用户生成内容或通讯中不当传播的信息所引发的索赔或处罚，在调查和抗辩中产生重大费用，我们的业务、财务状况、经营业绩及发展前景可能受到重大不利影响。即使我们与用户签订了惯常的用户协议，我们也无法控制用户对我们基础模型及人工智能原生产品的使用方式。此类滥用行为可能影响客户认知、公众舆论、政策制定者和监管机构的看法，并导致基础模型技术的采用率下降。

－64－

RISK FACTORS

In addition, flaws or deficiencies in foundation model technologies could compromise the accuracy, reliability, integrity and thoroughness of the recommendations, forecasts or analyses generated by our products. If the recommendations, forecasts or analyses that our foundation models or our AI-native products assist in producing are inaccurate, misleading or otherwise flawed, or if our users rely on our products in inappropriate or harmful ways, we could be subjected to competitive harm, potential legal liability, and ethical or reputational harm. By "competitive harm," we refer to negative impacts on our ability to compete effectively in the market, including but not limited to loss of users or customers, damage to our brand, erosion of trust, increased customer acquisition costs, and the strengthening of our competitors' market positions. There can be no assurance that we will be able to detect and rectify such issues in a timely manner, or at all. Any flaws or inappropriate usage of foundation model technologies and related products, whether actual or perceived, could materially and adversely affect our business, reputation, results of operations and prospects.

The competitiveness of our foundation models and offerings depends on our continuous and significant investment in research and development, and we intend to continue investing significantly in research and development. Such investment may negatively impact our profitability and operating cash flow in the short term and may not generate the results we expect to achieve.

Our technological capabilities and infrastructure are critical to our success. To maintain the competitiveness, we have made substantial investments in our research and development. Our research and development expenses increased from US$10.6 million in 2022 to US$70.0 million in 2023 and further to US$189.0 million in 2024, and subsequently increased from US$138.7 million in the nine months ended September 30, 2024 to US$180.3 million in the nine months ended September 30, 2025, representing nil, 2,023.2%, 619.1%, 712.9% and 337.4% of our total revenues for the corresponding period. The industries in which we operate are characterized by rapid technological changes and are constant technological innovation. We need to allocate significant financial and other resources to research and development in order to enhance the functionality, performance, and competitiveness of our products. As a result, we expect that our research and development expenses will remain relatively high for the foreseeable future.

However, our expenditures on research and development may not generate corresponding benefits. We have been constantly focusing on advancing the deep integration and performance optimization of algorithms, software, and hardware, with particular emphasis on usability, scalability, and alignment with evolving industry standards and user needs. However, there is no guarantee that all of our efforts on research and development can yield anticipated benefits. Research and development activities are inherently uncertain, and we may not be able to obtain and retain sufficient resources, including qualified research and development personnel. It is possible that our research and development efforts may ultimately prove unsuccessful. Even if we succeed in our research and development efforts and achieve the results we expect, such results may not materialize within the expected timeframe, and we may still encounter practical difficulties in commercializing our research and development results. In addition, even successfully developed products may not achieve market acceptance to the extent we

– 65 –

RISK FACTORS

anticipate. Moreover, given the rapid pace of innovation in the global foundation model industry, new technological developments could render our current or future technologies obsolete or less competitive, thereby limiting our ability to recover related development costs, which could result in a decline in our revenues, profitability and market share.

We have a limited operating history, which makes it difficult to forecast our future business prospects and results of operations.

We have a limited operating history, and we only commenced the development of models across multiple modalities in 2022. As a result, and particularly in light of the rapidly evolving nature of the global foundation model industry, it may make it difficult to evaluate our current business and reliably predict our future performance. Our historical results may not provide a meaningful basis for evaluating our business, results of operations, financial condition and prospects, and we may encounter unforeseen expenses, difficulties, complications, delays and other known and unknown factors, and may not be able to achieve promising results in future periods. If we cannot address these risks and overcome these difficulties successfully, our business and prospects will suffer.

In addition, our growth prospects should be considered in light of the risks and uncertainties that fast-growing companies with a limited operating history may encounter, including, among others, risks and uncertainties regarding our ability to:

• maintain and upgrade our AI architecture;

• upgrade our foundation model products and develop new technologies;

• further commercialize our foundation model products;

• retain existing users and attract new users to purchase our products;

• expand into new industry verticals and launch new products;

• further expand into international markets;

• increase brand awareness through marketing and promotional activities;

• successfully compete with other companies that are currently in, or may in the future enter, the industries and verticals we have entered;

• attract, retain and motivate talented employees, including research and development talents as well as staff with in-depth industry know-how;

• adapt to evolving regulatory environment; and

– 66 –

RISK FACTORS

•

defend ourselves against litigation, regulatory, intellectual property, privacy, data protection or other claims.

All of these initiatives involve inherent risks and will require significant research and development expenses, operating expenses and allocation of valuable management and employee resources. We cannot assure you that we will be able to effectively manage the expansion or growth of our operations and workforce or implement our business strategies effectively. If the markets for our products fail to develop as anticipated, or if we are unable to address the evolving needs of this dynamic market, our business, results of operations and financial condition may be materially and adversely affected.

As we continue to grow, we may not be able to effectively manage our growth and expand our operations, which could negatively impact our operation performance growth, financial condition, results of operations, and reputation.

We have experienced rapid growth in recent years. Driven by our monetization strategy, our revenues increased from nil in 2022 to US$3.5 million in 2023, and further to US$30.5 million in 2024, and subsequently increased from US$19.5 million in the nine months ended September 30, 2024 to US$53.4 million in the nine months ended September 30, 2025. We plan to further expand our business by, among other things, continuing to invest in technology, enhancing our brand recognition, and scaling our products globally. Our future operating results will depend to a large extent on our ability to manage our expansion and growth successfully.

Risks that we face in undertaking this expansion include, among others:

•

managing a larger organization with a greater number of employees across different divisions and geographic locations;

•

managing our supply chain to support rapid business growth;

•

controlling expenses and investments in anticipation of expanded operations;

•

establishing or expanding research and development, sales and service facilities;

•

implementing and enhancing administrative structure systems and processes;

•

executing our strategies and business initiatives successfully;

•

our ability to secure services from third parties who procure specialized hardware and software, that we rely on to establish the technology infrastructure supporting our foundation model products;

•

improving our operational, financial and management controls, compliance programs and reporting systems; and

– 67 –

RISK FACTORS •

addressing new markets and potentially unforeseen challenges as they arise.

If we fail to efficiently manage our business expansion, our costs and expenses may increase faster than anticipated, and we may not be able to respond promptly to competitive challenges or otherwise execute our business strategies successfully. Our growth requires significant financial resources and will place significant demands on our management. If we fail to effectively manage the growth of our business and operations, our reputation, overall prospects, and results of operations could be negatively impacted.

We may not be able to sustain our historical growth rates, and our historical growth may not be indicative of our future growth or financial results.

We have achieved rapid growth during the Track Record Period. Our total revenue increased from nil in 2022 to US$3.5 million in 2023, and further by 782.2% to US$30.5 million in 2024. We recorded total revenue of US$53.4 million in the nine months ended September 30, 2025, compared with US$19.5 million during the same period in 2024. However, there is no assurance that we will be able to sustain our historical growth rates in future periods. Our growth rates may decline for a number of reasons, including macroeconomic conditions, technology development in the global foundation model industry, availability of AI talents, increasing enterprise awareness to adopt foundation model products, our continued investment in technological innovation, and our ability to attract and retain our users. We cannot assure you that we will be able to effectively manage our growth or implement our business strategies. If the market for our products does not develop as we expect or if we fail to address the needs of this dynamic market, our business, results of operations and financial condition will be materially and adversely affected.

If our expansion or attempts to develop new products is not successful, our business, prospects and growth momentum may be materially and adversely affected.

Leveraging our foundation model technologies as well as software and hardware integration capabilities, we are able to provide foundation model-empowered products designed to address diversified needs of our users across different verticals. We have a track record of successfully developing new products. We cannot assure you, however, that we will be able to maintain this growth momentum in the future. Expanding offering categories into new business areas involves new risks and challenges. Our lack of familiarity with new verticals may make it more difficult for us to keep pace with evolving customer demands and preferences. In addition, there may be one or more existing market leaders in any vertical that we decide to expand into. Such companies may be able to compete more effectively than us by leveraging their experience in doing business in that market as well as their deeper industry insight and greater brand recognition among users. We will need to comply with new laws and regulations applicable to these businesses. Expansion into any new vertical and development of new products may place significant strain on our management and resources and incur substantial research and development and other costs and expenses before generating any revenues, and failure to expand successfully could have a material adverse effect on our business and prospects. We may also experience downward pressure on our operating margin

– 68 –

RISK FACTORS

as a result of such business expansion of our business into new areas, which may have margins much lower than that of our existing business lines. Our operating margin may also be negatively impacted from a greater proportion of revenue contributed by new business areas, which may grow faster than our existing business lines.

We have limited experience in the commercialization of our products.

We have relatively limited experience in launching, commercializing, as well as the sales and marketing of our products. For example, we have limited experience in building a commercial team, conducting comprehensive market analysis, or managing the sales force for our products and services. Therefore, our ability to successfully commercialize our products may involve more inherent risks, take longer, and cost more than it would if we were a company with more experience in sales and marketing. In particular, the commercialization of new products requires additional resources. The success of our sales and marketing efforts depends on our ability to attract, motivate and retain qualified and professional employees in our commercialization team who have, among other things, adequate industry knowledge to communicate effectively with industry professionals, sufficient experience in sales and marketing of our foundation model products, and strong connections within the industry as well as with academic and research institutions. Furthermore, along with our market expansion after the commercialization of our products and services, we expect to hire more employees with relevant industry experience and knowledge to strengthen our sales and marketing workforce. However, competition for experienced sales and marketing personnel is intense. In certain emerging industries that rely significantly on AI, many players with sufficient funds would heavily devote their resources to compete for talents with us. If we are unable to attract, motivate and retain a sufficient number of qualified sales and marketing personnel to support our business, the commercialization of our products may be adversely affected. Our business, results of operations, and prospects may also be adversely affected if our investment and efforts to expand our sales force do not generate a corresponding increase in revenue.

The size of our addressable markets and the demand for our products may not increase as rapidly as we anticipate due to a variety of factors. If the market for our products fails to grow as we expect, or if our users or potential users fail to adopt our products, our business, results of operations and financial condition could be adversely affected.

We are pursuing opportunities in markets that are undergoing rapid changes, including technological and regulatory changes, and it is difficult to predict the timing and size of the opportunities for our key products. See "— Risks Related to the Commercialization of Our Products — We operate in a rapidly evolving and increasingly competitive global foundation model industry. Our business is subject to constant technological advancements and industry transformation. If we fail to continuously innovate and adapt to evolving customer needs, our competitive position would be impacted and our business, financial condition and results of operations may be materially and adversely affected."

– 69 –

RISK FACTORS

In particular, our business growth relies on our ability to identify and adapt to the needs of our users and develop products that meet their demands. Our ability to retain existing users, attract new users, and increase revenue from both new and existing users will depend on a number of factors. Such growth also depends, to a large extent, on our ability to provide products that meet our users' requirements, including more advanced products that address the evolving needs of our users at competitive prices, the strength of our technology, and our ability to continue improving and enhancing the functionality, performance, reliability, design, security and adaptability of our products. To the extent we are not able to provide products that meet our users' requirement, or we are not able to improve and enhance the functionality, performance, reliability, design, security, adaptability and scalability of our products in a manner that responds to our users' evolving needs, our existing users may not spend more on our products, and we may not be able to attract new users, under which circumstances our business, financial condition, results of operations and prospects may be materially and adversely affected.

Our future financial performance will depend on our ability to make timely investments in the correct market opportunities. If one or more of these markets experience a shift in customer or prospective customer demand, then our products may not compete as effectively, if at all. Given the evolving nature of the markets in which we operate, it is difficult to predict customer demand or adoption rates for our products or the future growth of the markets in which we operate. Even if the global foundation model market grow substantially, there is no guarantee that demand for our products will correlate with that growth if we fail to effectively pursue such opportunities. There is also no guarantee that our business will be successful simply because of the future addressable markets of our products, or because of the trends of the addressable markets of our products. If demand does not develop or if we cannot accurately forecast customer demand, then the size of our markets, our future business, results of operations and financial condition would be materially and adversely affected.

In addition, evolving market conditions and changes in customer expectations may impact our pricing models and further affect demand for our products. As the market for our products grows, as our competitors introduce new products that compete with ours or reduce their prices, or as we enter into new verticals or international markets, we may be unable to attract new users or retain existing users based on our historical pricing models. Given our limited operating history and limited experience with our historical pricing models, we may not be able to accurately predict customer demands. In addition, regardless of the pricing model used, certain users may demand higher price discounts. As a result, we may be required to reduce our prices or offer alternative pricing models, which could adversely affect our revenue, gross margin, profitability, financial position and cash flow.

Furthermore, our prices vary across our products. Our products have different margin profiles, which vary between products depending on the amount, number and type of components that we deliver. If we adjust our business mix or fail to maintain our gross margin and operating margin for our products, our business, results of operations and financial condition would be adversely affected.

– 70 –

RISK FACTORS

The execution of our growth strategies will also require substantial capital investment and resource allocation. In particular, we may fund some of our expansion plans through our internal financial resources, and may also seek external equity or debt financings to implement them. If we seek debt financings for such plans, we may incur interest costs, which may affect our profit. In addition, we may not be able to manage our current or future operations effectively and efficiently to compete successfully in our existing markets or the new markets that we enter. We may also need to adjust our business plans and growth strategies from time to time, which could involve uncertainties. If our business plans and growth strategies fail to perform as expected, our business, financial condition and results of operations could be materially and adversely affected.

If we fail to retain existing users or attract new users, our business, financial condition and results of operations will suffer.

As of September 30, 2025, our AI-native products had collectively served over 200 million cumulative users and more than 100 thousand enterprises and developers across more than 200 countries and regions. We expect to continue to maintain business relationships with these existing users by not only providing existing products but also exploring their evolving needs to develop and promote new products. We also intend to further grow our business by attracting new users and expanding our global footprint. As a result, retaining our existing users and engaging new users are critical to our future operating results. Factors that may affect our ability to retain and promote additional products to our developers and enterprise customers include:

• the demand of our users for products;

• the price, performance, and functionality of our products;

• the availability, price, performance, and functionality of competing products;

• the stability, performance, and security of our technological infrastructure;

• our ability to develop complementary products, applications and platforms that are tailored to our users' needs;

• the effectiveness of our products;

• the success of our upgraded products or technologies;

• the financial performance, the budget of the research and development activities and the overall business environment of our company; and

• the overall business environment of the industry.

– 71 –

RISK FACTORS

In addition, we deliver certain products on a subscription basis. As a result, our future revenue depends in part on our ability to retain existing users and encourage renewals of subscriptions, which is in turn dependent on our ability to scale and adapt our products to meet our users' evolving needs. Factors that are not within our control may result in a reduction in our revenue or profitability. The cancellation of subscriptions to our products could materially and adversely affect our business, financial condition and results of operations.

RISKS RELATED TO OUR BUSINESS AND INDUSTRY

We may be unable to successfully expand our user base globally, and the expansion of our international operations may expose us to additional regulatory, economic and political risks, the failure to handle which may adversely affect our business, financial condition and results of operations.

We aim to expand our user base globally. However, we may not succeed in this endeavor and our success will depend on our ability to provide foundation model products that meet the diverse needs of users across different regions. For example, given the high regulatory and market access challenges in specific markets, we may not actively explore these markets in the short term, which could limit our ability to successfully achieve this objective. In addition, we operate in a highly competitive industry, and we cannot assure you that the pace of our growth will meet expectations. Our expansion strategy also requires significant cash investments and management resources and there is no guarantee that our business can generate additional sales of our products to support our expansion. As we expand, we will face risks in doing business internationally that could adversely affect our business, including:

• the difficulty of managing and staffing international operations and the increased operations, travel, and network costs associated with numerous international locations;

• challenges of gaining acceptance for our products by users in different markets;

• our ability to effectively price our products in competitive international markets;

• global or regional health crises;

• tariffs and other non-tariff trade barriers, such as quotas and local content rules;

• the complexities of complying with current and future export control and economic sanctions administered by the U.S. Department of Commerce's Bureau of Industry and Security, the U.S. Department of the Treasury's ("Treasury") Office of Foreign Assets Control and other relevant authorities;

• protectionist or national security policies that restrict our ability to develop, import or export certain technologies or limit our ability to raise capital; and

– 72 –

RISK FACTORS •

more limited protection for intellectual property rights in some countries.

Our failure to manage any of these risks successfully could harm our international operations, and adversely affect our business, operating results and financial condition.

AI technologies carry certain inherent safety risks, which may adversely affect our business and reputation.

AI technologies carry certain inherent risks and challenges that may adversely affect public perception of our business. Any inappropriate, abusive or premature usage of AI technologies, whether actual or perceived, whether intended or inadvertent, and whether by us or by third parties, may dissuade prospective users from adopting AI-native products, may impair the general acceptance of AI-native products by the society, attract negative publicity and adversely impact our reputation. Specific risks relating to AI technologies may include, among others: (i) fraudulent activity, such as the creation of convincing fake images, videos, and text that can be used to create deepfakes, impersonations, and forged documents for fraudulent purposes; (ii) misinformation and disinformation, such as the generation of realistic and convincing synthetic media that could be used to spread misinformation and disinformation; (iii) privacy concerns, such as the creation of synthetic identities or manipulation of personal data, which may raise privacy concerns; (iv) cybersecurity threats, such as the creation of sophisticated phishing attacks or bypass of security measures, which may increase the risk of cyberattacks and data breaches; and (v) safety and alignment challenges, particularly as AI systems become more advanced and capable.

Adverse user behaviours and misuse of AI-native products may also cause, or be alleged to cause, harm to users or third parties, including by promoting or facilitating self-harm, suicide or other violent conduct, or by creating undue emotional dependence, particularly among minors or other vulnerable persons. In this regard, certain AI companies have been sued in connection with allegations that their chatbot products contributed to users' self-harm or suicide.

In addition, adverse user behaviours may result in heightened scrutiny by platform operators and other regulators, and could lead to enforcement actions such as takedowns, suspensions, distribution restrictions, age-gating requirements or other remedial measures that may materially reduce user acquisition, engagement and monetisation. For the temporary removal of the Talkie app from Apple's App Store, please refer to "Risk Factors – Any restriction on access to major distribution channels, such as the iOS App Store, Google Play or the Internet, or any failure to maintain stable relationships with such channels, could materially and adversely affect our user growth and business performance".

If similar incidents were to occur in relation to our products or services, we could be subject to significant legal and regulatory exposure (including civil claims alleging negligence, wrongful death, product liability, failure-to-warn or consumer-protection violations), as well as other enforcement actions, product changes or usage restrictions. Even if such claims are ultimately unsuccessful, defending against them could be costly and time-consuming, could – 73 –

RISK FACTORS

divert management attention, could require increased spending on safety, compliance and customer support, and could cause reputational harm, any of which could materially and adversely affect our business, operations and financial performance.

We may face significant challenges in ensuring that our AI-native products behave in a manner that is safe, reliable, and aligned with human values. As AI models grow more complex, there is an inherent risk that they may exhibit unintended behaviors, pursue goals misaligned with user or societal interests, or fail to perform as expected in high-stakes or novel situations. For example, advanced AI systems could develop emergent capabilities, such as strategic planning or deception, that were not anticipated during their development. Additionally, the rapid pace of AI progress may exacerbate safety risks, as competitive pressures or the deployment of untrustworthy AI systems by third parties could lead to harmful outcomes. If we are unable to address these safety challenges effectively, any real or perceived failures in the safety or alignment of our AI systems could result in reputational damage, regulatory scrutiny and a loss of user trust, all of which could materially and adversely affect our business, operations, and financial performance.

In addition to these safety-related concerns, we are subject to a complex and evolving regulatory landscape governing internet content. Applicable government and regulatory authorities have adopted regulations governing content contained within videos, audios, images and other information over the internet. Under these regulations, internet content providers are prohibited from posting or displaying content that, among other things, violates applicable laws and regulations, impairs the national dignity of certain countries or the public interest, or is obscene, superstitious, fraudulent, violent or defamatory on the internet. Internet content providers are also prohibited from displaying content that may be deemed by relevant government authorities as illegal or inappropriate.

We allow our users to produce content using our AI-native products and generate outputs via our foundation models. These outputs may include various forms of media such as text, video, audio, and image, some of which may feature our proprietary watermarks. We implement measures to identify and restrict content that may be prohibited under applicable government regulations. However, we cannot guarantee that all generated content will be identified or restricted in a timely manner, or that it will fully comply with all relevant laws and regulations.

If we are unable to protect or promote our brand and reputation, our business may be materially adversely affected. Negative publicity or rumors about us, our products, our management, directors, employees, shareholders, users, business partners or their affiliates or our industry in general may adversely affect our reputation and business.

We must maintain and enhance our brand identity while increasing market awareness of the reputation of our business and products. The successful promotion of our brand depends on our ability to achieve widespread acceptance of our products, attract and retain users, maintain our current market share, and successfully differentiate our products from those of our competitors.

– 74 –

RISK FACTORS

Achieving these goals require substantial expenditures, and we anticipate expenses to increase as we expand into new markets. In addition to marketing and advertising costs, we may need to invest in customer support, public relations, community engagement, and compliance mechanisms to reinforce positive brand perception. However, there is no assurance that these investments will result in increased revenue, or that any revenue growth will be sufficient to offset the associated expenses.

Moreover, even isolated incidents, such as unintended outputs generated by our foundation models, miscommunication by company representatives, or negative feedback from influential users or media, can quickly escalate online and undermine years of brand-building efforts. Damage to our brand or reputation could lead to user attrition, reduced pricing power, increased customer acquisition costs, or reluctance from potential partners and investors to engage with us. Any of these factors could materially and adversely affect our business, financial condition, results of operations and growth prospects.

The technology infrastructure we relied on and our products may experience system failures, interruptions, security breaches, cyberattacks, or other technical inadequacies. Our brand, reputation, business, financial conditions or results of operations may be materially and adversely affected if we fail to effectively identify and rectify these problems in a timely manner.

The technology infrastructure we relied on and products may encounter disruptions or other outages caused by problems or defects in our own technologies and systems, such as malfunctions in software or network overload. The technology infrastructure we relied on may also be vulnerable to damage or interruption caused by security breaches, cyberattacks, human error or other technical inadequacies. The occurrence of unanticipated problems that affect the technology infrastructure we relied on could result in interruptions in the availability of our products. It may be difficult for us to respond to such interruptions in a timely manner, or at all. Such interruptions may affect the ability of users to use our products, which would damage our reputation, reduce our future revenues, harm our future profits, subject us to regulatory scrutiny and lead our users to seek alternative products.

It is possible that our security controls and other security practices we follow may not prevent the improper access to or disclosure of personal data or proprietary information. We also rely on systems provided by third parties, which may also suffer security breaches or unauthorized access to or disclosure of personal data or proprietary information. Additionally, our business involves the processing, storage, transmission and processing of confidential and user data, including user data, and the deployment of our IT resources in a safe and secure manner that does not expose our network systems to security breaches or the loss of data. Any data security incidents, including internal malfeasance by our employees, unauthorized access or usage, virus or similar breach or disruption of us or our service providers could result in loss of confidential or proprietary information or personal data, damage to our reputation, loss of users, litigation, regulatory investigations, fines, penalties and other liabilities. Accordingly, if our cybersecurity measures or those of our users fail to protect against unauthorized access, attacks (which may include sophisticated cyber-attacks), the compromise or mishandling of

– 75 –

RISK FACTORS

data, or other misconduct or malfeasance, including by computer hackers, employees, contractors, vendors, users and business partners, as well as software bugs, human error or technical malfunctions, then our reputation, business, operating results and financial condition could be adversely affected.

Furthermore, the technology infrastructure we relied on is also vulnerable to damages from fires, floods, earthquakes and other natural disasters, power loss and telecommunications failures. Any network interruption or inadequacy that causes interruptions to our operations, or failure to maintain the network and server or solve such problems in a timely manner, could reduce our user satisfaction, which in turn could adversely affect our reputation, business and financial condition.

Flaws in our foundation model products, including programming errors or defects in our models, whether real or perceived, could adversely affect our user experience and market acceptance of our products, which may materially and adversely affect our reputation, business and results of operations.

The technology underlying our AI-native products is inherently complex and may contain material defects or errors, particularly when new products are first introduced, when new features or capabilities are released or when integrated with new or updated third-party hardware or software. Our foundation model products are subject to frequent updates, and may contain bugs or flaws that can only become apparent when the updates are accessed by a number of users, especially when we launch updates under a tight schedule. We have from time to time received user feedback pertaining to programming errors. We cannot assure you that we will be able to detect and resolve all these programming errors effectively and in a timely manner. Any real or perceived programming errors or defects may adversely affect user experience, cause users to refrain from subscribing for our products, or cause our enterprise customers to reduce their use of our products, result in negative publicity and performance issues, any of which could materially and adversely affect our business and results of operations. Correcting such defects or errors may be costly and time consuming. Moreover, the harm to our reputation and legal liability related to such real or perceived defects or errors may be substantial and would harm our business.

We make certain of our models and products available on an open-source basis and may use open-source technology, which may pose particular risks to our business.

We strongly believe in open-source collaboration and we make certain of our models and products available on an open-source basis. By opening our technologies, we allow third parties, including competitors, to access, use, modify, or redistribute them, which could limit our ability to commercialize those technologies or differentiate ourselves in the marketplace. Specifically, our competitors may develop their own products using our open-source technology to compete with us, potentially reducing the demand for our products.

– 76 –

RISK FACTORS

此外，我们可能不时在某些业务中使用开源技术，并预计未来将继续使用某些开源技术。第三方可能就所有权提出主张或寻求执行开源许可证条款，此风险依然存在。此类主张可能包括要求披露开源组件、衍生作品，甚至是我们使用开源技术开发的专有源代码。这些主张可能导致诉讼，并分散管理层的注意力和资源。此外，许多开源许可证的条款尚未经法院解释，存在这些许可证可能以某种方式被解读、从而施加意料之外的条件或限制我们将产品商业化能力的风险。在此情况下，我们可能须向第三方寻求许可证以继续商业提供我们的产品，将我们的专有代码以源代码形式公开，对我们的产品进行重新设计，或在无法及时完成重新设计的情况下停止销售我们的产品，上述任何情况均可能对我们的业务、财务状况及经营业绩产生不利影响。

任何对主要分发渠道（如iOS App Store、Google Play或互联网）的访问限制，或与此类渠道维持稳定关系的任何失败，可能对我们的用户增长和业务表现产生重大不利影响。

我们某些产品的用户需要访问互联网及主要应用分发渠道（如Apple的App Store、Google Play及其他知名应用商店），以下载我们某些AI原生产品。我们的移动应用程序在上述第三方分发渠道上遭受任何中断、限制、暂停或下架，或上述渠道的条款、审核政策或技术要求发生任何变化，均可能对我们获取用户及交付产品的能力产生重大影响。例如，2024年12月，我们的Talkie应用程序的一个早期版本在某些司法管辖区的Apple App Store上被临时下架，下架期约为两个月。据我们所知，Talkie应用程序被临时从Apple App Store下架并非由于任何产品缺陷或违法行为，且Apple未说明下架原因。因此，自2024年12月中旬至2025年2月中旬，该早期版本的Talkie应用程序在某些司法管辖区无法从Apple App Store下载，Talkie应用程序的平均日下载量与下架前的平均水平相比减少了约1.68万次。在此期间，我们对Talkie应用程序的产品功能进行了若干调整，以增强其风险管理能力和用户体验。自2025年2月中旬起，更新后的Talkie应用程序已在Apple App Store上重新开放下载。如果我们当前或未来的任何移动应用程序在访问上述分发渠道方面受到限制、被下架或遭受其他中断——无论是由于技术问题、平台顾虑、内容标准演变、地缘政治敏感性或其他因素——我们可能面临声誉损害、用户增长减缓，且我们的财务状况及经营业绩可能受到重大不利影响。

此外，法律法规或政府当局可能出于安全、保密、数据隐私或其他方面的考虑，封锁或限制对互联网或上述分发渠道的访问，且无法保证我们能够维持稳定

– 77 –

RISK FACTORS

relationships with these distribution channels. Any restriction on access to the Internet in general or these distribution channels or the failure to maintain relationship with these distribution channels could result in the loss of existing users, slower user growth, or increased distribution and customer acquisition costs. In case an important distribution channel is inaccessible, we will resort to other distribution channels available to us. That said, our business, results of operations and prospects may be materially and adversely affected by limited access to distribution channels.

Our success depends on the continued contributions of our senior management and key employees. Failure to attract, recruit, retain, and motivate such qualified personnel could materially and adversely affect our business and growth prospects.

The business of the Company is highly dependent on certain individuals whose leadership, vision, and technical expertise are not readily replaceable. The market for high-caliber workers and leaders in our industry is extremely competitive. To execute our business strategies successfully, we must attract, retain and motivate our senior management and key employees. In particular, hiring qualified executives, scientists, engineers, technical staff and research and development personnel is costly and critical to our business.

Competition for personnel results in increased costs in the form of cash and stock-based compensation. Nonetheless, we must recruit and develop diverse qualified personnel to remain competitive in our industry. If one or more of our key employees, including senior executives or core technical personnel, were to depart unexpectedly, such departures could disrupt our operations, delay product development or strategic initiatives, and result in the loss of valuable institutional knowledge. Effective succession planning is also important to our long-term success. Failure to ensure effective transfer of knowledge and smooth transitions involving key employees could hinder our strategic planning and execution. If we are less successful in our recruiting efforts, or if we cannot retain key employees or their knowledge, our ability to develop and deliver successful products may be adversely affected. Such events may also lead to reputational harm, decreased employee morale, and potential reluctance from customers or partners to continue existing or prospective engagements.

The interpretation and application of employment-related laws to our workforce practices may result in increased operating costs and less flexibility in how we meet our workforce needs. Changes in immigration and work permit laws and regulations or the administration or interpretation of such laws or regulations could impair our ability to attract and retain highly qualified employees. If we do not continue to anticipate and address the needs of our employees sufficiently and/or in a timely manner, their productivity could be impacted, or we could fail to retain them, which could have a material adverse impact on our future business operations, results of operations and financial condition.

– 78 –

RISK FACTORS

We face risks related to changes in global and regional macroeconomic conditions, geopolitical tensions, regional conflicts, terrorist activities, natural disasters, health epidemics and other outbreaks of contagious diseases, and other force majeure events, any of which could materially and adversely affect our business operations, financial condition, results of operations and prospects.

Uncertainties about global economic conditions, regulatory changes, geopolitical tensions and other factors, including fluctuation of interest rates, inflation level, unemployment, labor and healthcare costs, access to credit, consumer confidence and other macroeconomic factors may pose risks and materially and adversely affect demand for our products. A deteriorating global economic outlook could result in a slowdown in business investment, tighter budget allocations for technology spending, and greater pricing sensitivity among customers, which may in turn reduce the market adoption of our products and services.

The escalated Palestinian-Israeli conflict, the conflict in Ukraine and the imposition of broad economic sanctions on Russia have disrupted global supply chains and triggered uncertainty across financial markets. Unrest, terrorist threats and the potential for war in the Middle East and elsewhere may increase volatility and risk aversion in global capital markets. In addition, the evolving trade relationships between China and other countries — particularly regarding tariffs, treaties, and government regulations — may have profound implications on cross-border business activities, cost structures, and our ability to access certain markets. Such developments may affect the macroeconomic environment, both domestically and internationally, and could have a direct or indirect impact on the markets in which we operate.

Geopolitical, economic and market conditions, including factors such as the liquidity of the global financial markets, the level and volatility of debt and equity prices, interest rates, currency and commodities prices, investor sentiment, inflation and the availability and cost of capital and credit have been affecting, and will continue to affect the countries where we operate. There is considerable uncertainty over the long-term effects of the expansionary monetary and fiscal policies adopted by the central banks and financial authorities of some of the world's leading economies.

There have been concerns over unrest and terrorist threats in the Middle East, Europe and Africa and over the conflicts involving Ukraine and Syria. The slow economic recoveries around the world and the high inflation, high interest environment have contributed to higher global volatility. These developments may adversely impact global liquidity, heighten market volatility and increase U.S. dollar funding costs resulting in tightened global financial conditions and fears of a recession. It is unclear whether these challenges and uncertainties will be contained or resolved, and what effects they may have on the global political and economic conditions in the long term. It remains unclear whether these challenges and uncertainties will be resolved in the near future, and their potential long-term effects on the global political and economic landscape remain difficult to predict. Any severe or prolonged slowdown in the global or PRC economy may materially and adversely affect our business, results of operations and financial condition.

– 79 –

RISK FACTORS

In addition, natural disasters such as floods, earthquakes, sandstorms, snowstorms, fire or drought, the outbreak of a widespread health epidemic or any severe epidemic disease such as SARS, Ebola, Zika or the COVID-19, acts of war, terrorism or other force majeure events beyond our control may disrupt our research and development, manufacturing and commercialization activities and business operations, all of which could adversely affect our business, results of operations, financial condition and prospects.

We are subject to the risks associated with international trade policies, geopolitics and trade protection measures. Changes in international relationships, trade and investment policies, trade protection and investment restriction measures may adversely impact our business, financial condition and results of operations.

Our operations may be negatively affected by trade policies, geopolitics and other trade protection measures administered by the government authorities in the countries in and with which we operate, including, but not limited to, regulation of economic and labor conditions, increased duties, taxes and other costs. Margins on sales of our products in certain countries could be materially and adversely affected by international trade regulations, including duties, tariffs and antidumping penalties. In addition to trade policy measures, the United States and certain other governments have imposed and may adopt additional sanctions, export controls and other regulatory measures that directly or indirectly affect technology companies based in certain jurisdictions. Due to the global presence of our business, we are potentially impacted by changes in international trade and investment policies, escalations of tensions in international relations, and increased scrutiny from regulatory authorities, particularly given recent trade negotiations between the United States and China, which has resulted in and may continue to cause changes in international trade policies and additional barriers to trade. Such regulatory measures are complex and subject to frequent changes, and the interpretation and enforcement of the relevant regulations may change from time to time, which may be driven by political and/or other factors that are not within our control or that are heightened by national security and foreign policy concerns.

For instance, in recent years, the United States has expanded sanctions and export control restrictions on China through the EAR, administered by BIS. These regulations are designed, in part, to restrict the access by Chinese companies to sensitive U.S. technologies, particularly in industries like telecommunications, artificial intelligence, and semiconductors. In recent years, the United States has expanded sanctions and export controls restrictions on China through the Export Administration Regulations (the "EAR"), administered by the Bureau of Industry and Security of the United States Department of Commerce (the "BIS"). For instance, in October 2022, BIS issued an interim final rule (the "BIS October 2022 IFR") requiring license for exports, re-exports, or transfers of any item subject to the EAR when there is "knowledge" that the item is destined for end use in the development or production of ICs at a fab in China that fabricates ICs meeting certain criteria. On December 2, 2024, BIS issued an interim final rule (the "BIS December 2024 IFR") and a final rule (the "BIS December 2024 FR"), which expanded controls in the EAR on advanced computing and semiconductor manufacturing items. Separately, BIS issued the so-called "Affiliate Rule" that expanded the scope of the Entity List and Military End-User List to include entities owned 50 percent or

– 80 –

RISK FACTORS

more directly or indirectly, individually or in the aggregate, by one or more listed parties. The U.S. Government has indicated that implementation of the Affiliate Rule will be delayed for at least one year (i.e., until November 2026). These recent measures together with the U.S. export control regime regulate the export, reexport and transfer of U.S. products, software, and technology, including certain items manufactured outside the United States that contain greater than de minimis controlled U.S. content or are the foreign direct product of certain U.S. software or technology. Export licenses may be required depending on the nature of the items, destination, end-use, end-user and other parties to the relevant transactions.

Our AI products were developed by us without direct using material U.S. software or technology, or incorporating material components procured from U.S. suppliers. We engage certain U.S. service providers in our ordinary course of business to support the operations of our international business. The services provided by such U.S. service providers could generally be replaced by suppliers in other jurisdictions around the world, at comparable quality and price, and we therefore believe that our R&D activities and operations are not reliant on technology or raw materials of U.S.-origin to any material extent. However, if any uncertainties in U.S.-China relationship or any resulted disruption to our supply chains will make it necessary for us to make such transition, such transition may take time to complete, and cause certain delays or disruptions to our ordinary course of business, and may therefore adversely affect our business, results of operations and financial conditions.

In addition to disrupting our supply chain in the U.S., export controls could also adversely impact the ability of our technology vendors in China to procure certain hardware or related services for their provision of services to us, which may have a material adverse impact on our operations. In the future, as similar or more expansive restrictions may be imposed by different jurisdictions, we will need to maintain heightened internal control and risk management policies to ensure sound compliance with such restrictions, which requires significant resources and efforts. Furthermore, such potential restrictions may materially and adversely affect our and our technology partners' abilities to acquire technologies, systems, devices or components that may be critical to business operations. Any of these developments could affect us, our users and/or suppliers or economic conditions generally, any of which could adversely affect our business and financial condition.

As advised by our international sanctions legal advisor, during the Track Record Period and up to the Latest Practicable Date, our Group has not been subject to sanctions, and we have not engaged in any material activities in comprehensively sanction countries, or entered into material service contract with any customers that are targets of U.S. sanctions. Therefore, as advised by our international sanctions legal advisor, we have been in compliance with rule and laws in US export control and sanctions in all material aspects, and U.S. sanctions are not likely to have any material adverse impact on us.

Our business operations is impacted not only by rules and laws related to export control, but also by changes in regulations governing cross-border investment policies. Changes in international investment policies, particularly with regard to China, could materially and adversely impact our business and operating results. In particular, in January 2025, a U.S. rule

– 81 –

RISK FACTORS

went into effect that prohibits or requires the submission of notifications in connection with U.S. outbound investment in Chinese-affiliated companies engaged in certain activities involving specified sensitive technologies sectors (artificial intelligence ("AI"), semiconductors and microelectronics, and quantum information technologies) and issued a broadly worded "America First Trade Policy" and an "America First Investment Policy" that seek to further restrict U.S. investments involving China (including possibly expanding technologies subject to the U.S. outbound investment regime and narrowing related exceptions (including those related to publicly traded securities)). In addition, effective on January 2, 2025, the final rule issued by the U.S. Department of the Treasury to implement the executive order of August 9, 2023 (the "Final Rule") imposes investment prohibition and notification requirements on U.S. Persons for a wide range of investments in entities associated with China (including Hong Kong and Macau) that are engaged in activities relating to three sectors: (i) semiconductors and microelectronics, (ii) quantum information technologies, and (iii) AI systems. The Final Rule could limit our ability to raise capital or contingent equity capital from U.S. investors after this Global Offering given that relevant laws, regulations, and policies continue to evolve. See "— We are subject to the risks associated with sanctions and export controls laws and regulations, and developing domestic and foreign laws and regulations on AI and related technologies, and our business, financial condition and results of operations could be materially and adversely affected."

We are closely monitoring potential changes in tariff policy and assessing the potential impact of such policy changes on our business operations and financial performance. For example, recently, the United States proposed to impose multiple rounds of tariffs on a wide range of goods imported from multiple countries, including China, and China responded with retaliatory tariffs. Since February 2025, both countries raised reciprocal tariffs on each other's imported goods to 125%. However, on May 12, 2025, both the U.S. and China modified these tariff measures: the U.S. removed the 125% tariff and temporarily reduced tariffs on Chinese goods to 10% by suspending a 24% duty for 90 days. The PRC government announced the same tariff adjustments, removing the 125% retaliatory tariff and cutting tariffs on U.S. goods from 34% to 10% for the same period. These policies have adversely affected the global economy and financial markets. On August 12, 2025, both the U.S. and China announced the extension of these tariff measures for another 90 days. On October 30, 2025, the United States announced it would further reduce tariffs by 10% but otherwise the tariffs by China and the United States remains in place. As advised by our international sanctions legal advisor, U.S. import tariffs only apply to the export of physical goods to the United States. On such basis, it is the view of our Directors that, given that we do not export physical goods to the United States, U.S. tariffs are unlikely to have a material adverse impact on our business operations and financial performance. As relevant policies are rapidly evolving, it may be difficult to evaluate these tariff measures' potential future impacts.

Geopolitical conflicts may also lead to volatility in financial markets, fluctuations in currency exchange rates, increased procurement costs and declines in trading prices of our Class A Ordinary Shares. In extreme cases, such conflicts could result in economic downturns that materially and adversely impact our operations. It is unclear whether these challenges and

– 82 –

RISK FACTORS

uncertainties will be contained or resolved, and what effects they may have on the global political and economic conditions in the long term and the ability of Chinese companies to raise capital from U.S. investors.

We collaborate with third-party online payment channels for payment collection. Any interruption of their services or unintended leakage of confidential information may materially and adversely affect our reputation and business.

We collaborate with major third-party payment channels to facilitate and collect users' payment for subscriptions and in-app purchases. We are subject to various risks and uncertainties associated with these third-party online payment channels. Any interruption in their payment services could adversely affect our payment collection, and in turn, our revenue.

In all online payment transactions through third-party payment channels, secured transmission of users' confidential information over public networks is essential for maintaining user confidence. We do not have control over the security measures of the third-party payment channels, and their security measures may not be adequate at present or may not be adequate with the expected increased usage of online payment systems. We could be exposed to litigation and potential liabilities if we fail to safeguard users' confidential information, which could harm our reputation and our ability to retain or attract users and may have a material and adverse effect on our business.

Furthermore, our payment channels are subject to various laws and regulations regulating electronic funds transfers and virtual currencies, which could change or be reinterpreted in a way that will adversely affect their compliance. If our payment channels experience any non-compliance incidents, they may be subject to fines and even lose their ability to accept online payments from our users, which in turn would materially and adversely affect our ability to monetize our user base. In addition, if we or the third-party payment channels experienced any investigations or litigations arising from any non-compliance incidents, the assets maintained at these third-party payment channels could be frozen, seized or even forfeited, which would materially and adversely affect our business and financial conditions.

We allow our users to supply content through certain of our products to receive AI-generated outputs. If users have not obtained all necessary copyright licenses in connection with such inputted content, we may be subject to potential disputes and liabilities.

We allow users to input content for the purpose of obtaining AI-generated outputs on certain of our products, which may expose us to potential disputes and liabilities in connection with third-party copyright. When users register on our platform, they agree to our standard agreement, under which they agree not to disseminate any content infringing on third-party copyright. Given the amount of user-supplied contents, the way they were inputted by users and the passage of time since they were initially inputted, it is generally impracticable for us to

– 83 –

RISK FACTORS

accurately identify and verify the individual users that inputted or provided such content, the copyright status of such content, and the appropriate copyright owners from whom copyright licenses should have been obtained.

Applicable laws and regulations and certain other jurisdiction, online service providers, which provide storage space for users to supply works or links to other services or content, may be held liable for copyright infringement under various circumstances. For example, according to our PRC Legal Advisor, we and our PRC operations are subject to the following PRC laws and regulations:

• According to the Civil Code of the People's Republic of China (中華人民共和國民法典), the network service provider shall bear joint and several liability with the network user for the expanded part of the damage if it fails to take timely necessary measures after receiving the notice from the right holder. Where a network service provider knows or should know that a network user is infringing upon the civil rights and interests of others by using its network services but fails to take necessary measures, it shall bear joint and several liability with such network user.

• According to the Provisions by the Supreme People's Court on Several Issues Concerning the Application of Law in Hearing Civil Dispute Cases Involving Infringement of the Right of Communication to the Public on Information Networks (最高人民法院關於審理侵害信息網絡傳播權民事糾紛案件適用法律若干問題的規定), if a network service provider knows or should know that a network user is infringing upon the right to disseminate information through information networks by using its network services, but fails to take necessary measures such as deletion, blocking, or disconnection of links, or provides technical support or other assistance, the people's court shall determine that it constitutes an act of contributory infringement.

• Pursuant to the Interim Measures for the Administration of Generative Artificial Intelligence Services (生成式人工智能服務管理暫行辦法), the provision and use of generative artificial intelligence services shall abide by the following provisions: (i) Respect intellectual property rights and business ethics, keep trade secrets; (ii) Respect the legitimate rights and interests of others, shall not endanger others' physical and mental health, or infringe upon others' rights to portrait, reputation, honor, privacy, or personal information.

Providers of generative artificial intelligence services (the "Providers") shall carry out training data processing activities such as pre-training and optimization training in accordance with the law, and comply with the following provisions: (i) Use data and basic models with legal sources; (ii) Where intellectual property rights are involved, shall not infringe upon the intellectual property rights legally enjoyed by others; (iii) Where personal information is involved, shall obtain the individual's consent or meet other circumstances prescribed by laws and administrative regulations; (iv) Take

– 84 –

RISK FACTORS

effective measures to improve the quality of training data and enhance the authenticity, accuracy, objectivity, and diversity of training data; (v) Other relevant provisions of laws and administrative regulations such as the Cybersecurity Law of the People's Republic of China, the Data Security Law of the People's Republic of China, and the Personal Information Protection Law of the People's Republic of China, as well as relevant regulatory requirements of competent authorities.

•

Providers shall assume the responsibilities of network information content producers in accordance with the law and fulfill obligations related to network information security. Where personal information is involved, they shall assume the responsibilities of personal information processors in accordance with the law and fulfill obligations for personal information protection.

•

Furthermore, providers shall establish and improve complaint and report mechanisms, set up convenient complaint and report portals, publicize processing procedures and feedback time limits, promptly accept and handle public complaints and reports, and feedback the handling results.

•

According to the Copyright Law of the PRC (中華人民共和國著作權法), anyone who commits any of the infringing acts such as using another person's work without paying remuneration as required shall, or other acts of infringing copyright depending on the circumstances, bear civil liabilities such as ceasing the infringement, eliminating the effects, making an apology, or compensating for losses.

•

According to the Provisions by the Supreme People's Court on Several Issues Concerning the Application of Law in Hearing Civil Dispute Cases Involving Infringement of the Right of Communication to the Public on Information Networks (《最高人民法院關於審理侵害信息網絡傳播權民事糾紛案件適用法律若干問題的規定》) issued by the Supreme People's Court and which became effective on January 1, 2021, if a network service provider knows or should know that a network user is infringing upon the right to disseminate information through information networks by using its network services, but fails to take necessary measures such as deletion, blocking, or disconnection of links, or provides technical support or other assistance, the people's court shall determine that it constitutes an act of contributory infringement.

See "Regulatory Overview — Laws and Regulations in the PRC — Government Policies on Artificial Intelligence" and "Regulatory Overview — Laws and Regulations in the PRC — Regulations Relating to Intellectual Property" for details.

We may not be effective in preventing the unauthorized posting and use of third parties' copyrighted content or the infringement of other third-party intellectual property rights. In addition, individual users who supply infringing content on certain of our products may not have sufficient resources to fully indemnify us, if at all, for any such claims. Also, such measures

– 85 –

RISK FACTORS

may fail or be considered insufficient by courts or other relevant regulatory authorities. We may be subject to joint infringement liability with the users, which may materially and adversely affect our business, financial position, and reputation.

As an online service provider, we have adopted a comprehensive set of measures to reduce the likelihood of using, developing or making available any content without the proper licenses or necessary consents. Such measures include (i) requiring users to acknowledge and agree that they will not supply or perform content which may infringe upon others' copyright; (ii) putting in place procedures to block users on our blacklists from supplying or distributing content; (iii) implementing "notice and take-down" policies to be eligible for the safe harbor exemption for user-generated content. However, these measures may not be effective in preventing the unauthorized posting and use of third parties' copyrighted content or the infringement of other third-party intellectual property rights. Specifically, it is possible that such acknowledgments and agreements by users may not be enforceable against third parties who file claims against us. Furthermore, a plaintiff may not be able to locate users who generate content that infringes on the plaintiff's copyright and may choose to sue us instead. In addition, individual users who supply infringing content on our platforms may not have sufficient resources to fully indemnify us, if at all, for any such claims. Also, such measures may fail or be considered insufficient by courts or other relevant regulatory authorities. If we are not eligible for the safe harbor exemption, we may be subject to joint infringement liability with the users, and we may have to change our policies or adopt new measures to become eligible and retain eligibility for the safe harbor exemption, which could be expensive and reduce the attractiveness of our platform to users.

We may not be able to adequately protect or enforce our intellectual property rights throughout the world, and our efforts to do so may be costly.

We rely on proprietary technology, and we are dependent on our ability to protect such technology. If we are not able to adequately protect or enforce the intellectual property rights relating to our foundation model products and other technologies, competitors could be able to access and use them, and our operations and financial condition could be adversely affected. Other parties may unintentionally or willfully disclose, obtain or use our technologies. Unauthorized third parties, including our competitors, may be able to copy certain portions of our products or reverse engineer or obtain and use information that we regard as proprietary. Our competitors may also be able to independently develop similar or superior products without copying our proprietary technology or design around our patents. Further, we may not have adequate intellectual property rights in certain proprietary technology in jurisdictions that are important to the business or that one day may become important to the business where we do not currently own any issued or applied-for patents. In addition, the laws of some foreign countries do not protect our intellectual property rights as fully as do the laws of other countries, and our ability to protect our intellectual property rights will differ per jurisdiction. We did not adopt an aggressive or offensive global intellectual property strategy to enforce our intellectual property rights, which may expose us to greater risk of infringement by third parties.

– 86 –

RISK FACTORS

In addition, any litigation initiated by us concerning the infringement by third parties of our intellectual property rights is likely to be expensive and time consuming and could lead to the invalidation of, or render unenforceable, our intellectual property rights, or could otherwise have negative consequences for us. We may be a party to claims and litigation as a result of alleged infringement by third parties of our intellectual property rights. Even when we sue other parties for such infringement, that suit may have adverse consequences for our business. Any such suit may be time consuming and expensive to resolve and may divert our management's time and attention from our business. Furthermore, it could result in a court or governmental agency invalidating, narrowing the scope of, or rendering unenforceable our patents or other intellectual property rights upon which the suit is based, which may seriously harm our business. Additionally, monitoring unauthorized use and disclosures of our proprietary technology, intellectual property and confidential information can be difficult and expensive. We cannot be sure that the steps we have taken will prevent misappropriation, infringement and violation of our intellectual property or proprietary rights. If we are unable to adequately protect, establish, maintain or enforce our intellectual property or other proprietary rights, our business, financial condition and results of operations may be adversely affected.

In particular, we must actively protect and maintain the legal ownership of our trademarks under which we market our brand and offer various foundation model products. Any failure to register or maintain the registration of our trademarks in any jurisdictions where we operate our business may result in an adverse and material effect on our business, financial condition and results of operations. We currently have certain pending trademark applications across various jurisdictions, some of which may be subject to governmental scrutiny or third-party objection. We cannot assure you that we would not be subject to trademark infringement claims due to such trademark uses by us, or that we have duly registered all the trademarks necessary for our operations with competent governmental authorities. We may also be subject to other intellectual property infringement claims. As competition intensifies and as litigation becomes a more common method for resolving commercial disputes, we face a higher risk of intellectual property infringement claims. See "— We may become subject to litigation brought by third parties claiming infringement by us of their intellectual property rights."

If we are unsuccessful in obtaining protection for our trademarks, we may be required to change our brand names and may incur significant costs or substantial damages to our businesses in redirecting the existing users and potential users to our new name and may lose audience traffic to a material extent during the process. Any potential conflict over the usage of our brand may expose us to substantial legal costs and administrative penalties and take up the time and energy of our management which could have been used on development of our business. In addition to our registered patents, trademarks, and copyrights, we rely heavily on proprietary know-hows and internal trade secrets. These trade secrets and proprietary know-hows are critical to the performance of our foundation models but are inherently more difficult to protect. If any of our employees or partners misappropriate, disclose, or independently develop similar techniques, our competitive advantage may be materially harmed, and we may have limited legal remedies to enforce our rights.

– 87 –

RISK FACTORS

Moreover, our intellectual property strategy is closely linked to our ability to maintain and enhance our brand reputation. Adverse publicity, with or without merits, relating to events or activities attributed to us, our management, directors, employees, shareholders, business partners or their affiliates, industry, or products similar to ours, may tarnish our reputation and reduce the value of our brand. For instance, unfounded and adversarial statements or opinions could be misleading and could harm our business and reputation. Given the delicate and complex nature of the industry that we operate in, we are vulnerable to such statements or opinions. If we fail to respond to such statements or opinions in a proper manner, our business reputation, financial condition and results of operations may be adversely affected. Moreover, damage to our reputation and loss of brand equity may reduce demand for our products, have an adverse effect on our future financial results, or reduce the trading price of our Class A Ordinary Shares. Rebuilding our reputation and brand equity may also require additional time and significant resources. If we are unable to successfully enhance and protect our reputation, our business operations, results of operations, and financial condition could be materially and adversely affected.

We may become subject to litigation brought by third parties claiming infringement by us of their intellectual property rights.

The industry in which our business operates is characterized by a dense landscape of intellectual property rights, including rights whose scope, validity or enforceability may be uncertain. As a result, even though our business conduct with respect to intellectual property is generally in alignment with industry peers according to CIC, there remains a significant amount of uncertainty in the industry regarding intellectual property protection and infringement, and we cannot be certain that the conduct of our business does not and will not infringing, misappropriating or otherwise violating intellectual property or proprietary rights of third parties. User-generated content and the training materials used in our AI-native products present risks that could lead to allegations that we inadvertently violate intellectual property rights without our knowledge. For example, it could be alleged that datasets used for training may inadvertently include copyrighted material without proper licensing, or users may input content they do not own or have permission to use, potentially giving rights to claims of secondary liability for infringement. AI-generated outputs based on such data could be alleged to be unauthorized derivative works or closely mimic protected content. Additionally, the inclusion of open-source materials with specific usage restrictions, proprietary information such as trade secrets, or data alleged obtained through web scraping in violation of terms of service may further expose us to infringement claims. The complexity and opacity of sourcing and processing large datasets increase the risk of facing the allegations of unintentional violations.

In recent years, there has been significant litigation globally involving patents and other intellectual property rights, including litigation against the foundation model industry. As the legal frameworks governing the business of the global foundation model industry are still evolving, we and our industry peers could become subject to claims and litigation alleging infringement of third-party patents, copyrights or trade secrets. For example, on September 16, 2025, a group of major U.S. movie studio companies, including Disney, Universal and Warner

– 88 –

RISK FACTORS

Bros. Discovery（以下简称"原告"）就我们集团旗下视觉生成平台海螺AI（Hailuo AI）向美国加利福尼亚州中区联邦地区法院提起民事诉讼（以下简称"诉讼"）。由于本案仍处于早期阶段，我们无法确定预测其时间进程、结果、可能产生的损害赔偿金额或所需承担的费用。我们无法保证能够在此案的抗辩中取得胜诉。本案的任何不利结果均可能导致我们须支付金钱赔偿，并使管理层分心于日常运营，从而对我们的业务、经营业绩、财务状况及声誉产生不利影响。请参阅"业务——法律程序及合规"。此外，我们可能面临其他版权所有者提出的类似侵权索赔。任何此类诉讼均可能耗费大量时间和费用方可解决，并可能占用管理层处理业务的时间和精力。此外，若我们从其他科技公司（包括若干潜在竞争对手）招募员工，而该等员工已使用或被指控曾使用某些专有技术、技术或内容，或该等员工参与我们的研究与开发工作，我们可能因此面临该等员工不当使用或披露商业秘密或其他专有信息的索赔。上述索赔及由此引发的任何诉讼，若对我们作出不利裁决，可能使我们承担重大损害赔偿责任，对我们的产品或业务运营施加临时或永久禁令，或令我们的知识产权无效或不可执行。不利判决亦可能导致声誉受损，或迫使我们采取代价高昂的补救措施，例如重新设计产品。此外，由于专利申请从提交到获批可能历时多年，目前可能存在我们尚未知悉的待审申请，该等申请日后获批的专利可能被我们的产品所侵犯。若我们的任何产品侵犯有效且可执行的专利，或若我们希望避免就产品的任何涉嫌侵权行为引发知识产权诉讼，我们可能被禁止销售产品，或选择不销售产品，除非我们获得许可，而该许可可能无法获得，或仅能以商业上不合理、不利或其他不可接受的条款获得。或者，我们可能被迫支付大额特许权使用费，或被迫重新设计一项或多项产品以避免任何侵权行为或相关指控。此外，若用户、业务合作伙伴或第三方因使用我们的产品而遭受侵权诉讼，我们可能须承担向其提供赔偿或其他补救措施的责任。

我们亦可能无法成功重新设计产品以规避任何涉嫌侵权行为。若针对我们的侵权索赔获得胜诉，或我们未能或无法及时以可接受的条款开发和实施非侵权技术或获取被侵权技术的许可，可能对我们的业务及经营业绩产生重大不利影响。此外，无论上述诉讼结果如何，均可能耗费大量时间和费用方可解决，并可能占用管理层处理业务的时间和精力，从而对我们的业务造成严重损害。同时，无论上述诉讼结果如何，均可能严重损害我们在用户、开发者及企业客户中的声誉。

此外，第三方可能对我们提出侵权索赔，包括以从我们等运营公司获取专利许可收入为商业模式的非执业实体有时采取的激进且机会主义的行动。任何此类主张……

– 89 –

RISK FACTORS

regardless of merit, may be time consuming and expensive to resolve and result in litigation or may require us to obtain a license for the intellectual property rights of third parties. Such licenses may not be available or they may not be available on commercially reasonable terms. In addition, as we continue to develop software products and expand our portfolio using new technology and innovation, our exposure to threats of infringement may increase.

If we are unable to ensure compatibility of our products with a variety of hardware and software platforms and software applications developed by others, including our partners, and to ensure effective interoperation with mobile operating systems, networks and mobile devices whose standards we do not control, we may become less competitive and our results of operations may be materially and adversely affected.

Our foundation model products may be integrated with a variety of hardware and software platforms and software applications, and we need to modify and enhance our foundation model products to adapt to changes in hardware and software technologies in a timely and cost-effective manner. Compatibility of our products and hardware and software developed by others is critical to the performance of our products. Failure to ensure compatibility of our products may negatively affect our competitive edge, and our business, financial condition and results of operations could be materially and adversely affected.

We make our products available across a variety of mobile operating systems and devices. We are dependent on the interoperability of our products with popular mobile devices and mobile operating systems that we do not control, such as Android and iOS. Any changes in these mobile operating systems or devices that reduce the functionality of our products or give preferential treatment to competing products may negatively affect the user experience of our products or divert our users to our competitors. In addition, to deliver high-quality products, it is important that our products work well across a range of mobile operating systems, networks, mobile devices and standards that we do not control. If it becomes difficult for our users to access our products, our user growth and market acceptance of our products could be harmed. Furthermore, if the number of platforms for which we develop or adjust our products increases, it will result in an increase in our costs and expenses. Any of the above factors could adversely affect our business and results of operations.

We, our directors, management, employees and shareholders and their affiliates may be subject to lawsuits, contract disputes, employment-related controversies, and other legal and administrative proceedings or fines, which could have a material adverse effect on our business, results of operations, financial condition and reputation.

We may in the future be subject to or involved in lawsuits, contract disputes, employment-related controversies, and other legal proceedings or fines relating to our business operations inside and outside China. Lawsuits that may arise during our operations can involve substantial costs, including the costs associated with investigation, litigation and possible settlement, judgment, penalty or fine. Lawsuits may be costly and time consuming and may require a commitment of management and personnel resources that will be diverted from our normal business operations. There may also be negative publicity associated with litigation that

– 90 –

RISK FACTORS

could decrease consumer acceptance of our products, regardless of whether the allegations are valid or whether we are ultimately found liable. If any of these happens, our business, financial condition, results of operations or liquidity could be materially and adversely affected. In addition, our directors, management, shareholders and employees and their affiliates may from time to time be subject to litigation, regulatory investigations and/or negative publicity or otherwise face potential liability and expense in relation to commercial, labor, employment, securities or other matters, which could adversely affect our reputation and results of operations.

We or certain of our directors or officers may be a target for lawsuits, including putative class action lawsuits brought by shareholders and lawsuits against our directors and officers as a result of their position in other public companies. We cannot assure you that we or our directors or officers will be able to prevail in their defense or reverse any unfavorable judgment on appeal, and we and our directors or officers may decide to settle lawsuits on unfavorable terms. Any adverse outcome of these cases, including any plaintiffs' appeal of the judgment in these cases, could result in payments of substantial monetary damages or fines, or changes to our business practices, and thus materially and adversely affect our business, financial condition, results of operations, cash flows and reputation. Moreover, even if we or our directors or officers eventually prevail in these matters, we could incur significant legal fees or suffer significant reputational harm.

Confidentiality agreements and non-compete covenants with employees and other third parties may not adequately prevent the disclosure of proprietary information.

We have devoted substantial resources to the development of our technology and know-how. We cannot assure you that these agreements will not be breached, that we will have adequate remedies for any breach in time or at all, or that our proprietary technology, know-how or other intellectual property will not otherwise become known to third parties. Similarly, if we recruit employees who breached confidentiality, non-compete covenants with their prior employers, we may become subject to claims that such employees have improperly used or disclosed trade secrets or other proprietary information in violation of their confidentiality, non-compete covenants in a way that benefits us. In addition, others may independently discover trade secrets and proprietary information, limiting our ability to assert any proprietary rights against such parties. Costly and time-consuming litigation could be necessary to enforce and determine the scope of our proprietary rights, and failure to obtain or maintain trade secret protection could adversely affect our competitive position.

– 91 –

RISK FACTORS

Any investments or future acquisitions may have a material adverse effect on our business, reputation, financial condition and results of operations.

We may evaluate and consider a wide array of investment and acquisition opportunities that we believe can extend and solidify our leading market position as part of our overall business strategy. We may be engaged in discussions or negotiations with respect to one or more of these types of transactions. These transactions involve significant challenges and risks, including:

• difficulties in integrating the acquired personnel, operations, products into our operations;

• potential issues with technology, internal controls and financial reporting of the companies we acquire or invest in;

• disruptions of our ongoing business, distractions of the attention of our management and employees and increase of our expenses;

• loss of skilled professionals and established client relationships of the businesses we invest in or acquire;

• for investments over which we do not obtain management and operational control, lack of influence over the controlling partner or shareholder, which may prevent us from achieving our strategic goals in such investments;

• new regulatory requirements and compliance risks that we become subject to as a result of investments or acquisitions in new industries or otherwise;

• actual or alleged misconduct or noncompliance by any company we acquire or invest in (or by its affiliates) that occurred prior to our acquisition or investment, which may lead to negative publicity, government inquiry or investigations against such company or against us;

• unforeseen or hidden liabilities or costs that may adversely affect us following our acquisition of such targets;

• compliance matters including the anti-monopoly and competition laws, rules and regulations of the PRC and other countries in connection with any proposed investments and acquisitions;

• the risk that any of our pending or other future proposed investments or acquisitions does not close;

• the costs of identifying and consummating investments and acquisitions;

– 92 –

RISK FACTORS •

the use of substantial amounts of cash and potentially dilutive issuances of equity securities;

•

the occurrence of significant amortization expenses for other intangible assets; and

•

uncertainties in achieving the expected benefits of synergies and growth opportunities in connection with these acquisitions and investments.

Any such negative developments described above could disrupt our existing business and have a material adverse effect on our business, reputation, financial condition and results of operations.

我们面临与制裁及出口管制法律法规相关的风险，以及国内外人工智能及相关技术法规的不断发展，这可能对我们的业务、财务状况及经营业绩产生重大不利影响。

近年来，国际贸易摩擦持续升级。某些外国司法管辖区已经或可能以各种形式对特定国家、个人及法律实体实施出口管制、经济制裁或其他贸易相关措施，例如征收高额关税或施加苛刻的贸易条件，从而在一定程度上禁止或限制进出口活动。美国及其他司法管辖区或组织，包括欧盟、联合国、英国及澳大利亚，已通过行政命令、立法或其他政府手段，对相关国家或特定行业领域、企业群体或个人及/或组织实施经济制裁措施。

随着中美贸易争端的升级，美国政府可能对美国企业研发的零部件和技术实施额外的出口管制措施。例如，2025年4月9日，美国政府通知英伟达公司（NVIDIA Corporation），美国政府要求对向中国（包括香港及澳门）或总部位于中国或最终母公司位于中国的企业出口英伟达H20集成电路及任何其他达到H20内存带宽、互联带宽或其组合标准的电路进行许可证审批。美国政府表示，该许可证要求旨在应对相关产品可能被用于或转移至中国超级计算机的风险。如果我们的任何业务合作伙伴被列入实体清单或受到其他形式的出口管制，我们的运营可能受到负面影响，这可能导致我们无法获取关键零部件或访问源自美国的最新技术，进而对我们的业务、经营业绩、财务状况及业务前景产生重大不利影响。

国际贸易政策及国际出口管制与经济制裁法律法规正在持续演变。美国商务部工业和安全局（"BIS"）已发布实体清单（"实体清单"），并一直频繁更新该实体清单，将更多中国高科技企业纳入其中。

– 93 –

RISK FACTORS

persons and entities are regularly added to the Entity List and the list of Sanctioned Targets. PRC-based companies on the Entity List are subject to trade sanctions and export controls on a number of components and technologies developed by U.S. companies. Further, new requirements or restrictions could come into effect which might increase the scrutiny on our business or result in one or more of our business activities being deemed to have violated sanctions. We cannot provide any assurance that our future business will be free of sanctions risk, or our business will conform to the expectations and requirements of the authorities of U.S. or any other jurisdictions.

On August 9, 2023, U.S. President Biden issued an executive order and his administration issued an ANPRM providing a conceptual framework for outbound investment controls focused on China, including Hong Kong and Macau. Further to this ANPRM, on June 21, 2024, the U.S. Department of the Treasury issued a proposed rule on outbound U.S. investments involving China that generally follows the ANPRM. On October 28, 2024, the U.S. Department of the Treasury issued the Final Rule, which became effective on January 2, 2025. The Final Rule imposes investment prohibition and notification requirements on U.S. Persons for a wide range of investments in entities associated with China (including Hong Kong and Macau) that are engaged in activities relating to three sectors: (i) semiconductors and microelectronics, (ii) quantum information technologies, and (iii) AI systems, collectively defined as "covered foreign persons." U.S. persons subject to the Final Rule are prohibited from making, or required to report, certain investments in covered foreign persons, which are defined as "covered transactions," and include acquisitions of equity interests (including contingent equity interests), certain debt financing, joint ventures, and certain investments as a limited partner in a non-U.S. person pooled investment fund. The Final Rule excludes some investments from the scope of covered transactions, including certain ones in publicly traded securities. The Final Rule is aimed at exerting greater U.S. government oversight over U.S. direct and indirect investments involving China, and may introduce new hurdles and uncertainties for cross-border collaborations, investments, and funding opportunities of China-based issuers including us. Since our principal place of business is in China and we engage in the development of certain AI models, we are likely to be deemed as a "covered foreign person" as described in the Final Rule. The acquisition of our equity by U.S. persons may be deemed as a "covered transaction" as defined in the Final Rule, and such "covered transaction" is likely to be deemed as a "notifiable transaction," but not a "prohibited transaction," based on the level of computing power used to train the AI systems we develop and the end-uses of such systems. As a result, such U.S. persons may need to make a notification pursuant to the Final Rule. U.S. persons' acquisitions of certain publicly traded securities may be exempted from the prohibition and the notification requirement under the Final Rule (e.g., the publicly traded securities of the Company following the completion of the Global Offering). Investors, including those that are U.S. persons or are subsidiaries of U.S. persons, should consult their legal counsel regarding any potential notification obligations. Certain Underwriters have informed us that they may consider making notifications with the U.S. Department of the Treasury. None of the Underwriters has any obligation to inform us or any investor if they later decide that they will not file such notifications. No publicly available precedent exists regarding the application of the OIP regulations by the U.S. Department of the Treasury or by any court or other regulatory, judicial or legal authority to specific transactions. In addition, the technologies of our business could change such that we are engaged in "covered

– 94 –

RISK FACTORS

activities" that trigger the OIP's prohibitions, or the OIP may be changed by executive actions of the U.S. government, including modifications to the scope of activities and technologies that are subject to prohibitions or notification requirements, or changes to the scope and availability of applicable exceptions to the OIP's prohibitions or notification requirements.

Specifically, on January 20, 2025, the U.S. government issued a national security presidential memorandum entitled "America First Trade Policy", which, among other things, directs the Secretary of the Treasury and several other executive departments and agencies to review the OIP to determine whether it contains "sufficient controls to address national security threats" and to determine whether the executive order implementing the OIP "should be modified or rescinded and replaced." In addition, on February 21, 2025, the U.S. government issued another national security presidential memorandum entitled "America First Investment Policy", which, among other things, states that the U.S. government will consider potential expansion of the OIP to a wider range of technology sectors, including biotechnology, hypersonics, aerospace, advanced manufacturing, directed energy, and other areas "implicated by the PRC's national Military-Civil Fusion strategy," and application of restrictions to a broader range of investments, including "publicly traded securities". On April 3, 2025, the U.S. government further stated that it intends to evaluate whether the scope of outbound investment restrictions should be expanded "to be responsive to developments in technology and the strategies of countries of concern."

Changes to our technologies or to the OIP could limit, or in the worst-case scenario, eliminate our ability to raise capital or contingent capital (such as convertible instruments) from U.S. investors in the future. Our ability to raise such capital may be significantly and adversely affected, which could negatively impact our capital-raising capacity and our business, financial condition and prospects. In addition, changes to the Publicly Traded Securities Exception or other aspects of the OIP could restrict or prohibit the purchase or trading of our Shares by U.S. persons, impose new notification or other regulatory requirements, or otherwise make our Shares less attractive to investors. In such circumstances, the value and liquidity of our Shares may be materially and adversely affected, and in extreme cases, our Shares could experience significant declines in trading value.

The successful operation of our business depends on the performance and reliability of the Internet infrastructure and telecommunications networks in the countries where we operate.

Our business depends in part on the performance, reliability and security of the telecommunications and Internet infrastructure in the countries where we operate. In the event of disruptions, failures or other problems with the relevant Internet infrastructure, our business operation may be adversely affected. In addition, the Internet infrastructure in the countries in which we operate may not support the demands associated with continued growth in Internet usage.

– 95 –

RISK FACTORS

电信网络运营商未能向我们提供所需带宽，亦可能影响我们产品的速度及可用性。我们无法控制电信运营商所提供服务的成本。若我们为电信及互联网服务所支付的价格大幅上涨，我们的利润率可能受到不利影响。此外，若互联网接入费用或其他向用户收取的费用有所提高，我们产品的用户群体可能随之减少，进而可能导致我们的收入大幅下降。

尤其是，若域名安全受到损害，我们将无法在业务运营中使用相关域名，可能对我们的业务运营、声誉及品牌形象造成重大不利影响。若我们未能对通过所依赖的电信及互联网运营商网络传输的数据实施充分加密，则存在电信及互联网运营商或其业务合作伙伴盗用我们数据的风险，可能对我们的业务运营及声誉造成重大不利影响。

**我们依赖第三方运营的云服务及基础设施，若该等第三方服务及基础设施出现中断或干扰，将对我们的业务、经营业绩及财务状况产生不利影响。**

我们通过多家第三方云服务及基础设施提供商提供基础模型产品。我们的第三方云服务及基础设施提供商可能遭遇各类问题，包括但不限于软硬件故障、电力短缺或自然灾害，此类情况可能使我们面临第三方云服务及基础设施中断、延迟或停运的风险。上述第三方提供商所提供的云服务及基础设施水平，或相关云服务或基础设施出现的定期或长期中断，可能影响用户对我们产品的使用体验及满意度，并可能损害我们的声誉。

此外，在某些情况下，我们的云服务及基础设施提供商可能停止或限制我们对一项或多项服务的访问，或终止与我们的合同关系，或寻求终止与我们的合同关系。若我们与现有第三方提供商的合同关系遭到终止，我们向用户提供服务的能力可能出现暂时中断，且在寻找替代云服务及基础设施提供商的过程中可能产生额外成本。

由于上述原因，我们可能遭遇运营暂时中断，导致用户不满，亦可能产生额外成本或面临实际或潜在的法律责任，上述任何情况均可能对我们的业务、经营业绩及财务状况产生不利影响。

– 96 –

RISK FACTORS

网络攻击等对我们IT系统或第三方服务提供商IT系统的干扰和未经授权的访问可能对我们的业务运营、经营业绩、声誉及财务状况产生重大不利影响。

我们的产品和技术可能使我们能够访问数据或信息，这对可能寻求对我们或我们的供应商或服务提供商实施网络攻击的恶意行为者而言具有很强的吸引力。我们或我们的服务提供商的安全措施遭受实际或被认为存在的漏洞，或未能维持我们产品和技术平台（包括我们所依赖的第三方云平台及信息技术（IT）服务）的可靠性、安全性和完整性，可能使我们面临重大后果。我们无法保证我们的IT系统或第三方服务提供商的IT系统能够完全防范第三方入侵、病毒、黑客攻击、勒索软件攻击及其他网络攻击、信息或数据盗窃或其他类似威胁。此外，经第三方授权或许可并纳入我们产品和技术中的软件可能带来某些与网络安全相关的风险，例如此类软件普遍缺乏支持，可能导致漏洞，进而危害我们系统的安全性。有关我们使用开源软件相关风险的更多详情，请参阅"——我们将某些模型和产品以开源方式提供，并可能使用开源技术，这可能对我们的业务带来特定风险"。

因此，我们的系统、服务器和设备以及我们服务提供商的系统、服务器和设备可能遭受上述事件的影响，这可能导致我们IT系统受损、业务遭受重大中断，或我们或用户的商业信息、商业秘密、用户数据及其他机密或专有信息遭到盗窃、无法访问、不当披露或被盗用。即使我们利用备份信息进行恢复，任何此类事件也可能对我们的业务产生重大不利影响。后果可能包括法律和财务风险敞口、业务和用户流失、商业秘密或其他专有信息或个人信息的丢失或未经授权披露，并可能引发诉讼（包括集体诉讼以及用户基于客户协议和其他商业安排对我们提出的诉讼及赔偿索赔）、监管行动和罚款、消费者保护行动、其他相关费用（包括与我们的调查和补救工作相关的费用）以及对我们声誉的重大损害。这可能阻碍我们留住现有用户和商业合作伙伴以及吸引新合作伙伴和用户的能力。在我们遭受网络攻击或安全漏洞的情况下，我们可能无法成功实施补救计划以应对风险敞口和未来损害。此外，我们未购买与网络安全事件相关的保险，因此因任何网络攻击或安全漏洞所产生或与之相关的任何费用或成本（可能相当可观）将由我们自行承担。任何此类实际或被认为存在的干扰、访问、漏洞、不确定性或事件均可能对我们的业务运营、经营业绩及财务状况产生重大不利影响。

– 97 –

RISK FACTORS

We may not have sufficient insurance coverage to cover our business risks.

We believe we maintain insurance policies in line with industry standards. We do not maintain business interruption insurance, key-man life insurance or litigation insurance. Any uninsured occurrence of business disruption, litigation or natural disaster, or significant damages to our uninsured equipment or facilities could have a material adverse effect on our results of operations. Our current insurance coverage may not be sufficient to prevent us from any loss and there is no certainty that we will be able to successfully claim our losses under our current insurance policy on a timely basis, or at all. If we incur any loss that is not covered by our insurance policies, or the compensated amount is significantly less than our actual loss, our business, financial condition and results of operations could be materially and adversely affected. If such risk materializes, we may also suffer substantial losses as we do not have insurance coverage.

与我们财务状况及额外资本需求相关的风险

我们需要承担大量经营支出，且未来可能需要筹集额外资本，而此类资本可能无法以我们可接受的条款筹得，或根本无法筹得。若我们无法在需要时以有利条款筹集额外资金，我们的运营及前景可能受到不利影响。

我们产品的开发将要求我们定期承担经营支出，以维持我们的服务水平。竞争条件的变化或基础模型产品领域出现的任何重大技术进步，可能要求我们投入大量资本以保持竞争力。截至2025年9月30日，我们的股东权益总亏损为13.035亿美元（US$1,303.5 million），且我们在2022年、2023年、2024年及截至2025年9月30日的九个月内均录得净亏损。2024年，我们65.1%的经营支出用于研究与开发活动。若我们无法为上述投资提供资金，或未能在研究与运营方面进行必要投入，我们的业务、经营业绩或财务状况可能受到不利影响。我们的经营支出需求将取决于多种因素，包括但不限于：

• 技术进步；

• 市场对我们产品的接受程度及我们产品的整体销售水平；

• 研究与开发支出；

• 我们的成本控制能力；

• 销售及营销支出；

• 对我们系统及设施的升级改善；

— 98 —

RISK FACTORS •

potential acquisitions of businesses and solution lines; and

•

general economic conditions, including the effects of international conflicts and their impact on the global foundation model industry in particular.

Furthermore, if our capital requirements are materially different from those currently planned, we may need additional capital sooner than anticipated. If additional funds are raised through the issuance of equity or convertible debt securities, the percentage ownership of our shareholders at that point in time will be reduced. Additional financing may not be available on favorable terms, on a timely basis, or at all. If adequate funds are not available or are not available on acceptable terms, we may be unable to continue our operations as planned, develop or enhance our products, expand our sales and marketing programs, take advantage of future opportunities or respond to competitive pressures.

未能履行合同负债相关义务可能对我们的流动性和财务状况产生不利影响。

Our contract liabilities primarily arise from advance payments made by our users for services that have not yet been delivered. As of December 31, 2022, 2023, 2024 and September 30, 2025, we had contract liabilities of approximately nil, US$0.6 million, US$1.6 million and US$4.7 million. For further details, see "Financial Information — Liabilities — Contract Liabilities." There is no assurance that we will be able to fulfil our obligations in respect of contract liabilities. If we have any difficulties or fail to perform our obligations under our contracts, our relationships with our users will be adversely affected and we will be unable to recognize such contract liabilities as revenue, exposing us to the risk of shortfalls in liquidity, which may have a material adverse effect on our operational performance and prospects.

Failure to fulfill our obligations in respect of contract liabilities could adversely affect our liquidity and financial condition.

Our contract liabilities primarily arise from advance payments made by our users for services that have not yet been delivered. As of December 31, 2022, 2023, 2024 and September 30, 2025, we had contract liabilities of approximately nil, US$0.6 million, US$1.6 million and US$4.7 million. For further details, see "Financial Information — Liabilities — Contract Liabilities." There is no assurance that we will be able to fulfil our obligations in respect of contract liabilities. If we have any difficulties or fail to perform our obligations under our contracts, our relationships with our users will be adversely affected and we will be unable to recognize such contract liabilities as revenue, exposing us to the risk of shortfalls in liquidity, which may have a material adverse effect on our operational performance and prospects.

We are subject to credit risk related to delay in payment and defaults of users or related parties, which would adversely affect our liquidity and financial condition.

We are exposed to credit risk related to delay in payment and defaults of our various users. As of December 31, 2022, 2023, 2024 and September 30, 2025, our trade receivables amounted to nil, US$1.3 million, US$7.0 million and US$8.1 million, respectively, and our current version of prepayments, other receivables and other assets amounted to US$0.6 million, US$4.4 million, US$13.5 million and US$11.8 million respectively. We may not be able to collect all such trade receivables and prepayments, other receivables and other assets due to a variety of factors that are beyond our control, including long payment cycles of certain of our suppliers, adverse operating condition or financial condition of users, and users' inability to pay. If our users delay or default in their payments to us, we may have to make impairment provisions and write-off the relevant receivables and hence our liquidity and financial condition would be adversely affected.

– 99 –

RISK FACTORS

Fluctuations in changes in fair value of our financial assets at fair value through profit or loss would affect our financial results.

We have invested in, and intend to continue to selectively invest in, businesses, assets and technologies that complement our existing business and may make other financial investments. We recorded financial assets at fair value through profit or loss of US$65.8 million, US$15.8 million, US$390.6 million and US$714.4 million as of December 31, 2022, 2023, 2024 and September 30, 2025, respectively. These financial assets at fair value through profit or loss included our investments in structured wealth management products. The fair value changes in our financial assets measured at fair value through profit or loss may negatively affect our financial performance. The fair value of financial instruments that are not traded in an active market is determined by using valuation techniques. These valuation techniques maximize the use of observable market data where it is available and rely as little as possible on entity specific estimates. Any change in the estimates and assumptions may lead to a change in the fair value of the financial assets, which in turn could negatively affect our financial conditions and results.

Share-based payments may have a material and adverse effect on our financial performance and cause shareholding dilution to our Shareholders.

The share incentive plan was established for the benefit of our directors, senior management and core employees as remuneration for their services provided to us and to incentivize and reward the eligible persons who have contributed to the success of our Company. For the principal terms of the employee incentive scheme, see "Appendix IV — Statutory and General Information — D. Share Incentive Plans." In 2022, 2023, 2024 and the nine months ended September 30, 2025, we recorded US$1.1 million, US$3.3 million, US$6.8 million and US$8.6 million, respectively, in share-based payments.

To further incentivize our employees, we may incur additional share-based payment expenses in the future. We believe such share-based awards are important to our ability to attract, retain and motivate our key personnel, and we may continue to grant share-based awards in the future. Expenses incurred with respect to such share-based payments may also increase our operating expenses and therefore have a negative effect on our financial performance. Issuance of additional Shares with respect to such share-based payments may dilute the shareholding of our Shareholders and could result in a decline in the value of our Class A Ordinary Shares.

We may be subject to higher income tax rates if certain preferential tax treatments granted to us become unavailable or are not renewed.

Our PRC subsidiaries are subject to the PRC corporate income tax at a standard rate of 25% on their taxable income, but certain of our PRC subsidiaries were accredited as "High and New Technology Enterprises," and are entitled to a preferential income tax rate of 15%. We cannot assure you that the PRC policies on preferential tax treatments will not change or that the current preferential tax treatments we enjoy or will be entitled to enjoy will not be

– 100 –

RISK FACTORS

canceled. Moreover, we cannot assure you that our PRC subsidiaries will be able to renew the same preferential tax treatments upon expiration. If any such change, cancelation or discontinuation of preferential tax treatment occurs, the relevant PRC subsidiaries will be subject to the PRC enterprise income tax, or EIT, at a rate of 25% on taxable income. As a result, the increase in our tax charge could materially and adversely affect our results of operations.

To address any ESG-related risks, we may incur additional costs, which may materially and adversely affect our financial performance.

To identify, manage, and mitigate ESG-related risks, we may incur additional costs and expenses which could impact our financial performance. Given the nature of our business, we do not produce any material generation of emissions and wastes, and we do not produce any heavy pollution. Nonetheless, we monitor environmental and climate-related risks that may impact our business, strategy and financial performance. We also evaluate the magnitude of the resulting impact over the short-, medium- and long-term horizons. We monitor a wide range of indicators to manage our environmental and climate-related risks arising from our operations and are committed to providing adequate support to our employees to nurture a friendly and inspirational corporate culture. This commitment may entail incurring substantial additional costs and would potentially impact our profitability. For further details, see "Business — Environmental, Social and Governance."

In addition, the increasing ESG-related regulatory requirements, including various ESG disclosure mandates in the jurisdictions where we operate, may lead to rising compliance costs and cost of sales may rise. Failure to adapt to new regulations or meet evolving industry expectations and standards could result in consumers choosing products from other companies, which may materially and adversely affect our results of operations and financial conditions.

RISKS RELATED TO DOING BUSINESS IN THE GEOGRAPHIC MARKETS IN WHICH WE OPERATE

Changes in the political, economic and social conditions of the geographic markets in which we operate may materially and adversely affect our business, financial condition and results of operations.

Our business, financial condition and results of operations may be influenced to a significant degree by political, economic and social conditions in the geographic markets in which we operate, particularly those with emerging or evolving regulatory frameworks for AI and foundation model technologies.

In certain markets, local governments continue to play an active role in shaping the technology sector, including through licensing regimes, export controls, foreign ownership restrictions, or preferential treatment for domestic players. In some cases, governments may impose restrictions on access to computing infrastructure, or introduce national security reviews of AI and foundation model technologies, which could adversely affect our

– 101 –

RISK FACTORS

commercialization pace, product deployment or international expansion. Actions taken by governments to manage inflation, devalue currencies, impose capital controls, or regulate technology exports or imports may also materially impact our operations.

Additionally, political or social instability—including policy unpredictability, trade tensions, protests, or deteriorating diplomatic relations—could increase compliance costs, limit access to key resources or partnerships, or disrupt our operations and strategic planning. These risks are heightened in jurisdictions where the regulatory environment is rapidly changing or where geopolitical tensions may affect access to talent, infrastructure, or markets for AI and foundation model technologies.

We may be subject to the approval, filing or other requirements of the CSRC or other PRC governmental authorities in connection with capital raising activities.

On July 6, 2021, the relevant PRC government authorities issued the Opinions on Strictly Cracking Down Illegal Securities Activities in Accordance with the Law (《關於依法從嚴打擊證券違法活動的意見》). These opinions emphasized the need to strengthen the administration over illegal securities activities and the supervision on overseas listings by China-based companies and proposed to take effective measures, such as promoting the construction of relevant regulatory systems to deal with the risks and incidents faced by China-based overseas-listed companies.

On February 24, 2023, the CSRC and other relevant government authorities published the Provisions on Strengthening Confidentiality and Archives Administration of Overseas Securities Offering and Listing by Domestic Companies (《關於加強境內企業境外發行證券和上市相關保密和檔案管理工作的規定》) (the "Archives Rules"), which came into effect on March 31, 2023. The Archives Rules require that, in relation to the overseas securities offering and listing activities of domestic enterprises, either in direct or indirect form, such domestic enterprises, as well as securities companies and securities service institutions providing relevant securities services, are required to strictly comply with relevant requirements on confidentiality and archives management, establish a sound confidentiality and archives system, and take necessary measures to implement their confidentiality and archives management responsibilities. The interpretation and implementation of the Archives Rules may keep evolving, failure to comply with which may materially affect our business, results of operations or financial conditions.

Furthermore, we cannot assure you that new rules or regulations promulgated in the future will not impose additional requirements or restrictions on us, our shareholders or our financing activities. We or our shareholders may not be able to comply with such additional requirements in a timely manner. In addition, we or our shareholders may be subject to sanctions by the CSRC or other PRC regulatory authorities for failure to seek CSRC filing or other government authorization or approval for this listing or any subsequent change in shareholding structure, it is uncertain whether we can or how long it will take us or our shareholders to obtain such approval or complete such administrative procedures and these regulatory authorities may impose fines and penalties on us or our shareholders, limit our operating activities in the PRC,

– 102 –

RISK FACTORS

limit our ability to pay dividends outside the PRC, delay or restrict the repatriation of the proceeds from the Global Offering into the PRC or take other actions to restrict our financing activities, which could have a material adverse effect on our business.

We face exposure to foreign currency exchange rate fluctuations, and such fluctuations could adversely affect our financing arrangements, business operations, results of operations, and financial condition.

As we expand globally with our users, we become increasingly exposed to the effects of fluctuations in currency exchange rates, especially its potential impact on our financing arrangements. The value of the Renminbi against the U.S. dollar and other currencies has fluctuated significantly in the past, and may in the future continue to do so, affected by, among other things, changes in political and economic conditions and the foreign exchange policy adopted by the PRC government. We recorded other comprehensive gains from exchange differences on translation of foreign operations of US$0.1 million, US$0.4 million, US$0.3 million and other comprehensive losses of US$0.1 million and US$1.3 million in 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively, due to the fluctuations of U.S. dollar/RMB exchange rate when translating results and financial positions of the Company and its subsidiaries inside mainland China from their functional currency RMB into our presentation currency U.S. dollar. We recorded net foreign exchange gains of US$0.2 million, US$0.3 million, US$2 thousand, US$1.4 million and US$1.6 million in 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively, due to the fluctuation of U.S. dollar/RMB exchange rate when translating monetary assets and liabilities denominated in foreign currencies in terms of the functional currency of the Company and its subsidiaries. For details, see Note 32 to the Accountant's Report set out in Appendix I to this Prospectus.

We are a holding company, and we may rely on dividends paid by certain of our subsidiaries for our cash needs. We face translation exposure to fluctuations in currency exchange rates, which could hinder our ability to predict our future results and earnings and affect our operating results. To the extent that we need to convert any foreign currencies we receive from this Global Offering into Renminbi for our operations, appreciation of the Renminbi against such foreign currencies would have an adverse effect on the Renminbi amount we would receive. We cannot assure you that the Renminbi will not appreciate or depreciate significantly in value against the foreign currencies in the future. If we decide to convert our Renminbi into foreign currencies for making payments toward our financing, for dividends on our Offer Shares, or for other business purposes, appreciation of the foreign currency against the Renminbi would have a negative effect on the foreign currency amount, adversely affecting our financial position. Therefore, any significant fluctuation of Renminbi against the foreign currency could adversely affect our business, results of operations and financial condition, and the value of any dividends payable in foreign currencies.

– 103 –

RISK FACTORS

You may experience difficulties in effecting service of legal process, enforcing foreign judgments or bringing actions in China against us or our management named in the document based on foreign laws.

Substantially all of our operations are located in the PRC. In addition, almost all of our Directors, supervisors and officers reside in China and substantially all of their assets are located in China. It may be difficult for investors to effect service of process upon those persons residing in China or to enforce against us or them in China any judgments obtained from non-PRC courts. The PRC does not have treaties providing for the reciprocal recognition and enforcement of judgments of courts of most other jurisdictions. As a result, recognition and enforcement in the PRC of judgments of a court in any of these jurisdictions outside China may be difficult.

On July 14, 2006, the Supreme People's Court of the PRC and the Government of the Hong Kong Special Administrative Region signed an Arrangement on Reciprocal Recognition and Enforcement of Judgments in Civil and Commercial Matters Pursuant to Choice of Court Agreements between Parties Concerned (《最高人民法院關於內地與香港特別行政區法院相互認可和執行當事人協議管轄的民商事案件判決的安排》) (the "Arrangement"). Under the Arrangement, a party with an enforceable final court judgment rendered by any designated people's court of China or any designated Hong Kong court requiring payment of money in a civil and commercial case according to a written choice of court agreement, may apply for recognition and enforcement of the judgment in the relevant people's court of China or Hong Kong court. A written choice of court agreement is defined as any agreement in writing entered into between parties after the effective date of the Arrangement in which a Hong Kong court or a PRC court is expressly designated as the court having sole jurisdiction for the dispute.

On January 18, 2019, the Supreme People's Court of the PRC and Hong Kong entered into an Arrangement on Reciprocal Recognition and Enforcement of Judgments in Civil and Commercial Cases by the Courts of the Mainland and of the Hong Kong Special Administrative Region (《關於內地與香港特別行政區法院相互認可和執行民商事案件判決的安排》) (the "New Arrangement"). The New Arrangement will broaden the scope of judgments that may be enforced between China and Hong Kong under the Arrangement. Whereas a choice of jurisdiction needs to be agreed in writing in the form of an agreement between the parties for the selected jurisdiction to have exclusive jurisdiction over a matter under the Arrangement, the New Arrangement provides that the court where the judgment was sought could apply jurisdiction in accordance with certain rules without the parties' agreement. The New Arrangement will replace the Arrangement when the former becomes effective. The New Arrangement became effective on January 29, 2024 both in China and in Hong Kong. Under the New Arrangement, any party concerned may apply to the relevant PRC court or Hong Kong court for recognition and enforcement of the effective judgments in civil and commercial cases subject to the conditions set forth in the New Arrangement. Moreover, under the Civil Procedure Law of the PRC (《中華人民共和國民事訴訟法》), if a court of China rules that a foreign judgment violates the basic principles of PRC laws or national sovereignty, security, or public interest, the PRC court may not enforce the foreign judgment against our assets or managements in China.

– 104 –

RISK FACTORS

We may rely on dividends and other distributions on equity paid by our subsidiaries to fund any cash and financing requirements we may have, and any limitation on the ability of our subsidiaries to make payments to us could have a material and adverse effect on our ability to conduct our business.

We are a Cayman Islands holding company, and we may rely principally on dividends and other distributions on equity from our subsidiaries for our cash requirements, including for services of any debt we may incur. For example, our PRC subsidiaries' ability to distribute dividends is based upon their distributable earnings. Current PRC regulations permit our PRC subsidiaries to pay dividends to their respective shareholders only out of their accumulated profits, if any, determined in accordance with PRC accounting standards and regulations. In addition, each of our PRC subsidiaries are required to set aside at least 10% of its after-tax profits each year, if any, to fund a statutory reserve until such reserve reaches 50% of each of their registered capitals. These reserves are not distributable as cash dividends. If our PRC subsidiaries incur debt on their own behalf in the future, the instruments governing the debt may restrict their ability to pay dividends or make other payments to us. Any limitation on the ability of our PRC subsidiaries to distribute dividends or other payments to their respective shareholders could materially and adversely limit our ability to grow, make investments or acquisitions that could be beneficial to our businesses, pay dividends or otherwise fund and conduct our business.

We are subject to PRC regulation of loans to and direct investment in PRC entities by offshore holding companies and governmental regulations of currency conversion when we use the proceeds of this Global Offering to make loans or additional capital contributions to our PRC subsidiaries.

We are an offshore holding company conducting our operations in China through our PRC subsidiaries. We may make loans to our PRC subsidiaries subject to the approval from governmental authorities and limitation of amount, or we may make additional capital contributions to our PRC subsidiaries in China. Any loans to our PRC subsidiaries in China, which are treated as foreign-invested enterprises under PRC law, are subject to PRC regulations and foreign exchange loan registrations. For example, loans by us to our PRC subsidiaries in China to finance their activities cannot exceed statutory limits and must be registered with the local counterpart of SAFE. In addition, a foreign invested enterprise shall use its capital pursuant to the principle of authenticity and self-use within its business scope. The capital of a foreign invested enterprise shall not be used for the following purposes: (i) directly or indirectly used for payment beyond the business scope of the enterprises or the payment prohibited by relevant laws and regulations; (ii) directly or indirectly used for investment in securities investments other than banks' principal-secured products unless otherwise provided by relevant laws and regulations; (iii) the granting of loans to non-affiliated enterprises, except where it is expressly permitted in the business license; and (iv) paying the expenses related to the purchase of real estate that is not for self-use (except for the foreign-invested real estate enterprises). See "Regulatory Overview — Regulations Relating to Foreign Exchange" for details on foreign exchange related regulations.

– 105 –

RISK FACTORS

In light of the various requirements imposed by PRC regulations on loans to and direct investment in PRC entities by offshore holding companies, we cannot assure you that we will be able to complete the necessary government registrations or obtain the necessary government approvals on a timely basis, if at all, with respect to future loans to our PRC subsidiaries or future capital contributions by us to our wholly foreign-owned subsidiaries in China. As a result, uncertainties exist as to our ability to provide prompt financial support to our PRC subsidiaries when needed. If we fail to complete such registrations or obtain such approvals, our ability to use the proceeds we expect to receive from this Global Offering and to capitalize or otherwise fund our PRC operations may be negatively affected, which could materially and adversely affect our liquidity and our ability to fund and expand our business.

Governmental regulation of currency conversion may limit our ability to utilize our revenue effectively and affect the value of your investment.

The conversion of Renminbi is subject to applicable laws and regulations in the PRC. We receive most of our payments from users in Renminbi and may need to convert Renminbi into foreign currencies for the payment of dividends, if any, to holders of our Class A Ordinary Shares. Under the Chinese existing foreign exchange regulations, following the completion of the Global Offering, we will be able to pay dividends in foreign currencies without prior approval from SAFE or its local branches by complying with certain procedural requirements. However, we may not be able to pay dividends in foreign currencies to our Shareholders if access to foreign currencies for current account transactions is restricted in the future. Foreign exchange transactions under our capital account continue to be subject to foreign exchange controls and require the approval of the SAFE or its local branches. These limitations could affect our ability to obtain foreign exchange through equity financing, or to obtain foreign exchange for capital expenditures.

Most of our revenue and costs are denominated in Renminbi. Any significant revaluation of the Renminbi may materially and adversely affect our results of operations, cash flows and financial condition. Since 1994, the conversion of the Renminbi into foreign currencies, including U.S. dollars, has been based on rates set by the People's Bank of China, which are set daily based on the previous business day's interbank foreign exchange market rates and current exchange rates on the world financial markets. It is difficult to predict how market forces or government policies may impact the exchange rate between the Renminbi and the Hong Kong dollar, the U.S. dollar or other currencies in the future.

Changing international circumstances could result in appreciation of the Renminbi against the U.S. dollar, the Hong Kong dollar or other foreign currencies. If the Renminbi appreciates against other currencies significantly, and as we need to convert and remit the proceeds from the Global Offering and future financing into the Renminbi for our operations, appreciation of the Renminbi against the relevant foreign currencies would reduce the Renminbi amount we would receive from the conversion. On the other hand, because the dividends on our Class A Ordinary Shares, if any, will be paid in Hong Kong dollars, any devaluation of the Renminbi against the Hong Kong dollar could reduce the amount of any cash dividends on our Class A Ordinary Shares in Hong Kong dollar terms. In addition, there are

– 106 –

RISK FACTORS

limited instruments available for us to reduce our exposure to foreign currency risk at reasonable costs. Any of the foregoing factors may materially and adversely affect our businesses, results of operations, financial condition and prospects.

PRC regulations establish related procedures for some acquisitions of Chinese companies by foreign investors, which could make it complicated for us to pursue growth through acquisitions in China.

Among other things, the Regulations on Mergers and Acquisitions of Domestic Enterprises by Foreign Investors (《关于外国投资者并购境内企业的规定》), or the M&A Rules, adopted by six PRC regulatory agencies in 2006 and amended in 2009, established specific procedures and requirements for merger and acquisition activities by foreign investors. Such regulation requires, among other things, that MOFCOM be notified in advance of any change of control transaction in which a foreign investor takes control of a PRC domestic enterprise, if (i) any important industry is concerned, (ii) such transaction involves factors that have or may have impact on the national economic security, or (iii) such transaction will lead to a change in control of a domestic enterprise which holds a famous trademark or PRC time-honored brand. Moreover, the Anti-Monopoly Law of the PRC (《中华人民共和国反垄断法》) promulgated by the Standing Committee of the NPC which became effective in 2008 and last amended in 2022 requires that transactions which are deemed concentrations and involve parties with specified turnover thresholds must be cleared by the relevant anti-monopoly authority before they can be completed.

On December 19, 2020, the NDRC and MOFCOM jointly promulgated the Measures for the Security Review of Foreign Investment (《外商投资安全审查办法》), effective on January 18, 2021, setting forth provisions concerning the security review mechanism on foreign investment, and stating that any foreign investment that has or may have an impact on national security shall be subject to security review in accordance with the provisions thereof. According to the measures, foreign investment includes a foreign investor acquires the equity or assets of any enterprise in China by means of merger and acquisition, and a foreign investor makes investment in China by other means. Foreign investor or relevant parties in China must declare the security review to the working mechanism office prior to the investments in, among other industries, important cultural products and services, important information technology and internet products and services, important financial services, key technologies and other important fields relating to national security, while obtaining control over the enterprises invested in.

We may pursue potential strategic acquisitions that are complementary to our business and operations. Complying with the requirements of these regulations to complete such transactions could be costly, and any required approval processes, including obtaining approval or clearance from the competent governmental authority, may delay or inhibit our ability to complete such transactions, which could affect our ability to expand our business or maintain our market share.

– 107 –

RISK FACTORS

与中国居民设立境外特殊目的公司相关的中国法规可能使我们的中国居民实际权益拥有人或我们的中国子公司承担责任或受到处罚，限制我们向中国子公司注资的能力，限制我们中国子公司增加注册资本或向我们分配利润的能力，或对我们产生其他不利影响。

2014年7月，国家外汇管理局颁布了《国家外汇管理局关于境内居民通过特殊目的公司境外投融资及返程投资外汇管理有关问题的通知》（以下简称"汇发37号文"），以取代《关于境内居民通过境外特殊目的公司融资及返程投资外汇管理有关问题的通知》（以下简称"汇发75号文"）。汇发75号文自汇发37号文颁布之日起停止执行。汇发37号文要求中国居民（包括中国境内个人及中国境内企业法人）就其直接或间接的境外投资活动向国家外汇管理局或其地方分支机构进行登记。汇发37号文适用于我们持股的中国居民股东，亦可能适用于我们未来进行的任何境外收购。

汇发37号文要求中国居民就直接或间接控制境外实体事宜向中国政府主管部门进行登记并获得批准。汇发37号文对"控制"一词的定义较为宽泛，是指中国居民通过收购、信托、代理、表决权、回购、可转换债券或其他安排，在境外特殊目的载体（以下简称"SPV"）中取得的经营权、受益权或决策权。此外，任何直接或间接持有SPV股份的中国居民须就该SPV向国家外汇管理局地方分支机构更新其已备案的登记信息，以反映任何重大变更。2015年2月13日，国家外汇管理局颁布了《国家外汇管理局关于进一步简化和改进直接投资外汇管理政策的通知》（以下简称"汇发13号文"），并于2015年6月1日起正式实施。根据汇发13号文，包括汇发37号文项下所要求的外商直接投资及境外直接投资的外汇登记申请，将向具有资质的银行提交，而非向国家外汇管理局提交。具有资质的银行将在国家外汇管理局的监督下直接审核申请并受理登记。

上述法规可能对我们目前及未来的架构安排和投资活动产生重大影响。我们无法保证本公司的任何中国股东或我们所投资的任何中国企业均能遵守上述要求。上述个人或实体若未能遵守或无法遵守国家外汇管理局的相关规定，可能导致我们遭受罚款或法律制裁，例如我们的跨境投资活动或我们中国子公司向本公司分配股息、从本公司取得外币贷款的能力受到限制，或导致我们无法向本公司进行分配或支付股息。因此，我们的业务经营及向您进行分配的能力可能受到重大不利影响。

— 108 —

RISK FACTORS

Furthermore, with the promulgation of new laws, regulations and standards concerning foreign exchange regulations in the future, we are required to comply with these laws, regulations and standards concerning offshore or cross-border transactions, otherwise we may be subject to fines or other penalties, which could materially and adversely affect our business, results of operations and financial condition. This may restrict our ability to implement our acquisition strategy and could adversely affect our business and prospects.

If we are classified as a PRC resident enterprise for PRC enterprise income tax purposes, such classification could result in unfavorable tax consequences to us and our non-PRC shareholders.

Under the PRC Enterprise Income Tax Law (《中华人民共和国企业所得税法》) and its implementation rules, an enterprise established outside the PRC with its "de facto management body" within the PRC is considered a "resident enterprise" and will be subject to the enterprise income tax on its global income at the rate of 25%. The implementation rules define the term "de facto management body" as the body that exercises full and substantial control and overall management over the business, productions, personnel, accounts and properties of an enterprise. The State Administration of Taxation, or SAT, issued the Notice Regarding the Determination of Chinese-Controlled Offshore Incorporated Enterprises as People's Republic of China Tax Resident Enterprises on the Basis of De Facto Management Bodies (《关于境外注册中资控股企业依据实际管理机构标准认定为居民企业有关问题的通知》), known as SAT Circular 82, on April 22, 2009 and most recently amended on December 29, 2017. SAT Circular 82 provides certain specific criteria for determining whether the "de facto management body" of a PRC-controlled enterprise that is incorporated offshore is located in China. The criteria set forth in the circular may reflect the SAT's general position on how the "de facto management body" text should be applied in determining the tax resident status of all offshore enterprises.

According to SAT Circular 82, an offshore incorporated enterprise controlled by a PRC enterprise or a PRC enterprise group will be regarded as a PRC tax resident by virtue of having its "de facto management body" in China, and will be subject to PRC enterprise income tax on its global income only if all of the following conditions are met: (i) the primary location of the day-to-day operational management is in the PRC; (ii) decisions relating to the enterprise's financial and human resource matters are made or are subject to approval by organizations or personnel in the PRC; (iii) the enterprise's primary assets, accounting books and records, company seals, and board and shareholder resolutions are located or maintained in the PRC; and (iv) at least 50% of voting board members or senior executives habitually reside in the PRC.

We believe our Company is not a PRC resident enterprise for PRC tax purposes. However, the tax resident status of an enterprise is subject to determination by the PRC tax authorities and uncertainties remain with respect to the interpretation of the term "de facto management body." If the PRC tax authorities determine that our Company or any of our offshore subsidiaries is a PRC resident enterprise for enterprise income tax purposes, our Company or the relevant offshore subsidiaries will be subject to PRC enterprise income tax on its worldwide income at the rate of 25%. Furthermore, if we are treated as a PRC tax resident enterprise, we will be required to withhold a 10% tax from dividends we pay to our shareholders that are

– 109 –

RISK FACTORS

non-resident enterprises. In addition, non-resident enterprise shareholders may be subject to PRC tax at a rate of 10% on gains realized on the sale or other disposition of Offer Shares, if such gain is treated as derived from a PRC source. Furthermore, if we are deemed a PRC resident enterprise, dividends paid to our non-PRC individual shareholders and any gain realized on the transfer of Offer Shares by such shareholders may be subject to PRC tax at a rate of 20% (which, in the case of dividends, may be withheld at source by us). These rates may be reduced by an applicable tax treaty, but it is unclear whether our non-PRC shareholders would, in practice, be able to obtain the benefits of any tax treaties between their country of tax residence and the PRC in the event that we are treated as a PRC resident enterprise. Any such tax may reduce the returns on your investment in the Offer Shares.

Indirect transfers of equity interests in PRC resident enterprises by their non-PRC resident companies may be subject to tax obligation.

On February 3, 2015, the SAT issued the Public Notice Regarding Certain Corporate Income Tax Matters on Indirect Transfer of Properties by Non-Tax Resident Enterprises (《关于非居民企业间接转让财产企业所得税若干问题的公告》), or SAT Bulletin 7, which came into effect on February 3, 2015 and last amended in December 2017. Pursuant to this Bulletin 7, an "indirect transfer" of assets, including equity interests in a PRC resident enterprise, by non-PRC resident enterprises may be re-characterized and treated as a direct transfer of PRC taxable assets, if such arrangement does not have a reasonable commercial purpose and was established for the purpose of avoiding payment of PRC enterprise income tax. As a result, gains derived from such indirect transfer may be subject to PRC enterprise income tax. According to Bulletin 7, "PRC taxable assets" include assets attributed to an establishment in China, immovable properties located in China, and equity investments in PRC resident enterprises, in respect of which gains from their transfer by a direct holder, being a non-PRC resident enterprise, would be subject to PRC enterprise income taxes. SAT Bulletin 7 has introduced safe harbors for the purchase and sale of equity through a public securities market. SAT Bulletin 7 also brings challenges to both foreign transferor and transferee (or other person who is obligated to pay for the transfer) of taxable assets.

On October 17, 2017, the SAT issued the Announcement of the State Administration of Taxation on Issues Concerning the Withholding of Non-resident Enterprise Income Tax at Source (《关于非居民企业所得税源泉扣缴有关问题的公告》), or SAT Bulletin 37, which came into effect on December 1, 2017. SAT Bulletin 37 further clarifies the practice and procedure of the withholding of non-resident enterprise income tax.

We may be subject to tax obligation as to the reporting and other implications of future transactions where PRC taxable assets are involved, such as offshore restructuring, sale of the shares in our offshore subsidiaries and investments. As a result, we may be required to expend valuable resources to comply with SAT Bulletin 7 and/or SAT Bulletin 37 or to request the relevant transferors from whom we purchase taxable assets to comply with these circulars, or to establish that our Company should not be taxed under these circulars, which may have a material adverse effect on our financial condition and results of operations.

– 110 –

风险因素 与加权投票权架构相关的风险 我们B类普通股的投票权集中限制了股东对公司事务的影响力。

本公司将于全球发售完成后通过加权投票权进行控制。全球发售完成后，加权投票权受益人将为阎博士及云女士。阎博士及云女士预期于本公司的经济权益约为28.19%，代表本公司股东大会约78.81%的总投票权（假设发售规模调整权及超额配股权均未获行使），就股东就保留事项以外事项所作决议而言。因此，阎博士及云女士对本公司的管理及事务，以及所有须经股东批准的事项（包括董事选举（独立非执行董事的委任、选举或罢免除外）及重大公司交易，如合并或出售本公司或本公司资产）于可预见的将来具有重大影响力。此外，由于每股A类普通股仅享有每股B类股份十分之一的投票权（适用法律规定及保留事项相关情况除外），A类普通股的发行（包括未来以股份为基础的收购交易及雇员股权激励计划）可能影响其决定提交股东投票的大多数事项结果的能力。有关本公司股权架构的进一步详情，请参阅「股本——加权投票权架构」。此种集中控制限制或严重限制本公司股东对公司事务的影响力，因此，本公司可能采取股东认为对其不利的行动。因此，本公司A类普通股的价格可能受到不利影响。此种集中控制可能阻碍其他方寻求A类普通股持有人认为有利的任何潜在合并、收购或其他控制权变更交易，亦可能阻止、推迟或妨碍本公司控制权的变更，其效果可能使本公司其他股东丧失于出售本公司时就其A类普通股获得溢价的机会，并可能降低本公司A类普通股的价格。

我们B类普通股的持有人可能对我们施加重大影响，且可能不会以我们其他股东的最佳利益行事。

全球发售完成后，本公司加权投票权受益人将处于对本公司事务施加重大影响的地位，并能够影响任何股东决议的结果，而不论其他股东如何投票。本公司B类普通股持有人的利益未必与全体股东的利益一致，此种投票权集中亦可能导致本公司控制权的变更被推迟、延缓或阻止。

— 111 —

RISK FACTORS RISKS RELATED TO THE GLOBAL OFFERING AND OUR SHARES

There has been no prior public market for our Class A Ordinary Shares and the liquidity and market price of our Class A Ordinary Shares may be volatile.

Prior to the Global Offering, there has been no public market for our Class A Ordinary Shares. There can be no guarantee that an active trading market for our Class A Ordinary Shares will develop or be sustained after the completion of the Global Offering. The Offer Price of our Class A Ordinary Shares is the result of negotiations between our Company and the Overall Coordinators (for themselves and on behalf of the Underwriters), which may not be indicative of the price at which our Class A Ordinary Shares will be traded following the completion of the Global Offering. The market price of our Class A Ordinary Shares may drop below the Offer Price at any time after completion of the Global Offering.

The price and trading volume of our Class A Ordinary Shares may be volatile, which could result in substantial losses for investors purchasing our Class A Ordinary Shares in the Global Offering.

Factors such as fluctuations in our revenue, earnings, cash flows, new investments, regulatory development, additions or departures of key personnel, or actions taken by competitors could cause the market price of our Class A Ordinary Shares or trading volume of our Class A Ordinary Shares to change substantially and unexpectedly. In addition, stock prices have been subject to significant volatility in recent years. Such volatility has not always been directly related to the performance of the specific companies whose shares are traded. Such volatility, as well as general economic conditions, may materially and adversely affect the prices of shares, and as a result investors in our Class A Ordinary Shares may incur substantial losses.

Subscribers and purchasers of our Class A Ordinary Shares under the Global Offering will experience immediate dilution and may experience further dilution if we issue additional Shares in the future.

The Offer Price of our Class A Ordinary Shares is higher than our net tangible assets value per Share immediately prior to the Global Offering. Therefore, subscribers and purchasers of our Class A Ordinary Shares under the Global Offering will experience an immediate dilution in pro forma net tangible assets value per Share. In order to expand our business, we may consider offering and issuing additional Shares in the future or to raise additional funds in the future to finance our business expansion, for existing operations or new acquisitions. If additional funds are raised through the issuance of new equity or equity-linked securities of our Company, other than on a pro rata basis to existing Shareholders, then (i) the percentage ownership of the existing Shareholders may be reduced, and they may experience subsequent dilution and reduction in their earnings per share, (ii) such newly issued securities may have rights, preferences or privileges superior to those of the Shares of the existing Shareholders

– 112 –

RISK FACTORS

and/or (iii) subscribers and purchasers of our Class A Ordinary Shares may experience dilution in the net tangible assets value per Share if we issue additional Shares in the future at a price which is lower than our net tangible assets value per Share.

Future sale or major divestment of Shares by any of our substantial Shareholders could adversely affect the prevailing market price of our Class A Ordinary Shares.

The Shares held by certain Shareholders are subject to certain lock-up periods, the details of which are set out in the section headed "Underwriting" of this Prospectus. However, we cannot give any assurance that after the restrictions of the lock-up periods expire, these Shareholders will not dispose of any Shares. Sale of substantial amounts of our Class A Ordinary Shares in the public market, or the perception that these sales may occur, may materially and adversely affect the prevailing market price of our Class A Ordinary Shares.

The market price of the Shares when trading begins could be lower than the Offer Price.

The Offer Price will be determined on the Price Determination Date. However, the Shares will not commence trading on the Stock Exchange until they are delivered, which is expected to be a few Business Days after the expected Price Determination Date. Investors may not be able to sell or otherwise deal in the Shares during that period. As a result, holders of the Shares are subject to the risk that the price of the Shares when trading begins could be lower than the Offer Price as a result of adverse market conditions or other adverse developments that may occur during that period.

There can be no assurance of the accuracy or completeness of certain facts, forecasts and other statistics obtained from various government publications contained in this prospectus.

Facts, forecasts, estimates and other statistics in this Prospectus relating to the economy and the industry in which we operate our business on have been collected from materials from official government sources. The information from official government sources has not been independently verified by us, the Joint Sponsors, Joint Global Coordinators, Joint Bookrunners, Joint Lead Managers, the Underwriters, any of their respective directors, supervisors, and advisors, or any other parties involved in the Global Offering, and no representation is given as to its accuracy. In particular, due to possibly flawed or ineffective collection methods or discrepancies between published information and market practice, such information and statistics may be inaccurate or may not be comparable to other information and statistics produced. Statistics, industry data and other information relating to the economy and the industry derived from the official government sources used in this Prospectus may not be consistent with other information available from other sources and therefore, investors should not unduly rely upon such facts, forecasts, estimates and statistics while making investment decisions.

– 113 –

RISK FACTORS

If securities or industry analysts do not publish research reports about our business, or if they adversely change their recommendations regarding our Class A Ordinary Shares, the market price and trading volume of our Class A Ordinary Shares may decline.

The trading market for our Class A Ordinary Shares will be influenced by the research and reports that industry or securities analysts publish about us or our business. If one or more of the analysts who cover us downgrade our Class A Ordinary Shares, the price of our Class A Ordinary Shares would likely decline. If one or more of these analysts cease coverage of our Company or fail to regularly publish reports on us, we could lose visibility in the financial markets, which in turn could cause our stock price or trading volume to decline.

We may not be able to pay any dividends to our Shareholders.

We cannot guarantee when and in what form dividends will be paid on our Class A Ordinary Shares following the Global Offering. The declaration of dividends is proposed by the Board and is based on, and limited by, various factors, such as our business and financial performance, capital and regulatory requirements and general business and operation conditions. We may not have sufficient or any profits to enable us to make dividend distributions to our Shareholders in the future, even if our financial statements indicate that our operations have been profitable.

Investors may experience difficulties in enforcing Shareholder rights.

Our Company is an exempted company incorporated in the Cayman Islands with limited liability, and the laws of the Cayman Islands differ in some respects from those of Hong Kong or other jurisdictions where investors may be located. The corporate affairs of our Company are governed by the Memorandum and the Articles, as amended from time to time, the Companies Act and the common law of the Cayman Islands. The rights of Shareholders to take legal action against our Company and/or our Directors, actions by minority Shareholders and the fiduciary duties of our Directors to our Company under Cayman Islands laws are to a large extent governed by the common law of the Cayman Islands. The common law of the Cayman Islands is derived in part from comparatively limited judicial precedent in the Cayman Islands as well as from English common law, which has persuasive, but not binding, authority on a court in the Cayman Islands. The rights of the Shareholders and the fiduciary duties of our Directors under Cayman Islands laws may not be as clearly established as they would be under statutes or judicial precedents in Hong Kong or other jurisdictions where investors reside. In particular, the Cayman Islands has a less developed body of securities laws. As a result of all of the above, Shareholders may have more difficulty in exercising their rights in the face of actions taken by the management of our Company, Directors or major Shareholders than they would as shareholders of a Hong Kong company or company incorporated in other jurisdictions.

– 114 –

RISK FACTORS You should read the entire Prospectus carefully and should not place any reliance on any information contained in press articles or other media regarding the Global Offering. There may have been, prior to the publication of this Prospectus, and there may be, subsequent to the date of this Prospectus but prior to the completion of the Global Offering, press and media coverage regarding us and the Global Offering, such as the profit estimate information. You should rely solely upon the information contained in this Prospectus and any formal announcements made by us in Hong Kong in making your investment decision regarding the Global Offering. We do not accept any responsibility for the accuracy or completeness of any information reported by the press or other media, nor the fairness or appropriateness of any estimates, views or opinions expressed by the press or other media regarding the Global Offering or us. We make no representation as to the appropriateness, accuracy, completeness or reliability of any such information or publication. Accordingly, prospective investors should not rely on any such information, reports or publications in making their decisions whether to invest in the Global Offering. Prospective investors in the Global Offering are reminded that, in making their decisions as to whether to purchase our Class A Ordinary Shares, they should rely only on the financial, operational and other information included in this Prospectus. By applying to purchase our Class A Ordinary Shares in the Global Offering, you will be deemed to have agreed that you will not rely on any information other than that contained in this Prospectus. Forward-looking information contained in this Prospectus is subject to risks and uncertainties. This Prospectus contains certain statements and information that are forward-looking and uses forward-looking terminology such as "anticipate," "believe," "could," "going forward," "intend," "plan," "project," "seek," "expect," "may," "ought to," "should," "would" or "will" and similar expressions. You are cautioned that reliance on any forward-looking statement involves risks and uncertainties and that any or all of those assumptions could prove to be inaccurate and as a result, the forward-looking statements based on those assumptions could also be incorrect. In light of these and other risks and uncertainties, the inclusion of forward-looking statements in this Prospectus should not be regarded as representations or warranties by us that our plans and objectives will be achieved and these forward-looking statements should be considered in light of various important factors, including those set forth in this section. Subject to the requirements of the Listing Rules, we do not intend publicly to update or otherwise revise the forward-looking statements in this Prospectus, whether as a result of new information, future events or otherwise. Accordingly, you should not place undue reliance on any forward-looking information. All forward-looking statements in this Prospectus are qualified by reference to this cautionary statement.

– 115 –

WAIVERS AND EXEMPTION In preparation of the Listing, the Company has sought the following waivers from strict compliance with the relevant provisions of the Listing Rules and exemption from the Companies (Winding Up and Miscellaneous Provisions) Ordinance: WAIVER IN RELATION TO MANAGEMENT PRESENCE IN HONG KONG According to Rule 8.12 of the Listing Rules, except as otherwise permitted by the Stock Exchange at its discretion, all applicants applying for a primary listing on the Stock Exchange must have sufficient management presence in Hong Kong. This would normally mean that at least two of an applicant's executive directors must be ordinarily resident in Hong Kong. Our headquarters are based, and substantially all of the business operations of our Group, are managed and conducted in the PRC. Our executive Directors ordinarily reside in the PRC and they play very important roles in our Company's business operations. It is in our best interests for them to be based in places where our Group has significant operations. We consider it practically difficult and commercially unreasonable for us to arrange for two executive Directors to ordinarily reside in Hong Kong, either by means of relocation of our existing executive Directors or appointment of additional executive Directors. Therefore, our Company does not have, or does not contemplate in the foreseeable future that we will have sufficient management presence in Hong Kong for the purpose of satisfying the requirements under Rule 8.12 of the Listing Rules. Accordingly, the Company has applied for, and the Stock Exchange has granted the Company, a waiver from strict compliance with the requirements under Rule 8.12 of the Listing Rules, provided that the Company will implement the following arrangements: (i)

We have appointed Ms. Yun Yeyi (貟燁禕) and Mr. Xue Zizhao (薛子釗) as our authorized representatives (the "Authorized Representatives") pursuant to Rule 3.05 of the Listing Rules. The Authorized Representatives will act as our Company's principal channel of communication with the Hong Kong Stock Exchange. The Authorized Representatives will be readily contactable by phone, facsimile and email to promptly deal with inquiries from the Hong Kong Stock Exchange, and will also be available to meet with the Hong Kong Stock Exchange to discuss any matter within a reasonable period of time upon request of the Hong Kong Stock Exchange;

(ii) 当香港联合交易所希望就任何事宜联系我们的董事时，每位授权代表将始终具备随时迅速联系我们所有董事（包括独立非执行董事）及高级管理层的一切必要手段。本公司亦将就授权代表的任何变动及时通知香港联合交易所。我们已向香港联合交易所提供所有董事的联系方式（即手机号码、办公室电话号码及电子邮件地址），以便与香港联合交易所保持沟通。如任何董事预计将出行或因其他原因不在办公室，该董事亦将向授权代表提供其所在住所的电话号码；

(iii) 所有不常居香港的董事均持有或可申请有效旅行证件前往香港，并可在合理时间内与香港联合交易所会面；

(iv) 根据《上市规则》第3A.19条及第8A.33条，我们已委任盛力资本有限公司(Somerley Capital Limited)为本公司上市后的合规顾问，任期自上市日期起计。合规顾问将可随时联系我们的授权代表、董事及高级管理层成员，当授权代表未能联系时，合规顾问将作为与香港联合交易所沟通的额外渠道。合规顾问的联系方式已提供予香港联合交易所，本公司将就合规顾问的任何变动及时通知香港联合交易所；及

(v) 本公司已于上市后在本公司总部指定员工担任通讯联络官，负责与授权代表及本公司在香港的专业顾问（包括本公司的香港法律顾问及合规顾问）保持日常沟通，以掌握来自香港联合交易所的任何往来函件及／或查询，并向执行董事汇报，以进一步促进香港联合交易所与本公司之间的沟通。

**有关联席公司秘书的豁免**

根据《上市规则》第3.28条及第8.17条，我们须委任一名公司秘书，该人须凭借其学术或专业资格或相关经验，在香港联合交易所看来能够胜任公司秘书职能。《上市规则》第3.28条附注1规定，香港联合交易所认为以下学术或专业资格可予接受：

(i) 香港特许治理公会会员；

(ii) 《法律执业者条例》（香港法例第159章）所界定的律师或大律师；及

(iii) 《专业会计师条例》（香港法例第50章）所界定的执业会计师。

– 117 –

**豁免及exemption**

《上市规则》第3.28条附注2规定，在评估「相关经验」时，联合交易所将考虑个人的：

(i) 在发行人及其他发行人的任职年期及其担任的职务；

（ii）熟悉《上市规则》及其他相关法律法规，包括《证券及期货条例》、《公司条例》、《公司（清盘及杂项条文）条例》及《收购守则》；

（iii）除《上市规则》第3.29条规定的最低要求外，已参加及／或将参加的相关培训；及

（iv）在其他司法管辖区的专业资格。

本公司已委任薛子釗先生（"薛先生"）为本公司联席公司秘书之一。薛先生在本公司资本市场事务方面具有充分经验，但目前不具备《上市规则》第3.28条及第8.17条规定的任何资格，且可能无法单独履行《上市规则》的相关规定。因此，本公司已委任陈秀玲女士（"陈女士"）为另一名联席公司秘书。陈女士为特许秘书及特许管治专业人士，同时为香港特许管治公会（HKCGI）（前称香港特许秘书公会）及英国特许管治公会（CGI）（前称特许秘书及行政人员公会）之资深会员，完全符合《上市规则》第3.28条及第8.17条所规定的条件。陈女士将自上市日期起的初始三年期间内，协助薛先生积累《上市规则》第3.28条附注2所指的"相关经验"，以使其完全符合《上市规则》第3.28条及第8.17条所载明的规定。

由于薛先生不具备《上市规则》第3.28条规定的公司秘书正式资格，本公司已就严格遵守《上市规则》第3.28条及第8.17条的规定向香港联合交易所申请豁免，香港联合交易所已批准该豁免，使薛先生得以获委任为本公司联席公司秘书。根据联交所发布的《新上市申请人指引》第13段的规定，该豁免将适用于固定期间（"豁免期"），并须符合以下条件：（i）拟委任的公司秘书须全程获得一名在整个豁免期内具备《上市规则》第3.28条所规定资格或经验并获委任为联席公司秘书的人士协助；及（ii）如发行人严重违反《上市规则》，豁免可予撤销。该豁免自上市日期起初始三年内有效，并以陈女士与薛先生紧密合作、共同履行公司秘书职责及义务，并协助薛先生积累《上市规则》第3.28条及第8.17条所要求的相关经验为前提条件。陈女士亦将协助薛先生组织本公司董事会会议及股东大会，以及处理本公司其他相关事项。

－118－

WAIVERS AND EXEMPTION

incidental to the duties of a company secretary. Ms. Chan is expected to work closely with Mr. Xue and will maintain regular contact with Mr. Xue, the Directors and the senior management of our Company. The waiver will be revoked immediately if Ms. Chan ceases to provide assistance to Mr. Xue as a joint company secretary for the three-year period after the Listing or where there are material breaches of the Listing Rules by our Company. In addition, Mr. Xue will comply with the annual professional training requirement under Rule 3.29 of the Listing Rules and will enhance his knowledge of the Listing Rules during the three-year period from the Listing. Mr. Xue will also be assisted by (a) the Compliance Adviser of our Company, particularly in relation to compliance with the Listing Rules; and (b) the Hong Kong legal advisers of our Company, on matters concerning our Company's ongoing compliance with the Listing Rules and the applicable laws and regulations.

Before the expiration of the initial three-year period, the qualifications of Mr. Xue will be re-evaluated to determine whether the requirements as stipulated in Rules 3.28 and 8.17 of the Listing Rules can be satisfied. We will liaise with the Hong Kong Stock Exchange to enable it to assess whether Mr. Xue, having benefited from the assistance of Ms. Chan for the preceding three years, will have acquired the skills necessary to carry out the duties of company secretary and the relevant experience within the meaning of Note 2 to Rule 3.28 of the Listing Rules so that a further waiver will not be necessary.

WAIVER IN RELATION TO CONTINUING CONNECTED TRANSACTIONS

Our Group has entered into certain transactions which would constitute partially-exempt and non-exempt continuing connected transactions under Chapter 14A of the Listing Rules after the Listing. Further particulars about such transactions together with the application for a waiver from strict compliance with the relevant requirements under Chapter 14A of the Listing Rules are set out in "Connected Transactions" in this prospectus.

WAIVER AND EXEMPTION IN RELATION TO THE PRE-IPO SHARE INCENTIVE PLAN

Rule 17.02(1)(b) of the Listing Rules stipulates that all material terms of a scheme adopted by a listing applicant prior to the listing must be clearly set out in the prospectus and ensure all relevant disclosures are adequately disclosed in the "Statutory and General Information" section and requires a listing applicant to, inter alia, disclose in this Prospectus full details of all outstanding options and awards and their potential dilution effect on the shareholdings upon listing as well as the impact on the earnings per share arising from the issue of shares in respect of such outstanding options.

Paragraph 27 of Appendix D1A to the Listing Rules requires a listing applicant to disclose, inter alia, particulars of any capital of any member of the group which is under option, or agreed conditionally or unconditionally to be put under option, including the consideration for which the option was or will be granted and the price and duration of the option, and the name and address of the grantee, or an appropriate negative statement, provided that where

– 119 –

(d)

the identities of all the Grantees who are Directors and connected persons of the Company are already disclosed in this Prospectus (in particular, in the section headed "Statutory and General Information — D. Share Incentive Plans" in Appendix IV to this Prospectus), and the remaining Grantees who are employees and Independent Third Party consultants are not otherwise required to be disclosed in the Prospectus; and

– 121 –

WAIVERS AND EXEMPTION

(e)

the grant of options under the Pre-IPO Share Incentive Plan has already been completed and no further grants will be made after the Listing, accordingly, the investing public will not be further affected by any subsequent undisclosed grants.

The Stock Exchange has granted a waiver from strict compliance with the requirements under Rule 17.02(1)(b) of and paragraph 27 of Appendix D1A to the Listing Rules. The SFC has granted a certificate of exemption from strict compliance with paragraph 10(d) of Part I of the Third Schedule to the Companies (Winding Up and Miscellaneous Provisions) Ordinance pursuant to section 342A of the Companies (Winding Up and Miscellaneous Provisions) Ordinance, subject to the condition that this Prospectus discloses the fact that the exemption has been granted and the reasons for the exemption.

豁免与豁除

鉴于上述情况，本公司董事认为，就本申请所寻求的豁免与豁除的批准以及对所需信息的不披露，不会妨碍潜在投资者对本集团的业务、资产及负债、财务状况、管理层及前景作出知情评估，亦不会损害公众投资者的利益。

联交所已就上市公司规则第17.02(1)(b)条及上市规则附录D1A第27段对本公司给予豁免，使本公司无需就上市前股份激励计划所授出期权严格遵守相关披露规定，但须符合以下条件：

(a) 证监会根据相关《公司（清盘及杂项条文）条例》要求发出豁除严格遵守相关规定的豁除证明书；

(b) 就本公司根据上市前股份激励计划授予本公司每名董事、高级管理层成员、关连人士、顾问及其他已获授代表200,000股A类普通股或以上期权的承授人，须在本招股章程中逐一披露所有期权的完整详情，包括上市规则第17.02(1)(b)条、上市规则附录D1A第27段及《公司（清盘及杂项条文）条例》第三附表第一部分第10段所要求的全部资料；

(c) 就本公司根据上市前股份激励计划授予上文第(b)段所述人士以外的其余承授人（「其他承授人」）的期权，须在本招股章程中以汇总方式披露以下详情：(i) 其他承授人的总人数及期权所涵盖的A类普通股数目；(ii) 授出期权所收取的代价；以及(iii) 期权的行使期及行使价；

(d) 于最后实际可行日期，已授出的尚未行使期权所涉及的股份总数以及该等股份数目占本公司已发行股本总额的百分比须在本招股章程中披露；

(e) 上市前股份激励计划的主要条款摘要，以及激励计划项下期权获全面行使后的摊薄效应及对每股收益的影响，须于本招股章程附录四「法定及一般资料 — D. 股份激励计划 — 1. 上市前股份激励计划」一节中披露；

(f) 本豁免与豁除的详情载于本招股章程；以及

— 122 —

豁免与豁除

(g) 已根据上市前股份激励计划获授认购股份期权的全体承授人的完整名单，包含上市规则第17.02(1)(b)条及上市规则附录D1A第27段以及《公司（清盘及杂项条文）条例》第三附表第一部分第10段所要求的全部详情，须按本招股章程附录五「送交香港公司注册处处长备案及可供查阅的文件」的规定供公众查阅。

证监会已根据《公司（清盘及杂项条文）条例》第342A条向本公司发出豁除证明书，豁除本公司无需严格遵守《公司（清盘及杂项条文）条例》第三附表第一部分第10(d)段，惟须符合以下条件：

(a) 就本公司根据上市前股份激励计划授予每名董事、高级管理层成员、关连人士、顾问及已获授认购200,000股A类普通股或以上尚未行使期权的承授人，须在本招股章程中披露所有尚未行使期权的完整详情，包括《公司（清盘及杂项条文）条例》第三附表第一部分第10段所要求的全部资料；

(b) 就本公司授予上文第(a)段所述人士以外的承授人的尚未行使期权，须在本招股章程中以汇总方式披露以下详情：

(i) 其他承授人的总人数及期权所涵盖的A类普通股数目；

(ii) 授出期权所收取的代价；以及

(iii) 期权的行使期及行使价；

(c) 已根据上市前股份激励计划获授购买A类普通股期权的全体承授人（包括上文第(a)及(b)段所述人士）的名单，包含《公司（清盘及杂项条文）条例》第三附表第一部分第10段所要求的全部详情，须按本招股章程附录五「送交香港公司注册处处长备案及可供查阅的文件 — 可供查阅的文件」的规定供公众查阅；以及

(d) 该豁除的详情须在本招股章程中披露，且本招股章程须于2025年12月31日或之前刊发。

— 123 —

WAIVERS AND EXEMPTION

关于Pre-IPO股份激励计划的进一步详情，请参阅本招股书附录四"法定及一般资料——D. 股份激励计划——1. Pre-IPO股份激励计划"一节。

根据《上市规则》第10.04条及《上市规则》附录F1第1C(2)段就现有股东及/或其紧密联系人作为基石投资者认购发售股份申请的豁免及同意

《上市规则》第10.04条规定，发行人的现有股东如拟以其本人名义或透过提名人认购或购买新申请人正在推销的、已申请上市的证券，须符合《上市规则》第10.03(1)及(2)条所载条件。

《上市规则》附录F1第1C(2)段规定（其中包括），除非符合《上市规则》第10.03条及第10.04条所订条件，否则不得向申请人的现有股东或其紧密联系人进行配售（无论以其本人名义或透过提名人），且须事先取得香港联合交易所的书面同意。

《指引》第2.5章第57段进一步规定，于IPO前持有专业技术公司少于10%股份的现有股东，可作为基石投资者或配售承配人认购IPO股份。若作为基石投资者认购，申请人及其保荐人须确认，除在IPO价格获保证配售的优惠待遇外，并无给予该现有股东任何优惠，且条款与其他基石投资者基本相同。

如本招股书"基石投资者"一节所进一步述及，(a) Alisoft China Holding Limited，本公司现有股东，于最后实际可行日期持有本公司13.66%的所有权及3.64%的投票权；(b) Aspex Master Fund，XEP-1 Holdings Limited的紧密联系人，XEP-1 Holdings Limited为本公司现有少数股东，于最后实际可行日期持有本公司1.06%的所有权及0.28%的投票权；(c) Abstract Enigma Limited，Nexus Vector Limited的紧密联系人，Nexus Vector Limited为本公司现有少数股东，于最后实际可行日期持有本公司0.71%的所有权及0.19%的投票权；(d) IDG Breyer Capital Fund L.P.，Lingham Beauty Limited及Forever Gain Limited的紧密联系人，Lingham Beauty Limited及Forever Gain Limited为本公司现有少数股东，于最后实际可行日期合共持有本公司1.89%的所有权及0.50%的投票权；(e) Janchor Partners Pan-Asian Master Fund及Janchor Partners Opportunities Master Fund III，本公司现有少数股东，于最后实际可行日期合共持有本公司0.85%的所有权及0.23%的投票权；及(f) MPC VII Pte. Ltd.，本公司现有少数股东，于最后实际可行日期持有本公司2.78%的所有权及0.74%的投票权（统称"现有股东基石投资者参与方"），已与本公司、

– 124 –

WAIVERS AND EXEMPTION

联席保荐人及整体协调人订立基石投资协议，据此，现有股东基石投资者参与方已同意作为基石投资者参与全球发售，认购本公司于国际发售项下发行的发售股份。

我们已就《上市规则》第10.04条申请豁免，并就《上市规则》附录F1第1C(2)段申请同意，以允许现有股东基石投资者参与方作为基石投资者参与全球发售，认购本公司于国际发售项下发行的发售股份。联交所已同意授予所申请的豁免及同意，惟须符合以下条件：

(a) 向现有股东基石投资者参与方的配售不会影响本公司分别满足《上市规则》第8.08(1)条、第18C.08条及第8.08A条相关规定的能力；

(b) 本公司及联席保荐人确认，在全球发售的任何配售中，并无直接或间接向现有股东基石投资者参与方作为基石投资者给予任何优惠待遇（基于其与本公司的关系），惟于发售价格下按基石投资获保证配售的优惠待遇除外，且条款与其他基石投资者基本相同；及

(c) 现有股东基石投资者参与方作为基石投资者于全球发售项下认购发售股份的详情已于本招股书中披露，而配售详情将于本公司的配发结果公告中披露。

有关相关基石投资的进一步资料，请参阅本招股书"基石投资者"一节。

– 125 –

INFORMATION ABOUT THIS PROSPECTUS AND THE GLOBAL OFFERING

DIRECTORS' RESPONSIBILITY FOR THE CONTENTS OF THIS PROSPECTUS

This Prospectus, for which our Directors (including any proposed director who is named as such in this Prospectus) collectively and individually accept full responsibility, includes particulars given in compliance with the Listing Rules, the Companies (Winding Up and Miscellaneous Provisions) Ordinance, the Securities and Futures (Stock Market Listing) Rules (Chapter 571V of the Laws of Hong Kong) and the Listing Rules for the purpose of giving information to the public with regard to our Group. Our Directors (including any proposed Director who is named as such in this Prospectus), having made all reasonable enquiries, confirm that, to the best of their knowledge and belief, the information contained in this Prospectus is accurate and complete in all material respects and not misleading or deceptive, and there are no other matters the omission of which would make any statement herein or this Prospectus misleading.

CSRC FILING

The Company has completed the PRC filing procedures with CSRC for the listing of our Class A Ordinary Shares on the Stock Exchange and the Global Offering.

INFORMATION ON THE GLOBAL OFFERING

This Prospectus is published solely in connection with the Hong Kong Public Offering. For applications under the Hong Kong Public Offering, this Prospectus contains the terms and conditions of the Hong Kong Public Offering. The Global Offering comprises the Hong Kong Public Offering of initially 1,269,480 Offer Shares (subject to reallocation on the basis as set out in "Structure of the Global Offering") and the International Offering of initially 24,119,740 Offer Shares (subject to reallocation and the Offer Size Adjustment Option on the basis as set out in "Structure of the Global Offering").

The Hong Kong Offer Shares are offered solely on the basis of the information contained and representations made in this Prospectus and on the terms and subject to the conditions set out herein. No person is authorized to give any information in connection with the Global Offering or to make any representation not contained in this Prospectus, and any information or representation not contained herein must not be relied upon as having been authorized by our Company, the Joint Sponsors, the Overall Coordinators, the Joint Global Coordinators, the Joint Bookrunners, the Joint Lead Managers, the Capital Market Intermediaries, the Underwriters, any of our or their affiliates or any of their respective directors, officers, employees, advisers, agents or representatives, or any other persons or parties involved in the Global Offering. Neither the delivery of this Prospectus nor any subscription or acquisition made under it shall, under any circumstances, create any implication that there has been no change in our affairs since the date of this Prospectus or that the information in this Prospectus is correct as of any subsequent time.

– 126 –

INFORMATION ABOUT THIS PROSPECTUS AND THE GLOBAL OFFERING

For details of the structure of the Global Offering, including its conditions and the arrangements relating to the Offer Size Adjustment Option, the Over-allotment Option and stabilization, see "Structure of the Global Offering."

PROCEDURES FOR APPLICATION FOR HONG KONG OFFER SHARES

The procedures for applying for the Hong Kong Offer Shares are set out in the section headed "How to Apply for Hong Kong Offer Shares."

RESTRICTIONS ON OFFER AND SALE OF THE SHARES

Each person acquiring the Hong Kong Offer Shares under the Hong Kong Public Offering will be required to, or be deemed by his/her acquisition of the Hong Kong Offer Shares to, confirm that he/she is aware of the restrictions on offers and sales of the Hong Kong Offer Shares described in this Prospectus.

No action has been taken to permit a public offering of the Offer Shares in any jurisdiction other than Hong Kong, or the distribution of this Prospectus in any jurisdiction other than Hong Kong. Accordingly, without limitation to the following, this Prospectus may not be used for the purpose of, and does not constitute, an offer or invitation in any jurisdiction or in any circumstances in which such an offer or invitation is not authorized or to any person to whom it is unlawful to make such an offer or invitation. The distribution of this Prospectus and the offering and sales of the Offer Shares in other jurisdictions are subject to restrictions and may not be made except as permitted under the applicable securities laws of such jurisdictions pursuant to registration with or authorization by the relevant securities regulatory authorities or an exemption therefrom. In particular, the Hong Kong Offer Shares have not been publicly offered or sold, directly or indirectly, in the PRC or the United States.

OFFER SIZE ADJUSTMENT OPTION, OVER-ALLOTMENT AND STABILIZATION

Details of the arrangement relating to the Offer Size Adjustment Option, the Over-allotment Option and stabilization are set out in the section headed "Structure of the Global Offering."

UNDERWRITING

The listing of our Class A Ordinary Shares on the Stock Exchange is sponsored by the Joint Sponsors and the Global Offering is managed by the Joint Global Coordinators and the Overall Coordinators. The Hong Kong Public Offering is fully underwritten by the Hong Kong Underwriters under the terms of the Hong Kong Underwriting Agreement and is subject to the Overall Coordinators (on behalf of the Underwriters) and us agreeing on the Offer Price on or before the Price Determination Date. An International Underwriting Agreement relating to the International Offering is expected to be entered into on or before Wednesday, January 7, 2026, subject to the Offer Price being agreed. The International Offering will be fully underwritten by the International Underwriters under the terms of the International Underwriting Agreement.

– 127 –

关于本招股说明书及全球发售的资料

待签订。如因任何原因，总协调人（代表其本身及承销商）与本公司未能于价格厘定日或之前就发售价达成协议，则全球发售将不会进行并将失效。有关承销商及承销安排的完整资料，请参阅"承销"部分。

申请于联交所上市

本公司已向联交所申请批准将以下股份上市及允许买卖：(a) 已发行的A类普通股（包括优先股转换所得的A类普通股及上市前发行的B类普通股转换所得的A类普通股）及根据全球发售将予发行的A类普通股（包括因行使发售规模调整权及超额配股权而可能发行的任何A类普通股），以及(b) 可根据上市后股份激励计划发行的A类普通股。本公司符合《上市规则》第18C.03条关于"未有商业收入公司"（定义见《上市规则》）的规定，以及《上市规则》第8A.06(1)条的规定，参考本公司于上市时的预期市值，该市值按指示性发售价区间下限计算超过400亿港元。

根据《公司（清盘及杂项条文）条例》第44B(1)条，如根据本招股说明书申请将股份于联交所上市的许可，在全球发售截止日期后三周届满前遭到拒绝，或在上述三周内由联交所或代表联交所通知本公司的不超过六周的较长期限届满前遭到拒绝，则根据本招股说明书的申请所作出的任何配发，无论何时作出，均属无效。

A类普通股开始买卖

A类普通股预期于2026年1月9日（星期五）上午9时正在联交所开始买卖。A类普通股将以每手20股A类普通股为单位进行买卖。A类普通股的股份代号为0100。

截至本招股说明书日期，本公司概无任何股份于其他任何证券交易所上市或买卖，亦无于联交所或任何其他证券交易所寻求或拟寻求上市或获准上市。所有A类普通股将登记于本公司的香港股东名册，以便在联交所进行买卖。

股份纳入中央结算系统

在联交所批准A类普通股上市及允许买卖，以及本公司符合香港结算的股份收纳规定的前提下，A类普通股将获香港结算接纳为合资格证券，可于上市日期起或由香港结算厘定的其他日期起在中央结算系统进行存管、结算及交收。交易所参与者之间的交易结算（如

— 128 —

INFORMATION ABOUT THIS PROSPECTUS AND THE GLOBAL OFFERING

defined in the Listing Rules) is required to take place in CCASS on the second settlement day after any trading day. All activities under CCASS are subject to the HKSCC Rules and HKSCC Operational Procedures in effect from time to time. Investors should seek the advice of their stockbroker or other professional adviser for details of the settlement arrangements as such arrangements may affect their rights and interests. All necessary arrangements have been made enabling the Shares to be admitted into CCASS.

HONG KONG REGISTER OF MEMBERS AND HONG KONG STAMP DUTY

Our Company's principal register of members will be maintained by our principal share registrar and transfer office, Maples Corporate Services Limited, in the Cayman Islands. All of the Shares issued pursuant to the Global Offering will be registered on our Company's Hong Kong Share Register to be maintained in Hong Kong by our Hong Kong Share Registrar, Tricor Investor Services Limited. Dealings in the Class A Ordinary Shares registered in our Company's Hong Kong Share Register will be subject to Hong Kong stamp duty. Unless determined otherwise by our Company, dividends payable in Hong Kong dollars in respect of the Class A Ordinary Shares will be paid to the Shareholders listed on the Hong Kong Share Register of our Company, by ordinary post, at the Shareholders' risk, to the registered address of each Shareholder.

PROFESSIONAL TAX ADVICE RECOMMENDED

Potential investors in the Global Offering are recommended to consult their professional advisers as to the taxation implications of subscribing for, purchasing, holding or disposing of, and/or dealing in the Class A Ordinary Shares or exercising rights attached to them. None of us, the Joint Sponsors, the Overall Coordinators, the Joint Global Coordinators, the Joint Bookrunners, the Joint Lead Managers, the Capital Market Intermediaries and the Underwriters, any of their respective directors, officers, employees, agents or representatives or any other person or party involved in the Global Offering accepts responsibility for any tax effects on, or liabilities of, any person resulting from the subscription, purchase, holding, disposition of, or dealing in, or the exercise of any rights in relation to, the Class A Ordinary Shares.

EXCHANGE RATE CONVERSION

Solely for your convenience, this Prospectus contains translations among certain Renminbi amounts into Hong Kong dollars and of Renminbi amounts into U.S. dollars at specified rates.

Unless indicated otherwise, the translation of Renminbi into Hong Kong dollars and of Renminbi into U.S. dollars, and vice versa, in this Prospectus was made at the following rates: HK$1.00 to RMB0.9068, US$1.00 to RMB7.0550, and US$1.00 to HK$7.7805. No representation is made that any amounts in Renminbi, Hong Kong dollars or U.S. dollars can be or could have been at the relevant dates converted at the above rates or any other rates or at all.

– 129 –

INFORMATION ABOUT THIS PROSPECTUS AND THE GLOBAL OFFERING

LANGUAGE

Translated English names of Chinese laws and regulations, governmental authorities, departments, entities (including subsidiaries of our Group), institutions, natural persons, facilities, certificates, titles and the like included in this Prospectus and for which no official English translation exists are unofficial translations for identification purposes only. In the event of any inconsistency, the Chinese name shall prevail.

ROUNDING

Certain amounts and percentage figures included in this Prospectus have been subject to rounding adjustments, or have been rounded to one or two decimal places. Accordingly, figures shown as totals in certain tables may not be an arithmetic aggregation of the figure preceding them. Any discrepancies in any table, chart or elsewhere between totals and sums of amounts listed therein are due to rounding.

– 130 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING

DIRECTORS

| Name | Address | Nationality | |------|---------|-------------| | **Executive Directors** | | | | Dr. Yan Junjie (閆俊傑) | Room 408, Building 3, Tianchangyuan, Chaoyang District, Beijing, PRC | Chinese | | Ms. Yun Yeyi (貟燁禕) | No. 51, Lane 1030, Zhongshan West Road, Changning District, Shanghai, PRC | Chinese | | Mr. Zhao Pengyu (趙鵬宇) | No. 501, 119th Floor, Block 4, Nanhu East Park 1, Wangjing, Chaoyang District, Beijing, PRC | Chinese | | Mr. Zhou Yucong (周彧聰) | No. 89, Lane 633, Wuzhong Road, Minhang District, Shanghai, PRC | Chinese | | **Non-executive Directors** | | | | Mr. Chen Yingjie (陳英傑) | No. 25 Gaoan Road, Xuhui District, Shanghai, PRC | Chinese | | Mr. Liu Wei (劉偉) | Building 1, Phase 4, Guangqiyuan, No. 519 Cangwu Road, Hongmei Road, Xuhui District, Shanghai, PRC | Chinese |

– 131 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING

| **Independent Non-executive Directors** | | | |------|---------|-------------| | Mr. Huang Guobin (黃國濱) | 17 MacDonnell Road, Central, Hong Kong | Chinese | | Dr. Wang Pengcheng (王鵬程) | Room 1005, Unit 1, Building 8, District 4, Yuandayuan, Century City, Haidian District, Beijing, PRC | Chinese | | Dr. Zhu Huaxing (朱華星) | 43-302, Zhong Guan Yuan, Haidian District, Beijing, PRC | Chinese |

For further information of our Directors, please see the section headed "Directors and Senior Management" in this Prospectus.

PARTIES INVOLVED IN THE GLOBAL OFFERING

Joint Sponsors

China International Capital Corporation Hong Kong Securities Limited 29/F, One International Finance Centre 1 Harbour View Street Central Hong Kong

UBS Securities Hong Kong Limited 52/F, Two International Finance Centre 8 Finance Street Central Hong Kong

(in alphabetical order)

– 132 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING

Overall Coordinators

China International Capital Corporation Hong Kong Securities Limited 29/F, One International Finance Centre 1 Harbour View Street Central Hong Kong UBS AG Hong Kong Branch 52/F, Two International Finance Centre 8 Finance Street Central Hong Kong (in alphabetical order) Goldman Sachs (Asia) L.L.C. 68/F, Cheung Kong Center 2 Queen's Road Central Central Hong Kong Morgan Stanley Asia Limited 46/F, International Commerce Centre 1 Austin Road West Kowloon, Hong Kong (in alphabetical order)

– 133 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING Joint Global Coordinators

China International Capital Corporation Hong Kong Securities Limited 29/F One International Finance Centre 1 Harbour View Street Central Hong Kong UBS AG Hong Kong Branch 52/F, Two International Finance Centre 8 Finance Street Central Hong Kong (in alphabetical order) Goldman Sachs (Asia) L.L.C. 68/F, Cheung Kong Center 2 Queen's Road Central Central Hong Kong Morgan Stanley Asia Limited 46/F, International Commerce Centre 1 Austin Road West Kowloon Hong Kong (in alphabetical order)

– 134 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING Joint Bookrunners

China International Capital Corporation Hong Kong Securities Limited 29/F One International Finance Centre 1 Harbour View Street Central Hong Kong UBS AG Hong Kong Branch 52/F, Two International Finance Centre 8 Finance Street Central Hong Kong (in alphabetical order) Goldman Sachs (Asia) L.L.C. 68/F, Cheung Kong Center 2 Queen's Road Central Central Hong Kong Morgan Stanley Asia Limited 46/F, International Commerce Centre 1 Austin Road West Kowloon Hong Kong (in alphabetical order) Futu Securities International (Hong Kong) Limited 34/F, United Centre No. 95 Queensway Admiralty Hong Kong Tiger Brokers (HK) Global Limited 23/F, Li Po Chun Chambers 189 Des Voeux Road Central Hong Kong (in alphabetical order)

– 135 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING Joint Lead Managers

China International Capital Corporation Hong Kong Securities Limited 29/F One International Finance Centre 1 Harbour View Street Central Hong Kong UBS AG Hong Kong Branch 52/F, Two International Finance Centre 8 Finance Street Central Hong Kong (in alphabetical order) Goldman Sachs (Asia) L.L.C. 68/F, Cheung Kong Center 2 Queen's Road Central Central Hong Kong Morgan Stanley Asia Limited 46/F, International Commerce Centre 1 Austin Road West Kowloon Hong Kong (in alphabetical order) Futu Securities International (Hong Kong) Limited 34/F, United Centre No. 95 Queensway Admiralty Hong Kong Tiger Brokers (HK) Global Limited 23/F, Li Po Chun Chambers 189 Des Voeux Road Central Hong Kong (in alphabetical order)

– 136 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING Capital Market Intermediaries

China International Capital Corporation Hong Kong Securities Limited 29/F One International Finance Centre 1 Harbour View Street Central Hong Kong UBS AG Hong Kong Branch 52/F, Two International Finance Centre 8 Finance Street Central Hong Kong (in alphabetical order) Goldman Sachs (Asia) L.L.C. 68/F, Cheung Kong Center 2 Queen's Road Central Central Hong Kong Morgan Stanley Asia Limited 46/F, International Commerce Centre 1 Austin Road West Kowloon Hong Kong (in alphabetical order) Futu Securities International (Hong Kong) Limited 34/F, United Centre No. 95 Queensway Admiralty Hong Kong Tiger Brokers (HK) Global Limited 23/F, Li Po Chun Chambers 189 Des Voeux Road Central Hong Kong (in alphabetical order)

– 137 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING Legal advisors to our Company

As to Hong Kong and United States laws: Davis Polk & Wardwell 10/F, The Hong Kong Club Building 3A Chater Road Central Hong Kong As to PRC laws (including as to matters concerning data compliance in the PRC): Jingtian & Gongcheng 34/F, Tower 3 China Central Place 77 Jianguo Road Chaoyang District Beijing PRC As to international sanctions laws: Hogan Lovells International LLP 11th Floor, One Pacific Place 88 Queensway Hong Kong As to United States laws (as to matters concerning data compliance in the United States): ZwillGen PLLC 1900 M Street NW, Suite 250 Washington, DC 20036 United States As to Singapore laws (including as to matters concerning data compliance in Singapore): Shook Lin & Bok LLP 1 Robinson Road #18-00 AIA Tower Singapore 048542

– 138 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING As to Cayman Islands laws: Maples and Calder (Hong Kong) LLP 26th Floor, Central Plaza 18 Harbour Road Wanchai, Hong Kong Legal advisors to the Joint Sponsors and the Underwriters

As to Hong Kong and United States laws: Freshfields 55th Floor, One Island East Taikoo Place, Quarry Bay Hong Kong As to PRC laws: Commerce & Finance Law Offices 12-15th Floor, China World Office 2 No. 1 Jianguomenwai Avenue Chaoyang District Beijing PRC

Auditor and Reporting Accountant

Ernst & Young Certified Public Accountants Registered Public Interest Entity Auditor 27/F, One Taikoo Place 979 King's Road Quarry Bay Hong Kong

Industry Consultant

China Insights Industry Consultancy Limited 10F, Block B, Jing'an International Center 88 Puji Road Jing'an District Shanghai PRC

– 139 –

DIRECTORS AND PARTIES INVOLVED IN THE GLOBAL OFFERING Compliance Adviser

Somerley Capital Limited 20/F, China Building 29 Queen's Road Central Hong Kong

Receiving Bank

Standard Chartered Bank (Hong Kong) Limited 18/F, Standard Chartered Tower 388 Kwun Tong Road Hong Kong

– 140 –

CORPORATE INFORMATION Registered Office

Maples Corporate Services Limited PO Box 309, Ugland House Grand Cayman, KY1-1104 Cayman Islands

Head Office and Principal Place of Business in the PRC

11th Floor, Building B Xinyan Mansion No. 65 Guiqing Road Xuhui District, Shanghai PRC

Principal Place of Business in Hong Kong

Room 1917, 19/F Lee Garden One 33 Hysan Avenue, Causeway Bay Hong Kong

Company's Websites

INDUSTRY OVERVIEW

The information and statistics set out in this section and other sections of this prospectus were extracted from different official government publications, available sources from public market research and other sources from independent suppliers, and from the independent industry report prepared by China Insights Industry Consultancy Limited ("CIC"). We engaged CIC to prepare an independent industry report in connection with the Global Offering (the "CIC Report"). The information from official government sources has not been independently verified by us, the Joint Sponsors, the Overall Coordinators, the Joint Global Coordinators, the Joint Bookrunners, the Joint Lead Managers, the Underwriters, any of their respective directors, supervisors, and advisors, or any other parties involved in the Global Offering, and no representation is given as to its accuracy.

OVERVIEW OF THE GLOBAL FOUNDATION MODEL INDUSTRY

Artificial intelligence (AI) is not only unlocking productivity, but also enriching creativity. Today, AI permeates various aspects of both people's professional and personal lives, from social media content recommendations, chatbots, and intelligent personal assistants, to autonomous driving systems, intelligent risk control models, and AI-assisted medical diagnostics. AI has emerged as a key enabler in the intelligent transformation of society and industries worldwide.

Foundation models in the past three years represent a significant technological paradigm shift compared to previous generations of AI — an inevitable trend fueled by societal developments. Traditional AI centered around small-scale models, which were custom-trained for application-specific scenarios. However, the goal is to enable intelligence that can perform the full range of human intellectual tasks. This requires AI to be more general-purpose, as user needs are becoming increasingly personalized and diverse. Foundation models are designed to address this challenge by offering scalability and generalization capabilities, and represent the most promising path towards achieving this goal.

Over the past three years, the field of foundation models has undergone rapid evolution. Benefitting from significant improvements in model scale and intelligence, the expansion of multi-modal capabilities, and the acceleration of commercialization, the industry has evolved at a remarkable pace.

SUSTAINED IMPROVEMENT IN MODEL INTELLIGENCE

Expansion of model scale and capabilities

Foundation models have scaled up dramatically in parameters in recent years, with significant performance improvements. OpenAI's GPT-3 launched in 2020 with 175 billion parameters, while GPT-4 launched in 2023 far outperformed it, scoring in the top 10% on a simulated bar exam versus GPT-3.5's bottom 10%, indicating near-human reasoning and comprehension capabilities.

– 144 –

INDUSTRY OVERVIEW Since then, more advanced models like GPT-4o, Gemini 2.5 Pro, and Claude 3.7 have pushed the boundaries of scale and intelligence further. A major breakthrough was the emergence of the Mixture-of-Experts (MoE) architecture, which uses expert sub-networks and a gating network to expand scale while keeping computational cost and latency low. In 2025, leading foundation model companies accelerated model updates from once every 4 months or longer in 2024 to no more than 3 months, driving continuous improvements in intelligence. For example, in 2024, Anthropic launched the Claude 3 series in March, followed by the Claude 3.5 generation — including the Sonnet and Haiku lineups released in June and October respectively — which together constituted a major upgrade to the Claude 3 family. However, in 2025, Anthropic launched Claude 3.7 Sonnet in February, followed by Claude 4 just three months later in May, and Opus 4.1 in August, nearly 50% faster than the 2024 update pace.

Improving context windows and reasoning efficiency

New models are not only more intelligent, but also capable of memorizing and processing more content. GPT-3 supported a context window of approximately 2,048 tokens, GPT-3.5 increased to 4,096, and GPT-4 extended to 32,768. Claude's 100,000-token context window enabled interaction with ultra-long documents.

However, longer context windows raised inference costs, prompting architectural innovations and retrieval-augmented generation. Among the most notable innovations are improvements to the attention mechanism — for example, MiniMax Text-01, a prior version of our text model, marked the first large-scale application of the linear attention mechanism architecture.

Alignment with humans

RLHF (reinforcement learning from human feedback) allows foundation models to be more receptive to user prompts. It has now become a standard procedure that enhances instruction adherence and response quality. OpenAI used RLHF to evolve GPT-3.5 into ChatGPT, which is capable of delivering coherent, tailored, and practical answers.

This approach has also inspired alternatives like Anthropic's "Constitutional AI," which guides model behavior through predefined principles instead of human feedback. Human-aligned models show marked improvements in accuracy, tone control, and handling of inappropriate queries.

Emergence of CoT (chain-of-thought) and reasoning models

The CoT (chain-of-thought) prompting technique, introduced in 2022, improved performance on complex reasoning tasks, such as mathematical problem-solving, and common-sense reasoning, by generating intermediate steps.

– 145 –

INDUSTRY OVERVIEW A major shift in 2024 saw models trained to break down problems step-by-step during inference, allocating more compute to iterative reasoning, reflection, and output refinement. Reasoning is increasingly regarded as a computable process rather than merely an emergent ability due to large model size. Reasoning models built on test-time compute (such as OpenAI o1 and DeepSeek R1) are explicitly trained with additional compute during the reasoning process to conduct iterative or structured reasoning during test-time. This trade-off between increased reasoning cost/latency and higher reasoning quality may lead to a divergence in future model development: one class of models will be optimized for fast, factually accurate responses, while the other will focus on deeper, more resource-intensive reasoning, with test-time compute being a key variable.

Agentic tool use as a new paradigm

A new paradigm is emerging with AI agents — models that autonomously plan and use external tools to accomplish more complex tasks. In 2023, GPT-4 introduced plug-ins and function calling to access external tools like browsers or Python for code execution, overcoming the limitations of models that could previously only operate within the bounds of their training data. Gemini further advanced this by running code autonomously within a sandbox environment. In 2025, a number of leading foundation model companies have been focused on enhancing their models' agentic capabilities.

These capabilities turn models into intelligent agents, expanding their roles beyond passive response towards active task orchestration. Other tool use includes generating structured outputs for other systems to read and integrating with knowledge retrieval. Agentic AI has not only greatly expanded the scope of AI applications, but also represents a critical step forward.

Parallel development of closed- and open-source models

Both closed- and open-source models are advancing in parallel over the past few years. OpenAI released closed-source models like GPT-4, while Meta drove open-source adoption with the introduction of the open-sourced LLaMA 2, lowering the barrier of fine-tuned model development. The academic community also work together with the industry to accelerate research progress. Projects like Stanford's Alpaca and LMSYS's Vicuna democratized the exploration of instruction adherence models. The academic community also contributed to foundational model architecture innovations while helping advance model evaluation and safety.

The open-source momentum is pushing closed-source developers to iterate faster, while giving users more customizable model options. Chinese companies are also launching competitive open-source models, including Alibaba's Qwen3, DeepSeek's V3 and R1, and MiniMax's M1 and M2.

– 146 –

INDUSTRY OVERVIEW Acceleration of progress

The intelligence level of foundation models worldwide continues to advance. According to OpenAI's five-level roadmap, current models have now reached the threshold of Level 3. Looking ahead, the trajectory points clearly towards accelerated progress.

| Levels | Name | Description | |--------|------|-------------| | L1 | Chatbots | AI with conversational language | | L2 | Reasoners | AI with human-level problem solving | | L3 | Agents | AI that can take actions | | L4 | Innovators | AI that can aid in invention | | L5 | Organizations | |

AI that can do the work of an organization

Source: OpenAI

模态的持续扩展 从单模态到多模态 基础模型已扩展至多模态领域，旨在将文本、图像、音频和视频的特征整合并对齐至共享语义空间，实现跨模态融合。

视觉理解 在多模态理解的早期阶段，CLIP、ViLBERT和VisualBERT等模型主要依赖双编码器架构来对齐视觉与文本输入。近期，发展趋势正转向更为统一的多模态能力。例如，GPT-4V将GPT-4框架扩展以支持图像输入，允许用户要求模型分析视觉内容、描述图像细节、解读表情包中的幽默信息及医学图像中的信息。Gemini基于纯解码器架构构建，支持图像、视频和音频模态，其中Gemini Ultra在多模态推理任务中树立了新的基准。

音频生成 文本与音频的融合使AI能够理解并自主生成音频内容。 OpenAI的Whisper于2022年发布，拥有16亿个参数，可对97种语言的音频进行转录和翻译，英语转录精度接近人类水平，使开发者能够将音频转换为文本以供进一步处理。

– 147 –

行业概述 音频合成技术同样取得了快速进展。2023年，ElevenLabs和MiniMax等服务提供商使模型能够以接近人类的声音进行语音合成。同年，OpenAI为ChatGPT新增了基于音频的对话功能，支持实时语音输入与合成语音输出，拓展了音频模型在智能助手和客户服务领域的应用。

文本与音频的融合催生了语音驱动AI智能体和智能设备等新产品。未来的模型将更好地理解语音中的情感与意图，生成更加自然的回应，从而改善人机交互体验。

视觉生成 至2022年，DALL-E、Imagen、Stable Diffusion和Midjourney等文本生成图像模型已能产出媲美真实照片和艺术作品的输出结果。这些模型通常基于扩散模型，将语言模型与生成模型相结合，依托大规模图文数据集和基于Transformer的文本编码技术。DALL-E 3支持自然语言交互，并可编辑图像内的文字。Stable Diffusion以照片写实性、可定制性和开源生态著称。Midjourney则在艺术风格化和易用性方面表现突出。

自2023年起，视频生成作为新的多模态领域开始崭露头角。OpenAI的Sora是一款基于DiT架构的视频模型，可根据文本、图像甚至视频形式的输入生成全新视频内容。海螺AI和Google Veo 3等产品也在全球范围内获得广泛关注。这些工具使创意内容生成变得更加普及，并提升了创意行业的工作流效率。

学术界已开始探索能够同时进行多模态理解与生成的统一模型。这些模型旨在于单一、完整的架构内处理多种输入模态，并跨越其中一种或多种模态生成输出。此类统一系统需要将自回归模型在推理和文本生成方面的优势，与扩散模型在高保真图像生成方面的优势相结合。这种融合的追求映射出人类智能更深层的本质——人类的理解与表达，以及跨越不同模态的输入与输出，是深度交织、不可分割的，而非模块化且相互独立的。

基础模型应用的广泛普及释放商业价值 基础模型应用的空前增长 过去三年间，新一代AI经历了超高速增长，其速度超越了人类历史上所有以往的技术浪潮，包括互联网革命和工业革命。

– 148 –

行业概述 ChatGPT成为史上用户增长最快的产品，仅用17个月便达到8亿用户，且实现了全球覆盖，超过90%的用户来自北美以外地区。从商业角度而言，新一代AI原生产品在单一年度内所达到的营收水平，相当于SaaS行业用十年才实现的成就。借助前几波技术浪潮所积累的基础设施和发展动能，AI技术正以闪电般的速度蔓延至互联网的每个角落。

人类目前正处于指数级技术增长轨迹的关键拐点。在当下，进步对人们而言似乎是线性的，但回望过去，指数级的飞跃往往在极短的时间内便已发生。

泛化能力驱动基础模型应用的巨大市场空间 目前，基础模型的主要应用领域包括生产力工具、娱乐、视觉生成、音频生成以及通用企业服务（2B服务）。在这些细分领域中，基础模型的泛化能力通过单一、高度可扩展的架构实现了大规模部署与个性化应用的有机统一——服务于从大型企业到个人创作者的广泛用户群体，并实现了正向的模型投资回报率（ROI）。

能够实现长期持续有机增长的产品，是那些由底层模型智能的持续提升所驱动的产品。意在维持长期市场领导地位的参与者，必须开发能够端到端优化、并持续保持顶尖性能的自有基础模型。基础模型智能的突破性进展能够迅速将其推至行业前列，因为用户自然会被能提供更优体验的技术所吸引。

基础模型应用的市场格局

| 领域 | 生产力 | 娱乐 | 视觉生成 | 音频生成 | | | |------|--------|------|----------|----------|---|---| | | 通用 | | 视觉 | 音频 | 垂直 | | | 2C | 豆包（Doubao） | | | | | | | 应用层 | | | | | | | | 2B | | | | | | |

INDUSTRY OVERVIEW

Productivity The productivity segment represents a massive opportunity with broad downstream use cases. Users often engage with multiple models simultaneously. Top use cases include information search, writing, coding, education, office administration, academic research, and business analysis, which cover all aspects of work and daily life.

The productivity segment has undergone a generational shift from chatbots in 2023 to agents in 2025. Leading chatbot players include ChatGPT and DeepSeek R1, with OpenAI Deep Research and MiniMax Agent driving the next generation of AI agents. Unlike chatbots that simply respond to prompts, agents can complete long-horizon tasks due to advances in multi-step reasoning and use of external tools, enabling them to learn and improve through interactions with their environment. An emerging application of agents is virtual co-workers that can be integrated into enterprise workflows. Companies such as OpenAI and Anthropic are training foundation models in reinforcement learning environments to operate professional business software. The ultimate goal is for these agents to independently handle complex tasks and deliver tangible business value. Long-term leaders in this field must possess end-to-end capabilities, the ability of models to enhance their capabilities via end-to-end reinforcement learning using proprietary models and rewards from application-specific environments.

Since the second half of 2024, AI coding applications experienced exponential growth, fueled by the breakthrough of Claude 3.5 Sonnet. Its capabilities in code design, debugging, and optimization have powered over 30 million developers worldwide. Notable products include Claude Code, Anthropic's agentic coding tool, Cursor, a code editor for professional developers, and Windsurf, an enterprise-level secure coding platform. Beyond the professional coding market, 2025 has seen a surge of "vibe coding" tools designed for everyday users with no programming background. Platforms such as Lovable and Bolt.new enable anyone to create applications simply through natural language input. The overall trend is shifting from simple code completion towards more advanced coding agent capabilities, with a long-term potential for enabling personalized software generation from a single chat interface. This would not only lower barriers for professional product development, but also unlock a new market for users with little experience.

Entertainment Entertainment is the second-largest segment following productivity, with tens of millions of young users worldwide creating and interacting with personalized AI agents. The use cases are highly diverse, spanning role-play, companionship, and a wide range of everyday Q&A interactions.

Competition in the entertainment segment is in the process of stabilizing. Leading products include Character AI and Talkie/Xingye, which can enhance user experience with model optimization and inspire users' creativity with rich multi-modal creative tools. This combination drives high engagement, user stickiness, and highly interactive experiences.

INDUSTRY OVERVIEW

The new generation of AI-native users are naturally inclined to interact with AI companions. As societal productivity continues to rise and material needs are increasingly met, entertainment AI products will tap into users' emotional and psychological needs and unlock long-term market potential through personalized, emotionally resonant experiences.

Visual generation

Image generation has emerged as the first AI domain to achieve commercialization. In 2022, models like Midjourney impressed the world with their visually striking outputs, sparking exponential growth in social media engagement as image quality progressively met commercial standards. The primary user base consists of professional creators and enthusiasts in graphic design, film/TV production, advertising and e-commerce. These tools not only inspire creativity but also significantly enhance design workflow efficiency.

Leading applications in this space, including GPT-4o, Midjourney and Flux, continue to make breakthroughs in image quality, editing customizability, and diversity of styles. In 2025, Google's Nano Banana advanced image generation to a new level, enabling precise natural language editing and powering commercial-grade uses from marketing visuals to game design.

These advancements have unlocked greater end-user application scenarios, such as professional-standard product design and commercial marketing materials. AI-generated images have already achieved widespread popularity, marking their evolution from being just creative tools to becoming mainstream content.

Video generation has emerged as a rapidly growing segment in 2024, with a clear product-market fit. Demand comes from a wide range of industries, including film and television, short videos, mini-dramas, advertising, and e-commerce, leading to a massive market opportunity. In these industries, conventional video production often requires an entire team, whereas foundation models open up new market opportunities for individual professional creators to act as "one-person studios" to produce high-value content as well as enhancing their productivity.

Leading players in this segment include Sora, Veo, Hailuo AI, and Kling, among others. Their core competitiveness lies in maintaining our model R&D capabilities and cost efficiencies, coupled with fast-iterating creative workflow features and a vibrant creator ecosystem.

AI-generated videos are beginning to go viral increasingly frequently on social media, signaling that model performance is beginning to break through the boundaries of consumer-level content. Sora 2, launched in October 2025, sparked viral sharing on social media, signaling a major shift in the content industry with opportunities on the scale of the next short-video boom. In the future, relevant products may evolve into a "real-time personalized video generation engine", lowering barriers for anyone to create and consume personalized content.

– 151 –

INDUSTRY OVERVIEW

Audio generation

Audio is the universal interface of interaction in the AI era, with a broad downstream application market. For enterprises, AI voice agents overcome the limit of human capacity in sales and customer service, including recruitment, finance, healthcare; for content creators, it enables lifelike and emotionally expressive audio generation for audiobooks, education, dubbing and gaming, and others.

Leading players include OpenAI, MiniMax, and ElevenLabs. Their core competitiveness lies in delivering hyper-realistic audio model quality while maintaining low cost and low latency.

Numerous agentic AI applications and smart devices are empowered by audio in the AI era. OpenAI's GPT-4o introduced real-time audio interaction in May 2024, setting a new standard for chatbots; Google's NotebookLM saw viral success in September 2024 with its podcast generation feature. As human—AI interactions grow exponentially, the audio submarket holds vast untapped potential.

General 2B services

To accelerate AI adoption in various fields, foundation model companies such as OpenAI and Anthropic typically offer model capabilities to developers and enterprise clients via APIs with an open-platform strategy. Cloud service providers such as Microsoft, Amazon, Google, and Alibaba also provide models, toolkits and professional services through APIs, industry-tailored solutions, and on-premise deployment.

The core competitiveness in this segment includes model performance, cost-efficiency, and stability during high concurrencies, which are the top concerns for developers and enterprise customers. Secondary considerations include security, compliance, and customer support. A multiple-model strategy is now common, with enterprises often using three or more models and routing different models to specific tasks based on use-case requirements.

Enterprise demand is surging across industries. As agentic models become increasingly capable of delivering satisfying outcomes and inference costs continue to drop rapidly, foundation models are set to become a new productivity norm, continuously unlocking value across sectors.

SCALE OF THE GLOBAL FOUNDATION MODEL MARKET

Market size

The global foundation model market comprises revenue generated by model-based and deployment-based approaches. Model-based revenue is primarily generated from (i) a wide range of end-user applications such as AI chatbots, social and entertainment AI products, video generation and audio generation products, that are offered to both consumers and enterprises

– 152 –

INDUSTRY OVERVIEW

主要通过订阅方式获取收入，以及(ii) MaaS（模型即服务），即通过基于云端的API和许可协议提供基础模型能力，使开发者和企业能够按使用量访问并将模型功能集成到其自有产品或系统中。基于部署的收入来源于在本地部署定制化解决方案。

基础模型技术仍处于快速发展阶段。与基于部署的方式相比，基于模型的方式使用户能够持续受益于模型的迭代改进，而无需承担版本迁移成本。用户还可以根据实际需求动态调整模型使用规模，从而减少与硬件和基础设施相关的前期投入及持续维护费用。此外，该方式支持自动资源扩展，以满足用户不断变化的需求。

据灼识咨询（CIC）数据，全球基于模型的基础模型市场仍处于商业化早期阶段。随着技术持续成熟以及用户付费意愿的稳步提升，全球基于模型的基础模型市场预计将从2024年的107亿美元快速增长至2029年的2,065亿美元，复合年增长率（CAGR）为80.7%。受基础模型技术持续进步与成熟的推动，基础模型应用的市场规模预计将从2024年的71亿美元扩展至2029年的1,515亿美元，复合年增长率为84.3%；基础模型MaaS的市场规模预计将从2024年的36亿美元增长至2029年的550亿美元，复合年增长率为72.7%。

全球基础模型市场规模（按基于模型的收入口径） 2023–2029年预测 单位：十亿美元

| 类别 | 2023–2024年CAGR | 2024–2029年CAGR | |------|----------------|----------------| | 应用（Application） | 169.3% | 84.3% | | MaaS | 171.8% | 72.7% | | 合计（Total） | 170.1% | 80.7% |

| 年份 | 应用 | MaaS | 合计 | |------|------|------|------| | 2023 | 1.3 | 2.6 | 4.0 | | 2024 | 7.1 | 3.6 | 10.7 | | 2025E | 14.6 | 7.4 | 22.0 | | 2026E | 29.0 | 12.4 | — | | 2027E | 52.9 | 20.7 | — | | 2028E | 91.2 | 34.2 | 125.4 | | 2029E | 151.5 | 55.0 | 206.5 |

来源：灼识咨询（CIC） 注：基于模型的收入主要包括基础模型应用订阅产生的收入，以及基础模型API调用和许可产生的收入。

– 153 –

INDUSTRY OVERVIEW

市场驱动因素

技术飞跃

基础模型市场以颠覆性技术突破为显著特征，每一代新基础模型的性能提升均不断拓展潜在应用场景的边界。

GPT-3使娱乐聊天率先实现产品与市场的契合（Product-Market Fit）；GPT-3.5将聊天机器人应用提升至高度可用的水平，而GPT-4则深入金融、法律等专业领域。Sora推动视频生成达到商业化所需的质量标准。多模态模型GPT-4o促使ChatGPT迎来新一轮用户增长浪潮。Claude 3.5 Sonnet增强的代码能力，助推了Claude Code、Cursor和Windsurf等开发者工具实现产品与市场的契合。OpenAI的o1凭借增强的推理能力，以及Claude 3.7 Sonnet凭借更强的工具调用能力，正推动AI智能体（AI agents）的崛起。

技术是基础模型浪潮最强有力的底层驱动力，基础模型企业处于这一变革性增长的前沿阵地。每一代新模型都会催生新的应用场景，这些场景从零散的探索性实验逐步演变为主流应用。真正的突破通常发生于两代模型之间，每一次技术跃迁都开辟出新的能力曲线，并催生出全新类别的产品与服务。例如，MiniMax（海螺AI）最新文本模型MiniMax-M2引入了"交错思考"（interleaved thinking）这一新颖的非共识框架，使智能体推理更加稳健可靠，并已被证明具有高度有效性。该模型发布后第一周内，即跻身OpenRouter（全球使用最广泛的平台之一，允许开发者通过统一API便捷访问多个基础模型）全球日均Token使用量前三名，同时成为首个在OpenRouter上日均Token使用量突破500亿的中国模型。

对于基础模型企业而言，这些技术拐点直接转化为新产品和解决方案需求的扩张。通过探索模型演进方向，企业能够将产品路线图与新兴市场需求相匹配，加速客户采纳，并在每个技术周期的最早阶段把握增长机遇。

规模化定律（Scaling Law）

基础模型市场增长的根本驱动力在于，基础模型技术能够持续进行规模扩展。

预训练规模化定律（Pre-training Scaling Law）已广为人知——模型性能随模型规模、数据量和算力的增加而同步提升。这一定律在从GPT-3.5、GPT-4到近期发布的Claude 4等模型中仍然有效成立，其参数量从数千亿扩展至数万亿。此外，该原则不仅适用于文本模型，同样适用于视频、音频等其他模态，近期视频和音频生成技术的突破同样得益于模型规模和训练数据量的双重扩展。

– 154 –

INDUSTRY OVERVIEW

此外，自OpenAI的o1开始，行业见证了一种以测试时算力为核心的新扩展定律。模型的推理能力随着推理阶段算力的延伸而增强——模型用于思考的时间越长，其表现越佳。长思考过程所需的算力可能是单次推理的100倍，使模型能够解决传统模型几乎无法处理的极复杂任务。这一原理在OpenAI o3等后续模型中得到持续验证，截至2025年11月，Artificial Analysis智能指数排名前十的最佳模型均为推理模型，其中包括MiniMax-M2。

展望未来，基础模型的规模扩展预计将持续推进，预训练与推理的扩展相互强化。这一动态支撑了新的"摩尔定律"，通过集体扩展进步而非孤立创新惠及整个行业。它使基础模型公司能够以更高吞吐量、更低延迟部署日益复杂的模型，并把握行业扩展轨迹所带来的增长机遇。

成本下降

模型成本的持续下降是比模型能力提升更具可预测性的市场驱动因素，二者共同作用，预计将解锁越来越多跨越投资回报率门槛、实现产品市场契合的应用场景。

GPT-4发布之时，内容审核等许多垂直应用已满足性能要求，但由于成本高昂、投资回报率为负，在商业上仍不可行。基础模型的推理成本持续稳步下降，GPT-4自发布以来每令牌成本已下降逾99%，从而推动其在高频量、后端行业场景中的广泛普及。这一下降趋势在行业中普遍存在，由架构创新、推理效率提升、工程优化以及算力成本降低等多重因素共同驱动，预计将以可预测的速率持续压低成本。

推理成本的下降拓宽了经济可行应用的范围，降低了采用门槛，开启了新的市场机遇。这一趋势推动推理量提升及基础模型产品与解决方案的更广泛部署，使客户能够盈利性地扩大使用规模，并加速整体市场增长。

基础模型应用趋势

基础模型应用的商业化仍处于早期阶段，与智能体的接近程度是有望释放重大商业价值的拐点。

– 155 –

INDUSTRY OVERVIEW

智能体应用：智能体能够以专业水平运作、自主行动并交付端到端成果，最终推动数万亿美元级别的GDP增长

实现专业级表现涉及在专家领域内执行专项任务。自主行动使系统能够独立于人类的时间与精力而运转。交付端到端成果标志着创造经济价值的能力。这些特征标志着大语言模型从提供工具转向交付成果的拐点。由此，大语言模型的可寻址市场将突破企业软件预算的边界，进入更广阔的劳动力服务市场空间。

此外，智能体正持续增强其完成任务的能力。智能体同样遵循其自身的扩展定律——智能体能够自主处理的任务时长大约每七个月翻倍。如今，AI已能自主完成通常需要人类一小时完成的任务。预计到2026年，AI将能自主处理2至3小时的任务，并在五年内处理长达一个月的任务。这为"无限专家"的未来铺平了道路——AI软件工程师、金融分析师和研究科学家全天候无间断地为更高生产力和智能体经济做出贡献。

鉴于智能体扩展定律，未来的智能体不仅将执行任务，还将作为AI研究员，能够加速新专有算法的研究，并开发超越自身能力的新一代智能体。这种指数级、自我强化的演进正是这一代AI技术在核心层面有别于以往的关键所在。算法扩展与应用扩展之间将涌现出复利效应，二者相互强化、相互加速。这一正反馈循环有可能使AI成为人类历史上演进最快的技术革命。

娱乐与生成式应用：多个垂直领域的快速增长

新一代AI原生用户追求沉浸式、共创体验，推动娱乐产品向个性化AI陪伴方向演进。随着模型智能与记忆能力的持续提升，我们预见未来将出现"Her"式时刻——每个人都拥有一位真正了解自己、并在生活各方面主动提供协助的AI伴侣。这些AI伴侣将兼具智识与情感智能，通过富有表现力和共情力的互动与用户建立深厚的情感纽带。个性化AI作为一种趋势正在兴起，AI伴侣能够从日常互动中学习用户的个性化偏好、习惯和沟通风格，跨设备为其提供全方位协助。

视频生成技术的进步正在改变创意的边界，AI生成的视频有望在社交媒体上广泛传播。这标志着内容生产从专业工具向大众可及的创意引擎转变，显著改变了视频内容的供给格局。与此同时，音频生成与交互能力正在

– 156 –

INDUSTRY OVERVIEW

逐渐成为人工智能应用的标准配置，从基本的指令-响应功能向更具情感表达力的沟通方式演进。这一演变使人机交互更加自然，语音界面正成为多模态交互的核心枢纽。

多模态应用：统一多模态开启全新市场潜力

2025年3月，GPT-4o更新了图像生成能力，大幅提升了图像质量，引发ChatGPT新订阅用户激增，充分展示了多模态融合的商业潜力。与此前依赖DALL-E等独立模型进行图像生成的方式不同，GPT-4o基于原生多模态架构，可直接从文本提示生成图像，实现了全新水平的可控图像生成与编辑。

精确生成图像内文本以及对图像进行精细化编辑的能力，开辟了多种商业应用场景，例如生成教育可视化内容、带有风格化排版的产品海报以及科学插图。展望未来，文本、音频与视觉模态的深度融合将使完全可编辑视频的创作成为可能，并能够随视频内容同步生成音频与文字等，为下一场短视频革命开启市场机遇。

全球基础模型市场竞争格局

竞争排名

基础模型企业大致分为两类：基础模型技术公司和基础模型应用公司。前者指具备自主研发基础模型技术能力的企业，后者指在现有基础模型之上构建行业或场景专属应用及解决方案、自身不从事基础模型研发或维护的企业。

当前，基础模型行业仍处于底层技术驱动发展的阶段，基础模型的重大迭代能够显著拓展模型能力的边界。因此，更专注于推进底层技术的基础模型技术公司是创新的主要推动者，在塑造行业未来方面发挥着引领作用。

由于终端用户体验在很大程度上取决于基础模型的性能，当今市场上大量领先产品均由兼具端到端模型与应用开发能力的基础模型技术公司所开发。

根据CIC的数据，以2024年基于模型的营业收入计，MiniMax（名之梦）在全球基础模型技术公司中排名第十，市场份额为0.3%，详见下表。全球基础模型市场规模预计将于2025年达到220亿美元（US$22.0 billion），MiniMax预计将占据约0.3%的市场份额。鉴于多数同行均为资源雄厚的大型上市公司，一家资源相对有限的初创企业能够跻身全球前十、与行业巨头同台竞技，实属难能可贵。

2024年全球基础模型技术公司排名 （按基于模型的营业收入）

| 排名 | 公司 | 市场份额（%） | |------|------|--------------| | 1 | A公司 | 30.1% | | 2 | B公司 | 16.9% | | 3 | C公司 | 8.2% | | 4 | D公司 | 4.7% | | 5 | E公司 | 2.8% | | 6 | F公司 | 1.8% | | 7 | G公司 | 0.7% | | 8 | H公司 | 0.5% | | 9 | I公司 | 0.3% | | 10 | MiniMax | 0.3% | | 11 | J公司 | 0.3% | | 12 | K公司 | 0.3% | | 13 | L公司 | 0.3% | | 14 | M公司 | 0.2% | | 15 | N公司 | 0.2% |

来源：CIC

注： (1) 基于模型的营业收入主要包括基础模型应用订阅、基础模型API调用及授权许可所产生的收入。

(2) A公司为2015年在美国成立的基础模型公司，主要提供聊天机器人及视频生成应用等AI原生产品，为非上市公司。

(3) B公司为1998年在美国成立的科技公司，主要提供互联网相关产品与服务，包括搜索引擎、云计算、数字广告及AI产品与服务，在纳斯达克证券交易所上市。

(4) C公司为1975年在美国成立的科技公司，主要提供办公软件、云服务及AI产品与服务，在纳斯达克证券交易所上市。

– 158 –

INDUSTRY OVERVIEW

(5) D公司为2021年在美国成立的基础模型公司，主要提供大语言模型产品，为非上市公司。

(6) E公司为2021年在美国成立的基础模型公司，主要提供AI图像生成应用，为非上市公司。

(7) F公司为1994年在美国成立的科技公司，主要提供电子商务平台、云计算服务、数字流媒体及AI产品与服务，在纳斯达克证券交易所上市。

(8) G公司为2004年在美国成立的科技公司，主要提供社交网络平台及开源基础模型，在纳斯达克证券交易所上市。

(9) H公司为1999年在中国成立的科技公司，主要提供电子商务平台、云计算服务、数字支付服务及AI产品与服务，在联交所及纽约证券交易所双重上市。

(10) I公司为2006年在美国成立的社交媒体公司，主要提供全球社交网络平台及大语言模型应用，为非上市公司。

(11)

Company J是一家于2022年在美国成立的人工智能公司。其主要提供AI驱动的语音合成及配音服务，包括多语言语音生成和声音克隆。该公司为非上市公司。

(12)

Company K是一家于2000年在中国成立的科技公司。其主要提供搜索引擎、云服务及AI产品和服务。该公司同时在联交所及纳斯达克证券交易所双重上市。

(13)

Company L是一家于2018年在美国成立的人工智能公司。其主要提供AI驱动的视频及图像生成工具。该公司为非上市公司。

(14)

Company M是一家于2017年在英国成立的人工智能公司。其主要提供AI驱动的视频创作工具。该公司为非上市公司。

(15)

Company N是一家于1999年在中国成立的语音技术及人工智能公司。其主要提供语音识别软件及其他基于语音的AI产品。该公司为在深圳证券交易所上市的上市公司。

下表列示全球主要基础模型技术公司的产品服务比较。

全球主要基础模型技术公司的产品服务

| 公司 | 主要产品类型 | 主要变现方式 | |---|---|---| | Company A | 生产力、视觉生成 | 订阅、API调用 | | Company B | 生产力、通用2B服务 | 订阅、API调用 | | Company C | 生产力、通用2B服务 | 订阅、API调用 | | Company D | 生产力 | 订阅 | | Company E | 视觉生成 | 订阅 | | Company F | 通用2B服务 | API调用 | | Company G | 通用2B服务 | API调用 | | Company H | 通用2B服务 | 订阅、API调用 | | Company I | 生产力、通用2B服务 | 订阅、API调用 |

– 159 –

行业概览

| 公司 | 主要产品类型 | 主要变现方式 | |---|---|---| | MiniMax | 娱乐、视觉生成、生产力 | 订阅、在线营销服务、应用内购买、API调用 | | Company J | 音频生成 | 订阅、API调用 | | Company K | 生产力、通用2B服务 | 订阅、API调用 | | Company L | 视觉生成 | 订阅、API调用 | | Company M | 视觉生成 | 订阅、API调用 | | Company N | 生产力、通用2B服务 | 订阅、API调用 |

下表总结了全球主要基础模型技术公司在文本、图像、视频及音频模态方面所采用的核心底层技术，以高层次方式呈现各公司的技术方向。

全球主要基础模型技术公司所采用的底层技术

模型模态

| 公司 | 文本 | 图像 | 视频 | 音频 | |---|---|---|---|---| | Company A | RLHF、SFT、RAG、CoT、MoE | 扩散模型、交叉注意力、MoE、ViT | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company B | RLHF、SFT、RAG、CoT、MoE | 扩散模型、交叉注意力、MoE、ViT | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company C | RLHF、SFT、RAG、CoT、MoE | 扩散模型、交叉注意力、ViT | 时空注意力、视频扩散模型、多模态融合 | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company D | RLHF、SFT、CoT、交错思考 | N.A. | N.A. | N.A. | | Company E | N.A. | N.A. | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company F | RLHF、SFT、RAG、CoT、MoE | 扩散模型、交叉注意力 | 时空注意力、视频扩散模型、多模态融合、MoE | N.A. |

– 160 –

行业概览

模型模态

| 公司 | 文本 | 图像 | 视频 | 音频 | |---|---|---|---|---| | Company G | RLHF、SFT、CoT、MoE | 扩散模型、交叉注意力、MoE、ViT | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company H | RLHF、SFT、RAG、CoT、MoE、线性注意力 | 扩散模型、交叉注意力、MoE、ViT | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company I | RLHF、SFT、RAG、CoT | 交叉注意力、ViT | N.A. | N.A. | | MiniMax | RLHF、SFT、RAG、CoT、MoE、线性注意力、交错思考 | 扩散模型、交叉注意力、MoE、ViT | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company J | N.A. | N.A. | N.A. | TTS、基于编解码器的模型、基于扩散的声码器 | | Company K | RLHF、SFT、RAG、CoT、MoE | 扩散模型、交叉注意力、MoE、ViT | 时空注意力、视频扩散模型、多模态融合、MoE | TTS、基于编解码器的模型、基于扩散的声码器、MoE | | Company L | N.A. | N.A. | 时空注意力、视频扩散模型、多模态融合 | TTS、基于编解码器的模型、基于扩散的声码器 | | Company M | N.A. | N.A. | 时空注意力、视频扩散模型、多模态融合 | TTS、基于编解码器的模型、基于扩散的声码器 | | Company N | RLHF、SFT、RAG、CoT | N.A. | N.A. | TTS、基于编解码器的模型、基于扩散的声码器、MoE |

– 161 –

行业概览

注释： 1. RLHF（基于人类反馈的强化学习）是指一种训练方法，模型通过对人类评价较优的输出给予奖励，从而学习生成符合人类偏好的回应。

2. SFT（监督微调）是指通过在具有已知正确答案的高质量示例上对模型进行训练，以改进模型性能的过程，使模型学习模仿期望的行为。

3. RAG（检索增强生成）是指一种技术，模型从外部数据库或文档中检索相关信息，以生成更准确、更符合事实的答案。

4. CoT（思维链推理）是指一种推理方法，模型在给出最终答案前生成中间思考步骤，从而提升逻辑准确性和问题解决能力。

5.

INDUSTRY OVERVIEW

MoE（混合专家模型）是指一种包含多个专业化"专家"网络的模型架构，针对每次输入仅激活最相关的专家网络，从而提升效率与可扩展性。

6.

线性注意力（Linear Attention）是指一种经过优化的注意力机制，可降低内存和计算成本，使模型能够高效处理更长的输入序列。

7.

交错思考（Interleaved Thinking）是指一种推理方式，模型在多条思维路径或任务之间交替切换，从而更高效地处理复杂问题并生成更连贯、更具上下文感知能力的输出。

8.

扩散模型（Diffusion Model）是指一种生成式方法，从随机噪声出发，逐步将其精炼为清晰图像，类似于冲洗照片的过程。

9.

交叉注意力（Cross-Attention）是指一种使模型能够连接并对齐不同来源信息的机制，例如将文本提示与视觉特征相关联。

10.

ViT（视觉Transformer）是指一种专为图像理解设计的Transformer架构，将图像划分为小块（patch）并加以处理，以捕捉全局视觉模式。

11.

时空注意力（Spatiotemporal Attention）是指一种同时分析空间信息（每帧中的对象）和时间信息（对象跨帧的运动方式）以理解视频内容的注意力机制。

12.

视频扩散模型（Video Diffusion Model）是指将扩散模型扩展至视频领域的生成框架，通过逐步对含噪视频帧去噪，生成流畅连贯的动态效果。

13.

多模态融合（Multimodal Fusion）是指将文本、图像、音频、视频等多种类型的数据加以整合的过程，使模型能够跨模态理解或生成内容。

14.

TTS（文本转语音）是指将书面文字转换为口语语音的过程，使机器能够自然地"说话"。

15.

基于编解码器的模型（Codec-based Model）是指一种音频生成方法，将声音压缩为紧凑的数字编码（token）并以高保真度重建，类似于MP3或EnCodec的工作原理。

16.

基于扩散的声码器（Diffusion-based Vocoder）是指一种通过逐步去噪过程从编码表示中重建真实音频波形的模型，可提升语音和音乐质量。

基础模型技术公司可进一步分为两类：核心业务完全聚焦于基础模型的纯基础模型公司（pureplay companies），以及在现有业务基础上进入基础模型领域的非纯基础模型公司（non-pureplay companies），例如主要互联网平台和云服务提供商。

– 162 –

INDUSTRY OVERVIEW

纯基础模型公司将其核心资源、积累的技术积淀及商业模式集中于基础模型领域。这种高度专注和资源投入使其能够推动快速创新，并使其成为推动基础模型行业发展的核心力量。相比之下，非纯基础模型公司可能在资本获取和算力方面具有更强优势。此类公司还能够将基础模型技术整合至组织内其他业务部门的更广泛产品和服务中，从而有可能实现更快速、更便捷的新技术验证与商业化落地。

根据灼识咨询（CIC）的数据，MiniMax按2024年基于模型的收入计算，是全球第四大纯基础模型技术公司，具体情况如下表所示。

2024年全球纯基础模型技术公司排名 （按基于模型的收入）

| 排名 | 公司 | 市场份额（%） | |------|------|--------------| | 1 | A公司 | 30.1% | | 2 | D公司 | 4.7% | | 3 | E公司 | 2.8% | | 4 | MiniMax | 0.3% | | 5 | J公司 | 0.3% |

来源：灼识咨询（CIC）

注： （1） 基于模型的收入主要包括基础模型应用订阅、基础模型API调用及许可所产生的收入。

– 163 –

INDUSTRY OVERVIEW

模型基准测试

MiniMax基础模型在发布时，已在文本、视频和语音模态上实现领先性能，在Artificial Analysis基准测试中位列前茅，并在所有开源模型中荣登榜首。Artificial Analysis是基础模型行业广泛认可的权威、独立AI基准测试平台，从大模型用户视角提供评估，具体情况如下图所示。

Artificial Analysis智能指数（文本模型评估）

| 排名 | 公司 | 模型 | 指数 | |------|------|------|------| | 1 | OpenAI | GPT-5 Codex (high) | 68 | | 1 | OpenAI | GPT-5 (high) | 68 | | 3 | X | Grok 4 | 65 | | 4 | Anthropic | Claude 4.5 Sonnet | 63 | | 5 | MiniMax | MiniMax-M2 | 61 | | 5 | OpenAI | gpt-oss-120B (high) | 61 | | 7 | X | Grok 4 Fast | 60 | | 7 | Google | Gemini 2.5 Pro | 60 | | 9 | Anthropic | Claude 4.1 Opus | 59 | | 10 | Alibaba | Qwen3 235B A22B 2507 | 57 |

来源：Artificial Analysis

注：数据截至2025年11月7日，即MiniMax-M2发布后不久。Artificial Analysis是一家独立的AI基准测试与分析公司，为开发者、研究人员、企业及其他AI用户提供独立基准测试与分析。Artificial Analysis智能指数是各项评估的加权平均指标，综合考量通用知识（MMLU-Pro、HLE及GPQA Diamond等权重均等）、数学推理（MATH-500及AIME 2024等权重均等）以及编程能力组合（SciCode及LiveCodeBench等权重均等）。

– 164 –

INDUSTRY OVERVIEW

Artificial Analysis视频竞技场排行榜（视频模型评估）

| 排名 | 公司 | 模型 | 竞技场ELO分 | |------|------|------|------------| | 1 | ByteDance（字节跳动） | Seedance 1.0 | 1,355 | | 2 | MiniMax | Hailuo-02 | 1,331 | | 3 | Google | Veo 3 Preview (No Audio) | 1,244 | | 4 | Kuaishou（快手） | Kling 2.0 | 1,195 | | 5 | Kuaishou（快手） | Kling 1.6 (Pro) | 1,144 | | 6 | Runway | Runway Gen 4 | 1,120 | | 7 | Google | Veo 2 | 1,118 | | 8 | Lightricks | LTV Video v0.9.7 (13B) | 1,064 | | 9 | MiniMax | I2V-01-Director | 1,047 | | 10 | Runway | Runway Gen 3 Alpha Turbo | 1,005 |

Source: Artificial Analysis Note: As of June 22, 2025, shortly after the release of Hailuo-02. The Arena ELO scores are determined by responses from users in the Artificial Analysis Video Arena.

Artificial Analysis Speech Arena Leaderboard (evaluation of speech models)

| Rank | Company | Model | Arena ELO | |------|-----------|--------------------------------|-----------| | 1 | MiniMax | Speech-02-HD | 1,174 | | 2 | OpenAI | TTS-1 HD | 1,146 | | 3 | OpenAI | TTS-1 | 1,132 | | 4 | ElevenLabs | Multilingual v2 | 1,114 | | 5 | ElevenLabs | Turbo v2.5 | 1,108 | | 6 | Cartesia | Sonic English (Oct'24) | 1,103 | | 7 | Kokoro | Kokoro 82M v1.0 | 1,078 | | 8 | Microsoft | Azure Neural | 1,056 | | 8 | Amazon | Polly Long-Form | 1,056 | | 10 | Google | Studio | 1,039 |

Source: Artificial Analysis Note: As of June 22, 2025, shortly after the release of Speech-02. The Arena ELO scores are determined by responses from users in the Artificial Analysis Speech Arena.

– 165 –

INDUSTRY OVERVIEW

Competitive Barriers

R&D Capabilities of Foundation Models

The competitiveness of foundation model products is fundamentally based on the underlying foundation models. Performance improvements driven by the iteration of foundation models often far outweigh enhancements made at the application layer or through product refinement. As a result, leading foundation model products nowadays are typically developed by companies with in-house foundation model R&D capabilities, while users tend to gravitate toward top-tier products that offer the best experience. Given the rapid pace of technological advancement, players in the industry must continue investing heavily in R&D to maintain performance leadership and their competitive edge.

Commercialization Capabilities

Commercialization capabilities enable foundation model companies to translate research and technologies into usable products more rapidly, shortening the cycle from technological development to tangible commercial value. By strategically selecting and developing products with the greatest potential for scalable commercialization, foundation model companies can further amplify the market impact of technological breakthroughs, improve the ROI of model development, and support the long-term sustainability of ongoing research efforts.

Organizational Abilities

Developing foundation models requires the integration of expertise across multiple complex domains, including advanced algorithms, large-scale model training, infrastructure optimization, and deployment efficiency. As such, companies must rely heavily on a small pool of top-tier AI talents with deep technical capabilities. To attract and retain these individuals, companies need organizational abilities — including a compelling long-term vision, research environment, capital support, and a culture that fosters innovation and ownership. These organizational qualities form a critical barrier to entry, enabling leading players to continuously overcome technical bottlenecks and maintain a sustainable competitive advantage.

Key Costs and Trends of the Global Foundation Model Market

Inference cost is the major cost for companies engaged in the global foundation model market. It refers to the computational expense incurred each time a user query is processed by the model, and is typically charged on a per-token basis. With the continued maturation of foundation model technologies and increasing economies of scale in commercialization, inference costs are expected to decline significantly. According to CIC, the industry average inference cost declined from approximately US$20 per million tokens by the end of 2022 to below US$0.1 per million tokens by the end of 2024, and is expected to further decline at an approximate rate of 10 times per year.

– 166 –

INDUSTRY OVERVIEW

SOURCE OF INFORMATION

CIC was commissioned to conduct research and analysis of, and produce a report on the global foundation model industry at a fee of US$115,000. The commissioned report has been prepared by CIC independently without the influence from the Company or other interested parties. CIC offers industry consulting services, commercial due diligence, and strategic consulting. With a consultant team actively tracking the latest market trends in various industries such as TMT, consumer goods and services, agriculture, chemicals, marketing and advertising, culture and entertainment, energy and industry, finance and services, healthcare, and transportation, CIC possesses the most relevant and insightful market intelligence in these sectors. CIC undertook both primary and secondary research using a variety of resources. Primary research involved interviewing key industry experts and leading industry participants. Secondary research involved analyzing data from various publicly available data sources, including annual reports published by relevant industry participants, industry associations, CIC's own internal database, etc.

The market projections in the commissioned report are based on the following key assumptions: (i) the overall global social, economic, and political environment is expected to maintain a stable trend during the forecast period, (ii) key industry drivers are likely to continue to drive market growth during the forecast period, and (iii) there is no extreme force majeure or unforeseen industry regulations in which the market may be affected either dramatically or fundamentally during the forecast period. Except as otherwise noted, all of the data and forecasts contained in this section are derived from the CIC Report.

– 167 –

监管概览

概述

本节概述了可能对我们业务产生重大影响的主要法律、规则及法规。

中国境内的法律法规

人工智能政府政策

2017年7月8日，国务院发布《新一代人工智能发展规划》，明确了中国发展新一代人工智能的"三步走"战略目标：到2020年，人工智能总体技术和应用与世界先进水平同步，人工智能产业成为新的重要经济增长点，人工智能技术应用成为改善民生的新途径；到2025年，人工智能基础理论实现重大突破，部分技术与应用达到世界领先水平，人工智能成为带动我国产业升级和经济转型的主要动力，智能社会建设取得积极进展；到2030年，人工智能理论、技术与应用总体达到世界领先水平，使中国成为世界主要人工智能创新中心。

2021年12月31日，国家互联网信息办公室（"网信办"）与其他三个部门联合发布《互联网信息服务算法推荐管理规定》，该规定于2022年3月1日起施行。该规定适用于利用生成合成、个性化推送、排序精选、检索过滤、调度决策等算法技术向用户提供信息的企业（即算法推荐服务提供者）。根据该规定，算法推荐服务提供者应落实算法安全主体责任，定期审核、评估、验证算法机制机理、模型、数据和应用结果等，不得设置诱导用户沉迷、过度消费或违反法律法规及伦理道德的算法模型，并应加强信息安全管理。算法推荐服务提供者应保护用户权益，向用户提供不针对其个人特征的算法推荐选项，或向用户提供便捷的关闭算法推荐服务的选项。向未成年人提供服务的，应当依法履行网络空间未成年人保护义务。算法推荐服务提供者应建立便捷有效的用户申诉和公众投诉举报渠道，明确受理、处理及反馈结果的程序和时限，及时受理、处理并反馈结果。具有舆论属性或社会动员能力的算法推荐服务提供者，应当依法履行备案手续并按照国家有关规定开展安全评估。

— 168 —

REGULATORY OVERVIEW

2022年11月25日，国家互联网信息办公室（CAC）联合另外两个部门共同发布了《互联网信息服务深度合成管理规定》，该规定于2023年1月10日起正式施行。上述规定对深度合成服务提供者、技术支持者及用户规定了相应义务，包括：核验用户真实身份、落实数据安全和个人信息保护措施、加强深度合成内容管理，以及对利用深度合成技术生成或编辑的信息内容进行标注。具有舆论属性或社会动员能力的深度合成服务提供者，应当依照《互联网信息服务算法推荐管理规定》，完成深度合成服务算法的备案及变更、注销手续。深度合成服务的技术支持者在办理备案、变更及注销手续时，应参照本规定执行。深度合成服务提供者在开发上线具有舆论属性或社会动员能力的新产品、新应用、新功能时，应依据国家相关规定开展安全评估。

2023年7月10日，国家互联网信息办公室联合其他六个部门共同发布了《生成式人工智能服务管理暂行办法》（"AIGC管理办法"），该办法于2023年8月15日起正式施行。根据AIGC管理办法的定义，生成式人工智能技术是指具有文本、图片、音频、视频等内容生成能力的模型及相关技术。生成式人工智能服务提供者，是指利用生成式人工智能技术提供生成式人工智能服务（包括通过可编程接口等方式提供生成式人工智能服务）的组织或个人。就适用范围而言，AIGC管理办法适用于在中华人民共和国境内，利用生成式人工智能技术向公众提供文本、图片、音频、视频等内容生成服务的活动。国家对利用生成式人工智能服务从事新闻出版、影视制作、文艺创作等活动另有规定的，从其规定。行业组织、企业、教育科研机构、公共文化机构及其他相关专业机构研究、开发、应用生成式人工智能技术，但不向境内公众提供生成式人工智能服务的，不适用本办法。在治理机制方面，AIGC管理办法规定生成式人工智能服务提供者应当依法使用具有合法来源的数据和基础模型开展预训练、微调等数据处理活动。涉及知识产权的，不得侵害他人依法享有的知识产权。涉及个人信息的，应当取得个人同意或者符合法律、行政法规规定的其他情形。提供者应当采取有效措施提高训练数据质量，增强训练数据的真实性、准确性、客观性、多样性；制定标注规则，开展数据标注质量评估，并对标注人员进行培训。在生成式人工智能服务监管方面，AIGC管理办法要求生成式人工智能服务提供者采取有效措施，防止未成年用户对生成式人工智能服务产生过度依赖或沉迷。

— 169 —

REGULATORY OVERVIEW

to generative AI services; to lawfully assume the responsibilities of personal information processors; and to fulfill obligations to protect the input information and usage records of generative AI service users. Providers shall promptly accept and handle individuals' requests to access, copy, correct, supplement, or delete their personal information. Providers shall label generated content such as images and videos in accordance with the Provisions on the Administration of Deep Synthesis of Internet-Based Information Services. Upon discovering illegal content, providers shall promptly take measures such as ceasing generation, halting transmission, or removal, and take corrective measures such as model optimization training. If a provider discovers that a user is using generative AI services to engage in illegal activities, it shall take relevant measures in accordance with the law and contractual agreements, retain relevant records, and report the matter to the competent authorities. Providers shall also establish and improve mechanisms for complaints and whistleblowing.

On March 7, 2025, the CAC and three other departments jointly issued the Measures for the Identification of AI-Generated and Synthesized Content (《人工智能生成合成内容标识办法》) (the "Identification Measures"), which came into effect on September 1, 2025. According to the Identification Measures, internet information service providers engaging in the identification of AI-generated and synthesized content that falls within the scope of the Provisions on the Administration of Algorithm-generated Recommendations for Internet Information Services, the Provisions on the Administration of Deep Synthesis of Internet-Based Information Services, or the AIGC Administration Measures shall be subject to these Measures. The Identification Measures specify that service providers shall add explicit identification to AI-generated and synthesized content such as text, audio, images, video, and virtual scenes. When providing functions such as downloading, copying, or exporting such content, providers shall ensure that the files contain the required explicit identification. In addition, implicit identification shall be embedded in the metadata of files containing AI-generated and synthesized content. The implicit identification shall include information on the attributes of the content, the name or code of the service provider, the content identification number, and other production-related elements. Service providers shall also specify in their user service agreements the methods, formats, and standards for identification of AI-generated and synthesized content, and shall remind users to carefully read and understand the relevant identification management requirements. Article 9 of the Identification Measures provides that where a user takes the initiative to request content without explicit identification, the website platform may, on the condition of not violating relevant laws and regulations, provide such content to the user after clearly defining the responsibilities and obligations in the user agreement and retaining the relevant log information in accordance with the law. Meanwhile, in subsequent use, the user must comply with Article 10 and other relevant provisions of the Identification Measures, proactively declare the AI-generated or synthesized nature of the content, and add explicit identification before publishing or disseminating such content to the public. Furthermore, Article 10 of the Identification Measures stipulates that no organization or individual shall maliciously delete, alter, forge, or conceal the identification of AI-generated and synthesized content as required under these Measures, nor shall they provide tools or services to others for carrying out such malicious acts. It is also prohibited to infringe upon the lawful rights and interests of others through improper means of identification.

– 170 –

REGULATORY OVERVIEW

Guidance Catalogue for the Industrial Structure Adjustment

According to the Guidance Catalogue for the Industrial Structure Adjustment (2024 Edition) (《產業結構調整指導目錄(2024年本)》), which was issued by the National Development and Reform Commission on December 27, 2023 and came into effect on February 1, 2024, industries such as big data, cloud computing, information technology services, and blockchain information services within the extent permitted in the PRC are under the encouraged category.

Outline of the 14th Five-Year Plan for National Economic and Social Development

The 14th Five-Year Plan for National Economic and Social Development and the Long-Range Objectives through the Year 2035 of the PRC (《中華人民共和國國民經濟和社會發展第十四個五年規劃和2035年遠景目標綱要》), which was issued by the National People's Congress (the "NPC") on March 12, 2021 and came into effect on the same day, explicitly emphasizes focusing on critical sectors including high-end chips, operating systems, key algorithms for artificial intelligence, and sensors. It also underscores the importance of accelerating research and development breakthroughs in basic theories, fundamental algorithms, and equipment materials.

Regulations on Foreign Investment

The Company Law of the PRC (《中華人民共和國公司法》), which was promulgated by the Standing Committee of the NPC of the PRC (the "NPC Standing Committee") on December 29, 1993, and was most recently amended on December 29, 2023, with its latest revision taking effect on July 1, 2024, governs matters related to the incorporation, operation, and management of companies in China, including foreign-invested enterprises. Unless otherwise specified by laws related to foreign investment, foreign-invested companies are required to comply with the provisions of the Company Law of the PRC.

Foreign investment in China shall adhere to the Catalogue of Encouraged Industries for Foreign Investment (2022 Edition) (《鼓勵外商投資產業目錄(2022年版)》) (the "Catalogue") issued by the National Development and Reform Commission of the PRC (the "NDRC") and the Ministry of Commerce of the PRC (the "Ministry of Commerce"), which was revised on October 26, 2022 and became effective on January 1, 2023, as well as the Special Administrative Measures (Negative List) for Foreign Investment Access (2024 Edition) (《外商投資准入特別管理措施(負面清單)(2024年版)》) (the "Negative List"), which was promulgated on September 6, 2024 and became effective on November 1, 2024. The Catalogue and the Negative List delineate the basic framework for foreign investment in China, classifying industries with foreign investment into three categories: "Encouraged," "Restricted," and "Prohibited". Industries not listed in either the Catalogue or the Negative List are generally considered to be in the "Permitted" category unless otherwise expressly restricted by other PRC laws and regulations.

– 171 –

REGULATORY OVERVIEW

The Law of the PRC on Foreign Investment (《中華人民共和國外商投資法》) (the "Foreign Investment Law"), which was promulgated by the NPC Standing Committee on March 15, 2019 and came into effect on January 1, 2020, and the Regulations for the Implementation of the Foreign Investment Law of the PRC (《中華人民共和國外商投資法實施條例》) (the "Regulations for the Implementation of the Foreign Investment Law"), which was promulgated by the State Council on December 26, 2019 and came into effect on January 1, 2020, constitute the primary prevailing legal framework governing foreign investment in China. The promulgation of the Foreign Investment Law and the Regulations for the Implementation of the Foreign Investment Law aims to further expand opening-up, promote foreign investment actively, protect the legitimate rights and interests of foreign investors, and regulate foreign investment management.

On December 30, 2019, the Ministry of Commerce and the State Administration for Market Regulation jointly issued the Measures for the Reporting of Foreign Investment Information (《外商投資信息報告辦法》) (the "Reporting Measures"), which became effective on January 1, 2020. The Reporting Measures regulate the reporting of information related to foreign investment activities within China. According to the Reporting Measures, foreign investors and foreign-invested enterprises conducting investment activities within China, either directly or indirectly, are required to submit investment information to the competent commercial authorities through initial reports, change reports, deregistration reports, and annual reports.

Regulations Relating to Overseas Listing

On February 17, 2023, the China Securities Regulatory Commission of the PRC (the "CSRC") issued the Interim Measures for the Administration of Overseas Securities Offering and Listing by Domestic Enterprises (《境內企業境外發行證券和上市管理試行辦法》) (the "Interim Measures for Overseas Listing") along with five supportive guidelines, which took effect on March 31, 2023. Prior to this, the foundational regulations governing the overseas offering and listing by domestic enterprises, namely the Special Provisions of the State Council on Issuing and Listing of Shares Abroad by Companies Limited by Shares (《國務院關於股份有限公司境外募集股份及上市的特別規定》) and the Circular of the State Council on Further Strengthening the Management of Share Issuance and Listing Overseas (《國務院關於進一步加強在境外發行股票和上市管理的通知》), were simultaneously abolished on March 31, 2023.

According to the Interim Measures for Overseas Listing, domestic enterprises seeking to offer and list securities directly or indirectly in foreign markets are required to complete filing procedures with the CSRC and submit relevant documentation. The Interim Measures for Overseas Listing specify that no overseas offering and listing shall be conducted under any of the following circumstances: (i) Financing through listing is expressly prohibited by laws, administrative regulations or relevant rules of the State; (ii) the overseas offering and listing may endanger national security as determined by the relevant competent department under the State Council after examination according to the law; (iii) a domestic enterprise or its controlling shareholder or actual controller has committed a criminal crime of corruption, bribery, embezzlement, misappropriation of property or disrupting the economic order of the

– 172 –

REGULATORY OVERVIEW

socialist market in the last three years; (iv) a domestic enterprise is under formal investigation according to the law for being suspected of any crime or major violation of laws and regulations, but no clear conclusions have been made; or (v) there is a major dispute over ownership of the equity held by the controlling shareholder or a shareholder controlled by the controlling shareholder or the actual controller.

The Interim Measures for Overseas Listing also specify that any overseas offering and listing conducted by an issuer that concurrently meets the following conditions shall be determined as indirect overseas offering and listing by a domestic enterprise: (i) Among the operating revenue, total profits, total assets or net assets of the domestic enterprise in the most recent fiscal year, any index accounts for over 50% of the relevant data in the audited consolidated financial statements of the issuer for the same period; and (ii) the main parts of the business activities of the issuer are carried out in China Mainland or the main business places are located in China Mainland, or most of the senior executives in charge of business operation are Chinese citizens, or their habitual residences are located in China Mainland. An issuer applying to relevant offshore regulatory authorities for an initial public offering shall undergo the recordation formalities with the CSRC within three working days after the application documents for offering and listing are submitted overseas. Furthermore, the Interim Measures for Overseas Listing stipulate that upon the occurrence of any of the material events specified below after an issuer has offered and listed securities in an overseas market, the issuer shall submit a report thereof to the CSRC within three working days after the occurrence and public disclosure of the event: change of control; investigations or sanctions imposed by overseas securities regulatory agencies or relevant competent authorities; change of listing status or transfer of listing segment; voluntary or mandatory delisting.

To enhance confidentiality and archives administration related to domestic enterprises' overseas offering and listing, on February 24, 2023, the CSRC, jointly with the Ministry of Finance, the National Administration of State Secrets Protection and the National Archives Administration, issued the Provisions on Strengthening Confidentiality and Archives Administration of Overseas Securities Offering and Listing by Domestic Companies (《关于加强境内企业境外发行证券和上市相关保密和档案管理工作的规定》) (CSRC Announcement [2023] No. 44), which took effect on March 31, 2023 and supersedes the Provisions on Strengthening Confidentiality and Archives Administration Concerning Overseas Securities Offering and Listing (《关于加强在境外发行证券与上市相关保密和档案管理工作的规定》) (CSRC Announcement [2009] No. 29). These Provisions outline procedural requirements and specify enterprises' confidentiality responsibilities and accounting archives administration standards, in alignment with the Interim Measures for Overseas Listing.

Regulations Relating to Anti-monopoly and Anti-unfair Competition

According to the Anti-unfair Competition Law of the PRC (《中华人民共和国反不正当竞争法》) promulgated by the NPC Standing Committee on September 2, 1993, effective from December 1, 1993 and most recently amended on 27 June 2025, unfair competition means that in its production or operation activity, a business operator disrupts the order of market competition and causes damage to the lawful rights and interests of other business operators

– 173 –

REGULATORY OVERVIEW

或消费者，违反了《中华人民共和国反不正当竞争法》。根据《中华人民共和国反不正当竞争法》，经营者在市场交易中应当遵循自愿、平等、公平、诚实信用的原则，遵守法律和商业道德。违反《中华人民共和国反不正当竞争法》的经营者，根据具体情况承担相应的民事责任、行政责任或刑事责任。

2021年2月7日，国务院反垄断委员会发布《国务院反垄断委员会关于平台经济领域的反垄断指南》（"反垄断指南"），概述了若无正当理由可能构成滥用市场支配地位的若干行为。

2024年5月6日，国家市场监督管理总局颁布《网络反不正当竞争暂行规定》，于2024年9月1日起施行。该规定为预防和制止网络不正当竞争行为、维护公平竞争市场秩序、鼓励创新、保护经营者和消费者合法权益、促进数字经济规范持续健康发展提供了监管依据。

根据《中华人民共和国反垄断法》（"反垄断法"），该法由全国人民代表大会常务委员会于2022年6月24日修订，并于2022年8月1日起施行，反垄断法适用于中国境内经济活动中的垄断行为，以及在中国境外实施但对境内市场竞争具有排除、限制影响的垄断行为。反垄断法规定的垄断行为包括：经营者达成垄断协议、经营者滥用市场支配地位，以及具有或可能具有排除、限制市场竞争效果的经营者集中。国务院反垄断执法机构负责统一的反垄断执法工作。根据规定，国务院反垄断执法机构可以依照反垄断法的规定，授权省、自治区、直辖市人民政府的相应机构负责有关反垄断执法工作。违反反垄断法的经营者，将被执法机构责令停止违法行为，并处以罚款或其他限制性措施。

2024年1月22日，国务院发布《国务院关于经营者集中申报标准的规定》，进一步明确了判断一个企业是否取得对另一个企业的控制权或能够对另一个企业施加决定性影响所需考虑的因素。

– 174 –

REGULATORY OVERVIEW

Regulations Relating to Consumer Protection

The Law of the PRC on Protecting Consumers' Rights and Interests (《中華人民共和國消費者權益保護法》) (the "Law on Protecting Consumers' Rights and Interests") was first promulgated by the NPC Standing Committee on October 31, 1993 and last amended on October 25, 2013, and came into effect on March 15, 2014. The Law on Protecting Consumers' Rights and Interests sets out the obligations of business operators and the rights and interests of consumers. Business operators must guarantee the quality, function, usage and term of validity of the goods or services they sell or provide. Consumers whose rights and interests have been damaged due to their purchase of goods or acceptance of services on online platforms may claim damages from the sellers or service providers. Online platform operators may be subject to liabilities if the lawful rights and interests of consumers are infringed in connection with consumers' purchase of goods or acceptance of services on online platforms and the online platform operators fail to provide consumers with authentic contact information of the sellers or service providers. The Regulations for the Implementation of the Law of the PRC on Protecting Consumers' Rights and Interests (《中華人民共和國消費者權益保護法實施條例》) was promulgated by the State Council on March 15, 2024 and came into effect on July 1, 2024, according to which, if the business operators adopt automatic extension, automatic renewal, or other similar mechanisms in connection with the provisions of their services, the business operators must prominently draw the attention of the consumers before they accept the service and before the dates of automatic extension, automatic renewal, or effectiveness of other mechanisms. Business operators are prohibited from sending commercial information or making commercial calls to consumers without their prior consent. If a consumer agrees to receive commercial information and/or commercial calls, the business operator must provide clear and easily accessible options for opting out. Upon the consumer's request to opt out, the business operator shall immediately stop sending commercial information or making commercial calls.

Regulations Relating to Cybersecurity and Data Protection

On July 1, 2015, the NPC Standing Committee issued the National Security Law of the PRC (《中華人民共和國國家安全法》) (the "National Security Law"), which came into effect on the same day. The National Security Law stipulates that the State must safeguard national sovereignty, security and the development interests in cyberspace. It also requires the establishment of national security review and supervision systems to examine foreign investments, critical technologies, internet information technology products and services, and other significant activities that may impact China's national security.

On November 7, 2016, the NPC Standing Committee promulgated the Cybersecurity Law of the PRC (《中華人民共和國網絡安全法》) (the "Cybersecurity Law"), which came into effect on June 1, 2017. According to the Cybersecurity Law, the State implements a cybersecurity multi-level protection system. Network operators shall comply with laws and regulations when conducting business and providing services, and fulfill their obligations to protect cybersecurity. Service providers operating via the network are required to adopt technical measures and other necessary measures, in accordance with laws, administrative – 175 –

REGULATORY OVERVIEW

regulations, and mandatory national standards, to ensure the safe and stable operation of the network, effectively respond to cybersecurity incidents, prevent illegal criminal activities committed on the network, and maintain the integrity, confidentiality and availability of network data.

According to the Civil Code of the PRC (《中華人民共和國民法典》) promulgated by the NPC on May 28, 2020, and effective from January 1, 2021, the personal information of a natural person shall be protected by law. Any organization or individual needing to obtain the personal information of other persons shall legally obtain and ensure the security of such information, and shall not illegally collect, use, process, or transmit the personal information of other persons, nor illegally buy, sell, provide, or publish the personal information of other persons.

On August 20, 2021, the NPC Standing Committee promulgated the Personal Information Protection Law of the PRC (《中華人民共和國個人信息保護法》) (the "Personal Information Protection Law"), which took effect on November 1, 2021. The Personal Information Protection Law further emphasizes and specifies the obligations and responsibilities of personal information processors in personal information processing activities, and establishes a comprehensive set of rules for personal information processing, including but not limited to: personal information processing shall be for a clear and reasonable purpose; sensitive information processing shall have additional protection; the provision and entrusted processing of personal information to external parties shall be subject to the signing of special agreements to ensure security; the storage, deletion, disclosure, and automated decision-making of personal information shall comply with specific rules; and personal information processors shall have appropriate organizational safeguards, systematic safeguards, and technical measure safeguards.

On June 10, 2021, the NPC Standing Committee promulgated the Data Security Law of the PRC (《中華人民共和國數據安全法》) (the "Data Security Law"), which came into force on September 1, 2021. The Data Security Law specifies a categorized and classified system for data protection based on the importance of the data in economic and social development, as well as the extent of harm to national security, public interests, or the lawful rights and interests of individuals or organizations that will be caused once the data are altered, destroyed, leaked, or illegally obtained or used. Entities engaging in data processing activities shall, in accordance with the laws and regulations, establish a sound data security management system throughout the whole process, organize and conduct data security education and training, and adopt corresponding technical measures and other necessary measures to ensure data security. The law also provides for national security review procedures for data activities that affect or may affect national security. It stipulates that processors of important data shall be clear about the persons responsible for data security and the data security management bodies, and fulfill the responsibilities for data security protection, conduct risk assessments of their data processing on a regular basis and submit risk assessment reports to relevant competent departments. Additionally, the Data Security Law subjects activities involving the provision of important data to overseas parties by data processors other than critical information

– 176 –

REGULATORY OVERVIEW

infrastructure operators to special regulatory procedures for data export, and restricts the transfer of data stored within the territory of China to any overseas judicial or law enforcement body without the approval of the competent authorities of the PRC.

On December 28, 2021, the Cyberspace Administration of China (the "CAC"), together with 12 other authorities, jointly promulgated the Measures for Cybersecurity Review (《网络安全审查办法》) (the "Measures for Cybersecurity Review"), which took effect on February 15, 2022. The Measures for Cybersecurity Review stipulate that: (i) online platform operators engaged in data processing activities that influence or may influence national security shall conduct a cybersecurity review; (ii) online platform operators that hold personal information of more than one million users and plan to go public abroad shall report for cybersecurity review to the Cybersecurity Review Office; (iii) critical information infrastructure operators that purchase network products and services affecting or possibly affecting national security shall also undergo cybersecurity review; (iv) network products and services, as well as data processing activities that the cybersecurity review work mechanism member units believe affect or may affect national security, shall, after being submitted to the Central Cyberspace Affairs Commission for approval according to procedures, be reviewed by the Cybersecurity Review Office in accordance with the provisions of these Measures.

On July 7, 2022, the CAC promulgated the Measures for the Security Assessment of Outbound Data Transfer (《数据出境安全评估办法》), which came into effect on September 1, 2022. The Measures for the Security Assessment of Outbound Data Transfer stipulate that data processors, who provide important data and personal information collected and generated in their operations within the territory of China to recipients overseas, shall conduct a security assessment of outbound data transfers according to these Measures. On March 22, 2024, the CAC issued the Provisions on Promoting and Regulating Cross-border Data Flows (《促进和规范数据跨境流动规定》) (the "New Outbound Data Transfer Provisions"), which took effect on the same date. The New Outbound Data Transfer Provisions stipulate that in case of inconsistencies with the Measures for the Security Assessment of Outbound Data Transfer, the New Outbound Data Transfer Provisions shall prevail. The New Outbound Data Transfer Provisions clarify specific situations in which certain obligations regarding cross-border data transfers (including declaring security assessments for outbound data transfers, entering into standard contracts for outbound personal information, and passing personal information protection certification) can be exempted: (i) the outbound provision of data that is collected and generated in international trade, cross-border transportation, academic cooperation, multinational production and manufacturing, marketing, and other activities that do not involve personal information or important data; (ii) the domestic processing and subsequent outbound provision of personal information that is previously collected and generated outside the territory of the PRC after transmission into China, provided that no personal information within the territory of the PRC or important data is incorporated during the processing activities; (iii) the outbound provision of personal information when it is truly necessary for the conclusion or performance of a contract to which the individual is a party; (iv) the outbound provision of employee personal information when it is truly necessary for implementing cross-border human resources management in accordance with legally established labor regulations and legally signed collective contracts; (v) the outbound provision of personal

– 177 –

REGULATORY OVERVIEW

information when it is truly necessary to protect the life, health, and property safety of natural persons in emergency situations; and (vi) the cumulative outbound provision of personal information of less than 100,000 individuals (excluding sensitive personal information) by data processors other than critical information infrastructure operators from January 1st of the current year.

On September 24, 2024, the State Council issued the Regulations on Network Data Security Management (《網絡數據安全管理條例》), which came into effect on January 1, 2025. The Regulations on Network Data Security Management aim to implement the general requirements for data security management set forth in the Cybersecurity Law, the Data Security Law, and the Personal Information Protection Law. The regulations reiterate the general provisions on data processing activities, personal information protection rules, important data security protection, cross-border network data security management, and the obligations of online platform service providers. The Regulations on Network Data Security Management clarify the definition of important data, further specify the obligations of important data processors, and require network data processors whose data processing activities affect or may affect national security to undergo national security review in accordance with relevant national regulations.

Regulations Relating to Intellectual Property

Trademarks

The Trademark Law of the PRC (《中華人民共和國商標法》) (the "Trademark Law") and the Implementing Regulations of the Trademark Law of the PRC (《中華人民共和國商標法實施條例》) regulate trademark registration, protection, and use in China. The Trademark Law was promulgated on August 23, 1982, and most recently revised on April 23, 2019, effective from November 1, 2019. It follows the "first-to-file" principle. The law grants exclusive rights to trademark registrants, administered by the Trademark Office of the State Administration for Industry and Commerce (國務院工商行政管理部門商標局) (the "Trademark Office").

The validity period of a registered trademark is ten years, renewable for successive ten-year periods. Renewal procedures shall be completed within 12 months before the expiration date, with a grace period of six months available. The Trademark Office shall publish an announcement for the trademark involved in renewed registration. The trademark registrant may license others through a licensing agreement, but details of the license shall be filed with the Trademark Office. Failure to file shall not be asserted against bona fide third parties. The licensor shall supervise the quality of the products, and the licensee shall maintain the quality of the products when using the registered trademark.

– 178 –

REGULATORY OVERVIEW

Patent

The patent activities in China are regulated by the Patent Law of the PRC (《中華人民共和國專利法》) ("Patent Law") and the Detailed Rules for the Implementation of the Patent Law of the PRC (《中華人民共和國專利法實施細則》). The Patent Law was promulgated on March 12, 1984, and most recently amended on October 17, 2020, effective from June 1, 2021. The patent administration department under the State Council is in charge of patent work nationwide. The departments for patent administration of the people's governments of provinces, autonomous regions, and municipalities directly under the Central Government shall be responsible for patent administration within their respective administrative areas.

The Patent Law and its Implementing Regulations recognize three types of patents: "inventions", "utility models", and "designs". An invention patent refers to a new technical solution proposed for a product, method, or their improvement. A utility model patent refers to a new technical solution suitable for practical use, proposed for the shape or structure of a product, or combination thereof. A design patent refers to a new design that is aesthetically pleasing and suitable for industrial application, pertaining to the overall or partial shape, pattern, or combination thereof, as well as the integration of color with shape or pattern. The term of an invention patent is 20 years, the term of a design patent is 15 years, and the term of a utility model patent is 10 years, all counted from the date of filing.

China follows the principle of "first-to-file", granting patent rights to the earliest applicant for the same invention. Any invention or utility model for which patent right may be granted shall possess novelty, inventiveness and practical applicability. The rights of the patent holder are protected by law, and others are only permitted to use the patent with proper authorization. Unless otherwise stipulated by law, unauthorized use constitutes patent infringement.

Copyright

According to the Copyright Law of the PRC (《中華人民共和國著作權法》) promulgated by the NPC Standing Committee on September 7, 1990, last amended on November 11, 2020, and effective from June 1, 2021, as well as the Implementation Regulations for the Copyright Law of the PRC (《中華人民共和國著作權法實施條例》) issued by the State Council on May 30, 1991, last amended on January 30, 2013, and effective from March 1, 2013, Chinese citizens, legal entities or other unincorporated organizations shall enjoy the copyright in their works, whether published or not. Works refer to intellectual achievements in the fields of literature, art and science, which are original and can be expressed in a certain form, including written works, oral works, photographic works, audiovisual works, and computer software. Copyright owners shall have various rights including right of publication, right of authorship and right of reproduction. And anyone who commits any of the infringing acts such as using another person's work without paying remuneration as required, or other acts of infringing copyright, shall, depending on the circumstances, bear civil liabilities such as ceasing the infringement, eliminating the effects, making an apology, or compensating for losses.

– 179 –

REGULATORY OVERVIEW

According to the Civil Code of the People's Republic of China (《中華人民共和國民法典》) issued by the NPC on May 28, 2020, and became effective on January 1, 2021, where a network user commits an infringing act by using network services, the right holder shall have the right to notify the network service provider to take necessary measures such as deletion, blocking, or disconnection of links. The notice shall include preliminary evidence of the infringement and the right holder's true identity information. Upon receiving the notice, the network service provider shall promptly transmit the notice to the relevant network user and take necessary measures based on the preliminary evidence of the infringement and the type of service; if it fails to take timely necessary measures, it shall bear joint and several liability with the network user for the expanded part of the damage. Where a network service provider knows or should know that a network user is infringing upon the civil rights and interests of others by using its network services but fails to take necessary measures, it shall bear joint and several liability with such network user.

According to the Provisions by the Supreme People's Court on Several Issues Concerning the Application of Law in Hearing Civil Dispute Cases Involving Infringement of the Right of Communication to the Public on Information Networks (《最高人民法院關於審理侵害信息網絡傳播權民事糾紛案件適用法律若干問題的規定》) issued by Supreme People's Court on December 29, 2020 and became effective on January 1, 2021, if a network service provider knows or should know that a network user is infringing upon the right to disseminate information through information networks by using its network services, but fails to take necessary measures such as deletion, blocking, or disconnection of links, or provides technical support or other assistance, the people's court shall determine that it constitutes an act of contributory infringement.

According to the Regulations on the Protection of Computer Software (《計算機軟件保護條例》) promulgated by the State Council on June 4, 1991, last revised on January 30, 2013, and became effective on March 1, 2013, Chinese citizens, legal persons, or other organizations enjoy the copyright (including the right of publication, right of authorship, right of modification, right of reproduction, right of distribution, right of rental, right of dissemination through information networks, right of translation, and other rights to which the software copyright owner is entitled) in the software they develop, regardless of whether the software has been published.

According to the Measures for the Registration of Computer Software Copyrights (《計算機軟件著作權登記辦法》) promulgated by the National Copyright Administration on April 6, 1992, last revised on June 18, 2004, and with the latest revision effective on July 1, 2004, software copyright and proprietary software copyright licensing contracts and transfer contracts shall be registered. The National Copyright Administration is the competent authority for software copyright registration, and it recognizes the Copyright Protection Center of China as the registration body for software. Applications that meet the requirements shall be registered, and the Copyright Protection Center of China shall issue the corresponding registration certificates.

– 180 –

REGULATORY OVERVIEW

Domain Name

According to the Measures for the Administration of Internet Domain Names (《互聯網域名管理辦法》) promulgated by the Ministry of Industry and Information Technology on August 24, 2017 (effective from November 1, 2017), and the Implementation Rules for National Top-Level Domain Registration (《國家頂級域名註冊實施細則》) issued by the China Internet Network Information Center on June 18, 2019 (effective from the same date), domain name holders shall register their domain names. The Ministry of Industry and Information Technology conducts the supervision and management of Internet domain names in China, while telecommunications authorities in various provinces, autonomous regions, and municipalities directly under the Central Government supervise and manage domain name services within their administrative regions. In principle, domain name registration follows the "first application, first registration" policy. Applicants shall provide accurate information to the domain registration service provider and establish a registration agreement with it. Upon completing the registration process, the applicant becomes the domain name holder.

Regulations Relating to Property Leasing

According to the Civil Code of the PRC, with the consent of the lessor, the lessee may sublease the leased property to a third party. In the case of subleasing by the lessee, the lease contract between the lessee and the lessor remains valid; if the third party causes damage to the leased property, the lessee shall bear the compensation liability. Any transfer of ownership of the leased property during the lessee's possession period under the lease contract does not affect the validity of the lease contract. Based on the Urban Real Estate Administration Law of the PRC (《中華人民共和國城市房地產管理法》) promulgated by the NPC Standing Committee on July 5, 1994, and last amended on August 26, 2019 and effective from January 1, 2020, and the Administrative Measures for Commodity House Leasing (《商品房屋租賃管理辦法》) issued by the Ministry of Housing and Urban-Rural Development on December 1, 2010, and effective from February 1, 2011, the parties involved in house leasing shall conclude a lease contract in accordance with the law. Within 30 days after signing the lease contract, the parties involved shall complete the procedures for housing lease registration and filing with the competent authority for construction (real estate) of the people's government of the municipalities directly under the Central Government, cities or counties at the location of the leased property. In case of any violations of the above provisions, the competent authority for construction (real estate) of the people's government of the municipalities directly under the Central Government, cities or counties shall order corrections within a specified period; if an individual fails to make corrections within the deadline, a fine of less than RMB1,000 may be imposed; if an entity fails to correct within the deadline, a fine between RMB1,000 and RMB10,000 may be imposed. The Civil Code of the PRC states that if the parties involved fail to register and record the lease contract in accordance with laws and administrative regulations, it does not affect the validity of the contract.

– 181 –

REGULATORY OVERVIEW Regulations Relating to Labor and Social Security Labor Law and Labor Contract Law According to the Labor Law of the PRC (《中華人民共和國勞動法》), which was promulgated on July 5, 1994, and amended on August 27, 2009 and December 29, 2018, employers must establish and improve labor hygiene systems, strictly implement national labor safety and hygiene regulations and standards, and provide labor safety and hygiene education to employees. Labor safety and hygiene facilities must meet statutory standards. Enterprises and employers must provide employees with labor safety and hygiene conditions that comply with relevant labor protection laws and regulations. The Labor Contract Law of the PRC (《中華人民共和國勞動合同法》), promulgated on June 29, 2007 and amended on December 28, 2012, and the Implementation Regulations of the Labor Contract Law of the PRC (《中華人民共和國勞動合同法實施條例》), promulgated on September 18, 2008, specify the particular provisions regarding the signing, terms, and termination of labor contracts, as well as the rights and obligations of employees and employers. When recruiting employees, employers shall truthfully inform them of the job content, working conditions, work location, occupational hazards, safety production conditions, remuneration, and any other information that the employees request. Social Insurance and Housing Provident Fund In accordance with the Social Insurance Law of the PRC (《中華人民共和國社會保險 法》) which was promulgated on October 28, 2010 and was last amended on December 29, 2018, as well as the Interim Regulations on the Collection and Payment of Social Insurance Premiums (《社會保險費徵繳暫行條例》) issued by the State Council on January 22, 1999 and last revised on March 24, 2019, employees are required to participate in basic pension insurance, basic medical insurance, unemployment insurance, occupational injury insurance, and maternity insurance. The contributions to basic pension insurance, basic medical insurance, and unemployment insurance are shared by employers and employees; contributions to occupational injury insurance and maternity insurance are paid solely by employers, with employees not required to contribute. According to the Notice of the General Office of the State Council on Issuing the Plan for the Pilot Program of Combined Implementation of Maternity Insurance and Basic Medical Insurance for Employees (《國務院辦公廳關於印發< 生育保險和職工基本醫療保險合併實施試點方案>的通知》) and the Opinions of the General Office of the State Council on Fully Promoting the Combined Implementation of Maternity Insurance and Basic Medical Insurance for Employees (《國務院辦公廳關於全面推進生育保 險和職工基本醫療保險合併實施的意見》), which were issued on January 19, 2017 and March 6, 2019, respectively, the maternity insurance and basic medical insurance for employees must be combined. In accordance with the Social Insurance Law of the PRC, employers are required to register their employees with the local social insurance administrative authorities, provide social insurance coverage for their employees, and withhold and pay the relevant social insurance premiums on their behalf. If an employer fails to complete such registration, the social insurance administrative department will order a rectification within a specified period. – 182 –

REGULATORY OVERVIEW

If the employer does not comply within the deadline, it may be fined between one and three times the amount of the social insurance premium payable. Employers failing to promptly contribute social security premiums in full amount shall be ordered by the social security premium collection agency to make or supplement contributions within a stipulated period, and shall be subject to a late payment fine computed from the due date at the rate of 0.05% per day; where the payment is not made within the stipulated period, the administrative authorities shall impose a fine ranging from one to three times of the amount in arrears.

In accordance with the Regulations on the Administration of Housing Provident Fund (《住房公積金管理條例》), which were promulgated on April 3, 1999 and amended on March 24, 2002 and March 24, 2019, respectively, employers shall pay and deposit housing provident funds on time and in full amount. Late or insufficient payments shall be prohibited. Employers shall process housing provident fund payment and deposit registrations with the housing provident fund management center. If an employer fails to process housing provident fund payment and deposit registrations or go through the formalities of opening housing provident fund accounts for its employees in violation of the aforesaid laws and regulations, the housing provident fund management center shall order it to complete such formalities within a prescribed time limit. If the employer fails to comply with the deadline, a fine ranging from RMB10,000 to RMB50,000 may be imposed. An employer that fails to contribute or under-contributes to the housing provident fund by the prescribed deadline in violation of these regulations shall be ordered by the housing provident fund management center to make the contributions within a stipulated time. Where the contribution has not been made after the expiration of the time limit, an application may be made to the people's court for compulsory enforcement.

According to the Supreme People's Court's Interpretation (II) on Several Issues Concerning the Application of Law in Labour Dispute Cases, if an employer and an employee agree, or the employee promises to the employer, that the employer does not need pay the social insurance premiums for such employee, the people's court shall hold that such agreement or promise invalid. If the employer fails to pay social insurance premiums in accordance with the law, and the employee requests to terminate the labor contract and demands the employer to pay economic compensation in accordance with the Labor Contract Law due to the employer's failure to contribute social insurance premiums, the people's court shall support such request in accordance with the law. Where the circumstances specified in the preceding paragraph exist, and the employer has made up for the social insurance contribution for such employee in accordance with the law, and then the employer requests the employee to reimburse the economic compensation that the employer has already paid to the employee for the previous lack of social insurance premiums, the people's court shall support such the employer's request of reimbursement in accordance with the law. As the Group has not encountered any of the circumstances outlined in the preceding paragraph, the Supreme People's Court's Interpretation (II) on Several Issues Concerning the Application of Law in Labour Dispute Cases will not impact the analysis of the Group's compliance with social insurance and housing provident fund contribution requirements under PRC laws and regulations.

– 183 –

REGULATORY OVERVIEW

Laws and Regulations Relating to Foreign Exchange Registration for Overseas Investment by PRC Residents

Pursuant to the Notice of the State Administration of Foreign Exchange ("SAFE") on Relevant Issues Concerning Foreign Exchange Administrative for Domestic Residents to Engage in Overseas Investment and Financing and Round Trip Investment via Special Purpose Vehicles (《國家外匯管理局關於境內居民通過特殊目的公司境外投融資及返程投資外匯管理有關問題的通知》) (the "SAFE Circular 37"), issued and implemented by the SAFE on July 4, 2014, domestic residents (including domestic institutions and individual residents) who, for the purpose of investment and financing, directly establish or indirectly control overseas enterprises by using assets or equity interests of domestic enterprises legally held by them, or by using their legally held overseas assets or equity interests, are required to register such overseas enterprises with the local branch of SAFE as "special purpose vehicles" (SPVs) in accordance with SAFE Circular 37. In the event of changes to the basic information of a registered overseas SPV — such as changes in domestic individual shareholders, name, or operating period — or significant events such as capital increase or decrease, equity transfer or replacement, merger, or division involving domestic individual residents, the relevant parties must promptly complete the procedures for amendment registration of foreign exchange for overseas investment with the foreign exchange authority. Where a domestic resident fails to complete the relevant foreign exchange registration as required, fails to truthfully disclose the actual controller of the round-trip investment enterprise, or makes false representations, any outbound remittance, inbound remittance, or foreign exchange settlement may be subject to rectification orders, warnings, and fines imposed by the foreign exchange authority.

In addition, according to SAFE Circular 37, where a non-listed SPV uses its equity or stock options as the underlying for equity incentive plans targeting directors, supervisors, senior management, or other employees with employment or labor relationships in domestic enterprises under its direct or indirect control, the relevant domestic individual residents may submit the required materials to the foreign exchange authority to apply for foreign exchange registration for the SPV prior to the exercise of such rights. However, in practice, local branches of SAFE may have different interpretations and implementations of SAFE Circular 37. Moreover, as SAFE Circular 37 is the first regulation governing the granting of equity incentives by overseas non-listed companies to domestic residents, there remains a degree of uncertainty in its implementation.

Laws and Regulations Relating to Employee Equity Incentive Plans

According to the Notice of the SAFE on Issues concerning the Foreign Exchange Administration of Domestic Individuals' Participation in Equity Incentive Plans of Overseas Listed Companies (《國家外匯管理局關於境內個人參與境外上市公司股權激勵計劃外匯管理有關問題的通知》) ("SAFE Circular 7"), issued by the SAFE on February 15, 2012, employees, directors, supervisors, and other senior management personnel who are Chinese citizens or non-Chinese citizens who have continuously resided in China for not less than one

– 184 –

REGULATORY OVERVIEW

year and who participate in equity incentive plans of overseas listed companies, are, except in a few exceptional cases, required to register with SAFE through a qualified domestic agent (which may be the Chinese affiliate of the overseas listed company) and complete a series of other procedures.

In addition, the State Taxation Administration has issued several notices concerning employee stock options and restricted shares. According to these notices, employees working in China who exercise stock options or are granted restricted shares are required to pay individual income tax in China. The Chinese affiliates of overseas listed companies are required to submit documentation related to employee stock options and restricted shares to the relevant tax authorities and to withhold individual income tax on behalf of employees who exercise stock options or purchase restricted shares. If an employee fails to pay the required taxes in accordance with relevant laws and regulations, or if the Chinese affiliate fails to withhold such taxes, the Chinese affiliate may be subject to penalties by the tax authorities or other Chinese government agencies.

Regulations Relating to PRC Taxation

Income Tax Law

Pursuant to the Enterprise Income Tax Law of the PRC (《中華人民共和國企業所得稅法》) promulgated by the NPC on March 16, 2007 and most recently amended on December 29, 2018 (effective on the same day), and the Regulation on the Implementation of the Enterprise Income Tax Law of the PRC (《中華人民共和國企業所得稅法實施條例》) promulgated by the State Council on December 6, 2007 and most recently amended on December 6, 2024 (effective on January 20, 2025), enterprises are classified into "resident enterprises" and "non-resident enterprises." A resident enterprise refers to an enterprise that is established inside China, or which is established under the law of a foreign country (region) but whose actual office of management is inside China. A non-resident enterprise refers to an enterprise established under the law of a foreign country (region), whose actual institution of management is not inside China but which has offices or establishments inside China; or which does not have any offices or establishments inside China but has incomes sourced in China. Resident enterprises are subject to enterprise income tax at a rate of 25% on their worldwide income. The enterprise income tax on a small meagre-profit enterprise that meets the prescribed conditions shall be levied at a reduced tax rate of 20%. The enterprise income tax on important high- and new-tech enterprises that are necessary to be supported by the Chinese government shall be levied at the reduced tax rate of 15%.

– 185 –

REGULATORY OVERVIEW

Income Tax on Dividend Distribution

Pursuant to the Enterprise Income Tax Law of the PRC and its implementation regulations, dividends paid by foreign-invested enterprises in China to foreign investors that are classified as non-resident enterprises, and that arise on or after January 1, 2008, are generally subject to a withholding income tax at a rate of 10%, unless otherwise provided in a tax treaty entered into between China and the jurisdiction in which the foreign investor is resident.

According to the Arrangement between the Mainland of China and the Hong Kong Special Administrative Region for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with respect to Taxes on Income (《內地和香港特別行政區關於對所得避免雙重徵稅和防止偷漏稅的安排》), promulgated by the State Taxation Administration on August 21, 2006 and became effective on December 8, 2006, dividends paid to a Hong Kong enterprise that directly holds no less than 25% of the equity in a PRC company shall be subject to a withholding tax rate of 5%; otherwise, a 10% withholding income tax shall apply.

Pursuant to the Notice of the State Taxation Administration on the Issues concerning the Application of the Dividend Clauses of Tax Agreements (《國家稅務總局關於執行稅收協定股息條款有關問題的通知》), promulgated and became effective on February 20, 2009, if a transaction or arrangement is mainly for the purpose of obtaining preferential tax treatment, such transaction or arrangement shall not be deemed as a valid basis for applying the preferential provisions of the dividend clause in a tax agreement. Where a taxpayer improperly enjoys treaty benefits due to such transaction or arrangement, the competent tax authority is entitled to adjust the preferential tax treatment. According to the Announcement of the State Taxation Administration on Relevant Issues Concerning the "Beneficial Owner" in Tax Treaties (《國家稅務總局關於稅收協定中"受益所有人"有關問題的公告》), promulgated on February 3, 2018 and became effective on April 1, 2018, the determination of whether an applicant qualifies as a "beneficial owner" under a tax treaty will be made based on a comprehensive analysis of the actual circumstances of the case. This includes, but is not limited to, whether the applicant is obligated to pay over 50% of its income within twelve months to residents of a third country or region, whether the applicant engages in substantive business activities, and whether the counterpart jurisdiction under the tax treaty exempts or applies a minimal tax on the relevant income.

Under the Administrative Measures on Entitlement of Non-resident Taxpayers to Preferential Treatment under Tax Treaties (《非居民納稅人享受協定待遇管理辦法》), promulgated by the State Taxation Administration on October 14, 2019 and became effective on January 1, 2020, non-resident taxpayers may claim tax treaty benefits based on a "self-assessment, declaration, and retention of relevant materials for future inspection" approach. Non-resident taxpayers who, upon self-assessment, determine that they meet the conditions for enjoying treaty benefits, may claim such benefits at the time of tax filing or through the withholding agent at the time of withholding declaration. They shall collect and retain the relevant data in accordance with the regulations for potential future inspection, and shall be subject to subsequent administrative oversight by the tax authorities.

– 186 –

REGULATORY OVERVIEW

Value-Added Tax (VAT)

Pursuant to the Provisional Regulations of the PRC on Value-added Tax (《中華人民共和國增值稅暫行條例》), promulgated by the State Council on December 13, 1993 and last amended on November 19, 2017 (effective on the same date), and the Detailed Rules for the Implementation of the Provisional Regulations of the PRC on Value-added Tax (《中華人民共和國增值稅暫行條例實施細則》), promulgated by the Ministry of Finance on December 25, 1993 and last amended on October 28, 2011 (effective on November 1, 2011), all entities and individuals engaged in the sale of goods, provision of processing, repair, and replacement services, as well as the provision of services, the sale of intangible assets or real estate, or the importation of goods within the territory of China are required to pay value-added tax ("VAT"). Unless otherwise stipulated, taxpayers providing services or selling intangible assets shall be subject to a VAT rate of 6%.

Pursuant to the Notice on the Comprehensive Roll-out of the Pilot Program for Replacing Business Tax with Value-added Tax (《關於全面推開營業稅改徵增值稅試點的通知》) (Cai Shui [2016] No. 36), jointly promulgated by the Ministry of Finance and the State Taxation Administration on March 23, 2016 and amended respectively on July 11, 2017 (effective on May 1, 2016), and as approved by the State Council, the pilot program for replacing business tax with VAT has been implemented nationwide since May 1, 2016. Taxpayers previously subject to business tax in industries such as construction, real estate, finance, and lifestyle services have been included in the pilot scope and are now required to pay VAT instead of business tax. Pursuant to the Notice on Relevant Policies Regarding the Simplification of VAT Rates (《關於簡並增值稅稅率有關政策的通知》) (Cai Shui [2017] No. 37), jointly promulgated by the Ministry of Finance and the State Taxation Administration on April 28, 2017 and became effective on July 1, 2017, the VAT rate structure has been simplified by eliminating the 13% VAT rate since July 1, 2017. The notice also clarified the categories of goods subject to the 11% VAT rate and the rules for deducting input VAT.

Pursuant to the Notice of the Ministry of Finance and the State Taxation Administration on Adjusting VAT Rates (《財政部、國家稅務總局關於調整增值稅稅率的通知》) (Cai Shui [2018] No. 32), jointly promulgated by the Ministry of Finance and the State Taxation Administration on April 4, 2018 and became effective on May 1, 2018, since May 1, 2018, for taxable sales or importation of goods previously subject to VAT rates of 17% and 11%, the rates were adjusted to 16% and 10%, respectively.

Pursuant to the Announcement on Relevant Policies for Deepening the Value-Added Tax Reform (《關於深化增值稅改革有關政策的公告》) (Announcement [2019] No. 39 of the Ministry of Finance, the State Taxation Administration and the General Administration of Customs), which was jointly promulgated by the Ministry of Finance, the State Taxation Administration and the General Administration of Customs on March 20, 2019 and became effective on April 1, 2019, for general VAT taxpayers engaging in taxable sales or importation of goods, the previous 16% VAT rate was reduced to 13%, and the previous 10% VAT rate was reduced to 9%.

– 187 –

REGULATORY OVERVIEW

Regulations Relating to Foreign Exchange

The primary regulation governing foreign exchange in China is the Regulation of the PRC on Foreign Exchange Administration (《中华人民共和国外汇管理条例》), promulgated by the State Council on January 29, 1996 and last amended on August 5, 2008. Pursuant to these regulations and other applicable rules and regulations on currency exchange in China, Renminbi is generally freely convertible for current account transactions (such as foreign exchange transactions involving trade and services, and dividend payments). However, Renminbi may not be freely convertible for capital account transactions (such as outbound direct investment, loans, or securities investments) without prior approval from the SAFE or its local branches.

Pursuant to the Notice of the SAFE on Issues Concerning the Foreign Exchange Administration of Overseas Listing (《国家外汇管理局关于境外上市外汇管理有关问题的通知》), promulgated by SAFE on December 26, 2014, domestic companies must complete overseas listing registration with the local SAFE office at their place of registration within 15 business days from the closing date of their overseas offering. The proceeds raised by domestic companies through overseas listings may be remitted back to China or retained offshore, and their usage must be consistent with the contents disclosed in the offering documents and other public disclosures.

According to the Guidelines for the Foreign Exchange Business under the Capital Account (2024) (《资本项目外汇业务指引(2024年版)》), issued by SAFE on April 3, 2024 and became effective on May 6, 2024, in principle, the proceeds raised by domestic companies through overseas listings shall be remitted back to China in a timely manner, either in Renminbi or in foreign currency. The use of such proceeds shall be consistent with the relevant contents as disclosed in the prospectus or corporate bond offering documents, shareholder circulars, resolutions of the board of directors or shareholders' general meeting, and other public disclosures. If domestic companies use overseas listing proceeds to conduct outbound direct investments, offshore securities investments, offshore lending, or other related activities, they must comply with the relevant foreign exchange regulations.

On May 11, 2013, the SAFE promulgated the Circular on Promulgation of the Provisions on Foreign Exchange Control on Direct Investments in China by Foreign Investors and Supporting Documents (《国家外汇管理局关于印发<外国投资者境内直接投资外汇管理规定>及配套文件的通知》), which clearly stipulates that SAFE and its local branches shall implement a registration-based management system for direct investment in China by foreign investors. Institutions and individuals conducting direct investment in China must register with SAFE or its local branches. Banks shall handle relevant direct investment transactions in China based on the registration information provided by SAFE.

On February 13, 2015, SAFE promulgated the Circular of the SAFE on Further Simplifying and Improving the Policy on Foreign Exchange Management of Direct Investment (《国家外汇管理局关于进一步简化和改进直接投资外汇管理政策的通知》), which was amended on December 30, 2019. This circular allows entities and individuals to apply for

– 188 –

REGULATORY OVERVIEW

外汇登记手续可通过具备资质的银行办理。在国家外汇管理局的监管下，具备资质的银行可直接审查并批准相关申请。2015年3月30日，国家外汇管理局颁布《国家外汇管理局关于改革外商投资企业外汇资本金结汇管理方式的通知》，规定外商投资企业的外汇资本金实行意愿结汇。经核实相关文件后，外商投资企业可根据经营需要自主办理外汇资本金结汇。该通知强调，意愿结汇须遵循真实性原则，且结汇资金须在企业经营范围内自用。资金不得用于企业经营范围以外的支出、证券投资（另有规定的除外）、人民币委托贷款、企业间借贷或房地产相关支出（外商投资房地产企业自用除外）。

2017年1月26日，国家外汇管理局发布《国家外汇管理局关于进一步推进外汇管理改革完善真实合规性审核的通知》，就境内主体向境外主体汇出利润规定了若干资本管制措施，包括：（i）银行须根据真实交易原则，对与利润汇出相关的股东分红决议、原始税务申报表及经审计财务报表进行审查；（ii）境内机构须依法弥补以前年度累计亏损后，方可汇出利润。此外，根据同一通知，境内机构在办理相关对外投资登记手续时，须对投资资金的来源和用途作出详细说明，并提交股东决议、合同及其他证明文件。

2019年10月23日，国家外汇管理局发布《国家外汇管理局关于进一步促进跨境贸易投资便利化的通知》，该通知于2023年12月4日经修订。该通知规定，所有外商投资企业可将外汇资本金结汇所得人民币用于境内股权投资，但须确保股权投资真实合规、不违反相关法律法规，且符合外商投资准入负面清单的规定。

根据国家外汇管理局发布并于2023年12月4日起施行的《国家外汇管理局关于进一步深化改革促进跨境贸易投资便利化的通知》，境内机构向境内股权转让方（包括机构和个人）支付股权转让对价所需的外汇资金，以及境内企业境外上市募集的外汇资金，可直接存入资本项目结算账户。资本项目结算账户内的资金可由主体自主结汇使用。

– 189 –

REGULATORY OVERVIEW

根据国家外汇管理局发布并于2020年4月10日起施行的《国家外汇管理局关于优化外汇管理支持涉外业务发展的通知》，资本项目收入支付便利化改革在全国范围内推广。在资金使用真实合规、符合资本项目收入使用管理现行规定的前提下，满足规定条件的企业可将资本金、外债、境外上市等资本项目收入用于境内支付，无需事前逐笔向银行提供真实性证明材料。

新加坡法律法规

截至最后实际可行日期，本公司在新加坡设有2家主要子公司，即Subsup Pte. Ltd.及Nanonoble Pte. Ltd.（"新加坡子公司"），该两家公司在新加坡注册成立，须遵守新加坡的监管要求。新加坡子公司除须遵守适用于在新加坡注册成立及/或经营的公司的一般性法规外，不受任何特别立法或监管规定的约束。

1967年《公司法》

1967年《公司法》（"《公司法》"）是规管新加坡所有注册公司的主要立法。由于新加坡子公司均为股份有限私人公司，其受《公司法》及相关法规的条款规管。此外，新加坡子公司的股东亦须遵守并受各自公司章程条款的约束。

根据《公司法》，注册成立私人公司须满足以下条件：

• 向新加坡会计与企业管理局预留公司名称；

• 委任至少1名通常居住于新加坡的董事；

• 股东最少1名，最多不超过50名；

• 除获豁免外，须于注册成立后3个月内委任会计实体担任审计师；

• 须于注册成立后6个月内委任具备资格的公司秘书；

• 最低股本为1新加坡元；

• 提供本地地址作为私人公司的注册地址；及

– 190 –

REGULATORY OVERVIEW

• 为私人公司制定章程细则。

REGULATORY OVERVIEW

Personal Data Protection Act 2012

The Personal Data Protection Act 2012 of Singapore ("PDPA") governs the collection, use and disclosure of personal data by organisations. For the purposes of the PDPA, "personal data" refers to data, whether true or not, about an individual who can be identified using that data, or from that data and other information to which the organisation has or is likely to have access to.

The obligations that the PDPA imposes on organisations collecting, using or disclosing personal data of individuals ("relevant persons") are summarised as follows: (i) obligations to obtain consent, provide notification, and offer access and correction rights to relevant persons, (ii) limitations on the purpose for which personal data may be used, (iii) limitations on the retention and transfer of personal data, and (iv) requirements to ensure the accuracy and protection of data collected, as well as transparency in making information available through privacy policies and procedures.

Further to the above, in compliance with the PDPA, each company must designate at least 1 data protection officer (DPO) and the DPO's contact information must be made available to the public.

Corporate Income Tax

Corporate taxpayers (both resident and non-resident) are subject to Singapore income tax on income accrued in or derived from Singapore (i.e. Singapore-sourced) and, subject to certain exceptions, on income received in Singapore from outside Singapore (i.e. foreign-sourced income received or deemed received in Singapore) unless specifically exempt from income tax.

Foreign-sourced income in the form of branch profits, dividends and service fee income received or deemed received in Singapore by a Singapore tax resident company on or after June 1, 2003 are exempted from Singapore tax provided that the following qualifying conditions are met:

• such income is subject to tax of a similar character to income tax under the law of the territory from which such income is received;

• at the time the income is received in Singapore, the highest rate of tax of a similar character to income tax (by whatever name called) levied under the law of the territory from which the income is received is at least 15.0%; and

• the Comptroller of Income Tax is satisfied that the tax exemption would be beneficial to the recipient of the specified foreign income.

– 191 –

REGULATORY OVERVIEW

The prevailing corporate income tax rate in Singapore is 17.0%, which applies to both local and foreign companies. With effect from the year of assessment 2020, 75.0% of the first S$10,000, and 50.0% of the next S$190,000 of a company's chargeable income (otherwise subject to normal taxation) is exempt from corporate tax. The remaining chargeable income that exceeds S$200,000 will be fully taxable at the prevailing corporate tax rate.

For the year of assessment 2024, corporate taxpayers were entitled to corporate income tax rebates of 50.0% of the corporate tax payable (which were capped at S$40,000 less the corporate income tax rebate cash grant of $2,000 where applicable.) To be applicable for the rebate cash grant, a company must be active and have at least one local employee. The corporate income tax rebate will apply to income taxed at a concessionary tax rate but will not apply to income that is subject to a final withholding tax. Similarly, for the year of assessment 2025, a corporate income tax rebate of 50% of the corporate tax payable will be granted to all taxpaying companies, whether tax resident or not, with a rebate cash grant of $2,000 where applicable. As such, the total maximum benefits of corporate income tax rebate and rebate cash grant that a company may receive is $40,000.

A company is regarded as a tax resident in Singapore if the control and management of its business is exercised in Singapore. Control and management is defined as the making of decisions on strategic matters, such as those concerning the company's policy and strategy. Generally, the location of the company's board of directors meetings where strategic decisions are made determines where the control and management is exercised. However, under certain scenarios, holding board meetings in Singapore may not be sufficient and other factors will be considered to determine if the control and management of the business is indeed exercised in Singapore. The place of incorporation of a company is not necessarily indicative of the tax residency of a company.

Goods and Services Tax

The Goods and Services Tax in Singapore is a consumption tax that is levied on import of goods into Singapore, as well as nearly all supplies of goods and services in Singapore at a prevailing rate of 9.0%.

Other Taxes

Singapore does not currently impose withholding tax on dividends paid to resident or non-resident shareholders. Dividends payable by the Singapore Subsidiaries to its shareholders are exempt from Singapore income tax in the hands of the shareholders.

Further, there is also no tax on capital gains in Singapore. Thus, any gains derived from the disposal of our shares acquired for long-term investment will not be taxable in Singapore.

Foreign shareholders are advised to consult their own tax advisers to take into account the tax laws of their respective home countries/countries of residence and the applicability of any double taxation agreement which their country of residence may have with Singapore.

– 192 –

REGULATORY OVERVIEW Regulations on Anti-Money Laundering and Prevention of Terrorism Financing The primary anti-money laundering legislation in Singapore is the Corruption, Drug Trafficking and Other Serious Crimes (Confiscation of Benefits) Act 1992 of Singapore (the "CDSA") provides for the confiscation of benefits derived from, and to combat, corruption, drug dealing and other serious crimes. Generally, the CDSA criminalizes the concealment or transfer of the benefits of criminal conduct as well as the knowing assistance of the concealment, transfer or retention of such benefits. The Terrorism (Suppression of Financing) Act 2002 of Singapore (the "TSOFA"), is the primary legislation for the combating of terrorism financing. It was enacted to give effect to the International Convention for the Suppression of the Financing of Terrorism. Besides criminalizing the laundering of proceeds derived from drug dealing and other serious crimes and terrorism financing, the CDSA and the TSOFA also require suspicious transaction reports to be lodged with the Suspicious Transaction Reporting Office. If any person fails to lodge the requisite reports under the CDSA and the TSOFA, it may be subject to criminal liability. LAWS AND REGULATIONS IN THE UNITED STATES Regulations on Artificial Intelligence Technologies Although there are several private and public initiatives and organizations calling for regulations on AI technologies, including but not limited to the development of AI functionalities and the implementation of AI technology into another object or technology, as of the Latest Practicable Date, there is no unified federal law or regulation in the United States yet that was specifically adopted to govern AI technologies comprehensively. At the moment, AI-targeted, AI-based, or AI-related businesses are primarily regulated by the laws and regulations that apply to all types of technologies, products and services. For example, where AI system development and solution businesses involve software coding, they may be associated with concerns of copyright, privacy protection, and export controls. Other specific legal doctrines may have direct or indirect implications on AI operations. Common law doctrines in tort claims, for instance, raises questions about, including but not limited to, negligence, duty of care, and product liability. AI-related businesses might be held liable under tort law doctrines if they fail to exercise a reasonable standard of care in the design, manufacturing, or warning instructions for the product. Furthermore, AI-related businesses may also find themselves under common law doctrines in contract claims, particularly when statements or promises are made, with legal doctrines such as promissory estoppel serving as a potential safety net. In the absence of comprehensive federal legislation and regulation, individual states have taken initiative to regulate AI technologies within their jurisdictions. California has emerged as a leader in this space with Senate Bill 942, the California AI Transparency Act, and Senate Bill 2013, the California AI Training Data Transparency Act, both of which will become effective January 1, 2026. SB 942 mandates that "Covered Providers" — AI systems publicly accessible within California with more than one million monthly visitors or users — implement – 193 –

监管概述

采取全面措施披露人工智能生成或修改的内容。该法案还对人工智能检测工具、内容披露及许可实践提出了要求，违规者每日将面临5,000美元的罚款。SB 2013要求生成式人工智能系统的开发者发布有关用于训练系统的数据的详细信息，包括数据来源、数据集是否包含受知识产权保护的数据，以及开发者对数据集所做的修改。科罗拉多州凭借《科罗拉多人工智能法案》（CAIA）也跻身领先地位，该法案将于2026年2月1日生效。CAIA为"高风险人工智能系统"——即在部署时，在特定情境下作出或对作出重大决定具有实质性影响的人工智能系统——的开发者和部署者设定了应对算法歧视的义务，并规定了包括风险评估和信息披露在内的要求，以证明该义务已得到履行。其他州则通过了更具针对性的法规，主要聚焦于人工智能聊天机器人，着重确保用户了解其并非在与人类交流，并要求采取措施应对用户向旨在模拟人类互动的人工智能聊天机器人表达自我伤害意图的情形。

此外，尽管并非专门针对人工智能，加州《隐私权法案》（CPRA）和《科罗拉多隐私法》（CPA）等州隐私法正在整合人工智能相关条款。上述法规赋予消费者选择退出某些人工智能驱动的用户画像的权利，并对在决策具有法律效力或类似重大影响的情境中的自动化决策过程予以审查。企业可能还须针对人工智能实践开展数据隐私影响评估，尤其是在此类实践对消费者数据隐私存在重大风险的情况下。值得注意的是，并非每部州隐私法都对人工智能的复杂性进行深入探讨，这表明监管格局呈现多元化且仍处于演变之中。值得关注的是，随着人工智能迅速渗透至几乎每个行业，各国政府正逐步将人工智能纳入监管范畴。在联邦层面，以人工智能为重点的法案已被提交国会，但尚未获得通过。然而，人工智能监管似乎正在从美国联邦贸易委员会（FTC）层面逐步浮现。近年来，FTC发布了两份预示加强人工智能监管的文件，开始为人工智能的开发和使用制定基本规则，例如规定人工智能训练标准及部署前测试要求，以及建立问责和治理机制以记录公平负责任的人工智能开发、部署和使用情况。与此同时，FTC在现行法规框架下加大了人工智能执法力度，适用法规包括《公平信用报告法》、《儿童网络隐私保护法》及《联邦贸易委员会法》。

美国出口管制及经济制裁法规

出口管制法规受美国联邦法律管辖，主要包括《出口管理条例》（"EAR"）和《国际武器交通条例》（"ITAR"）。EAR由商务部工业和安全局（BIS）负责实施。EAR适用于商业、两用及特定军事硬件、软件和技术的出口、再出口及转让。受EAR管辖的硬件、软件和技术包括：美国原产物项、在美国境外制造但含有超过最低限度的受管制美国原产内容的特定物项，以及特定美国软件和技术的境外直接产品。根据硬件、软件或技术的性质、目的地国家、最终用途及最终用户的不同，

— 194 —

REGULATORY OVERVIEW

出口管制及制裁法规规定，受EAR管辖的物项在出口、再出口或转让前可能需要事先获得授权。ITAR由国务院国防贸易管制局（DDTC）负责实施。上述法规适用于国防物品、国防服务及相关技术数据的出口、再出口、转让、临时进口及中介活动。受ITAR管辖的所有出口、再出口、转让及临时进口均须事先获得授权。

财政部海外资产控制办公室（"OFAC"）负责对特定国家、实体及个人实施美国经济制裁。由于经济制裁旨在推进美国的对外政策目标，各项制裁计划之间存在较大差异。美国制裁计划通常适用于特定制裁计划中定义的"美国人"，以及其他与美国存在关联的交易。

美国出口管制及经济制裁依据严格责任原则予以执行。违反美国出口管制及经济制裁规定可能导致巨额民事罚款，以及针对故意违规行为的刑事处罚和/或监禁。

**美国对外投资审查计划**

2023年8月9日，美国总统拜登签发了一项行政令，其政府同时发布了一份预先通知拟制规则（ANPRM），为针对中国（包括香港和澳门）的对外投资管控提供了概念性框架。在此ANPRM的基础上，2024年6月21日，美国财政部发布了关于涉及中国的美国对外投资的拟议规则，该规则总体上与ANPRM保持一致。2024年10月28日，美国财政部发布了最终规则，并于2025年1月2日正式生效。

最终规则对美国人在从事以下三个领域活动的与中国（包括香港和澳门）相关联的实体（统称为"受涵盖外国人"）中进行的各类投资，规定了投资禁止和通知要求：（i）半导体及微电子、（ii）量子信息技术，以及（iii）人工智能系统。受最终规则约束的美国人被禁止对受涵盖外国人进行特定投资，或须就相关投资进行申报，上述投资被定义为"受涵盖交易"，包括收购股权权益（包括或有股权权益）、特定债务融资、合资企业，以及以有限合伙人身份对非美国人汇集投资基金进行的特定投资。最终规则将部分投资排除在受涵盖交易范围之外，包括对特定上市证券的投资（例如，本公司完成全球发售后的上市证券）。

2025年12月18日，美国总统特朗普签署颁布了《2026财年国防授权法案》，其中包含《2025年综合对外投资国家安全法》（"COINS法案"）。最终规则仍继续有效，但COINS法案要求财政部在2025年12月18日起450天内就最终规则提出若干修订。上述修订最终将包括（但不限于）以下内容：扩大关注国家范围、将覆盖技术扩展至高超音速系统、修订关键定义术语，以及建立正式的咨询意见程序。

– 195 –

REGULATORY OVERVIEW

**知识产权法**

美国同时具备联邦和州层面的知识产权法律。部分知识产权仅受联邦法律管辖，其他知识产权则同时受联邦法律和州法律管辖。

**受联邦法律管辖的知识产权**

版权和专利权专门受联邦法律管辖。

**版权**

版权是固定于有形载体中的原创作品创作者所享有的一系列专有权利。版权（i）保护创意表达，而非思想；（ii）不得纯属功能性内容；（iii）须为原创作品。美国版权法由1976年《版权法》规范，编纂于美国法典第17编第101条及以下条款。根据美国版权法，固定于有形媒介中的原创作品自创作完成之时起自动享有版权保护（《美国法典》第17编第102条）。版权所有人享有一系列专有权利（例如，复制权、创作衍生作品权、发行权、公开展示权等）（《美国法典》第17编第106条）。上述权利如无法定抗辩事由（如"合理使用"）适用，则可能构成侵权。如最终认定存在侵权责任，根据《美国法典》第17编第504条，版权所有人可获赔实际损失，加上人工智能公司因侵权所获得的、未包含在实际损失中的利润。此外，版权所有人亦可选择要求赔偿法定损害赔偿金。

**专利**

专利是政府授予的权利，赋予专利所有人在美国境内排除他人制造、使用、销售、要约销售或进口受保护发明，或在美国境内实施受保护方法的权利。专利通过向美国专利商标局（"USPTO"）提交申请而获得，申请须主张具有实用性、新颖性或非显而易见性的发明。申请须符合《专利法》（编纂于美国法典第35编第1条及以下条款）及USPTO（美国商务部下属机构）所制定的各项规定中载明的多项要求。专利所有人可向美国联邦法院提起侵权诉讼，或在涉及进口侵权商品的情形下，向国际贸易委员会提起诉讼。专利所有人可就侵权方获得相应救济，包括初步禁令及永久禁令、直接损害赔偿（包括利润损失或特许权使用费），以及在例外情形下的三倍赔偿和律师费。

– 196 –

REGULATORY OVERVIEW Intellectual Property Rights Governed by both Federal and State Law Trademarks and service marks A "mark" is the use of one or more words, symbols, or logos to identify and distinguish the mark owner's goods and/or services. A trademark is a mark used for goods; a service mark is a mark used in connection with providing services. U.S. trademarks and service marks generally must be used as a source identifier and (i) not be confusingly similar to prior marks when considered in connection with the goods or services with which they are used, (ii) not be generic, and (iii) not be merely descriptive. U.S. federal trademark law is governed by the Lanham Act, codified at 15 U.S.C. § 1051 et seq. The USPTO is responsible for examining trademark and service mark applications and either granting or rejecting applications to register marks. Once granted, a trademark or service mark registration provides its owner with nationwide exclusivity within one or more particular fields of use. State law is an alternative basis for trademark and service mark rights, either under specific state laws or under common law. States generally provide common law rights in trademarks and service marks upon their first use in commerce, without requiring registration. Some states have registries for trademarks and service marks. The rights inherent in such marks are limited to the state(s) where they are used. The owner of a trademark generally has a cause of action for infringement against a defendant who uses a mark that is likely to cause confusion in the relevant marketplace about the source of goods or services, or likely to cause consumers to falsely infer some association or affiliation between the trademark owner and the defendant. A plaintiff may be entitled to preliminary and permanent injunctions (including destruction of infringing articles), actual monetary damages, accounting of the defendant's profits, and in some cases, attorneys' fees. Trade secrets A trade secret is information that (i) has independent economic value from being generally unknown by the public and (ii) is the subject of reasonable efforts under the circumstances to maintain its secrecy. Trade secrets are governed by both federal and state law. The Defend Trade Secrets Act, codified at 18 U.S.C. § 1836, et seq. ("DTSA"), is the federal trade secret law. Enacted recently in 2016, the DTSA applies only to trade secrets used in interstate or foreign commerce. The DTSA provides specific remedies for trade secret misappropriation, including ex parte seizure in specific and generally rare instances. The DTSA is similar to the Uniform Trade Secret Act ("UTSA"), a model set of laws enacted by almost all fifty states within the U.S. A trade secret owner may often have a choice in enforcing its trade secret rights under the DTSA or a relevant state's version of the UTSA.

– 197 –

REGULATORY OVERVIEW United States Data Privacy and Security Laws and Regulations The U.S. does not have a comprehensive federal law that governs data privacy or data security. Instead, the U.S. has a complex patchwork of sector-specific data privacy and data security laws and regulations at the federal level, and sector-specific data and general privacy and data security laws and regulations at the state level. States have also enacted data breach notification laws, which generally require entities to notify affected customers of a data breach. Data Privacy Laws In the United States, approximately 20 states currently maintain comprehensive data privacy laws. Under these laws, companies must give consumers certain notices about how their personal data is collected, used, and disclosed, and must offer meaningful choices related to consumers' personal data (such as opt-in or opt-out rights, and rights of access, deletion, and correction, among others). Some of these laws also contain prescriptive privacy policy disclosure requirements (with the most prescriptive being in California under the California Consumer Privacy Act or "CCPA"). U.S. state privacy laws also generally require companies to limit information processing to the purposes and methods disclosed in their privacy notices or that are reasonably expected by consumers, and must implement safeguards that are appropriate to the risks presented by processing the specific types of personal data. In certain situations under these state laws, additional obligations apply — for example, when handling data classified as "sensitive," when engaging in processing activities that pose heightened risks to children (which may trigger a data protection impact assessment and additional consent requirements), or when relying on external processors or other third parties to carry out data processing on the company's behalf. There is no requirement to maintain a Cookie Notice in the U.S. Enforcement under the U.S. state laws generally occurs at the Attorney General level, but notably in California, both the State AG as well as the separately formed California Privacy Protection Agency ("CPPA") can enforce state privacy laws. Penalties for non-compliance with U.S. state data protection laws vary depending on state, but typically range from $2,500–$7,500 per violation, depending on whether the violation is deemed to be negligent or intentional/knowing. U.S. state regulators also have the authority to issue injunctive relief, which can include required deletion of data/accounts collected out of compliance, the implementation of required and prescriptive privacy controls that often go above and beyond actual legal requirements, and regulatory oversight (including the delivery of detailed, yearly compliance reports) for up to 20 years. In addition to enforcing specific privacy laws, U.S. state Attorneys General also have authority to issue injunctive and monetary penalties under state unfair and deceptive acts and practices laws.

– 198 –

REGULATORY OVERVIEW

在联邦层面，隐私法具有行业性质，目前联邦隐私法分别规范金融数据（《格雷姆-里奇-比利雷法案》，即"GLBA"）、儿童数据（《儿童在线隐私保护法》，即"COPPA"）、信用评估数据（《公平信用报告法》，即"FCRA"）以及健康数据（《健康保险流通与责任法案》，即"HIPAA"）。此外，美国联邦贸易委员会（FTC）依据《联邦贸易委员会法》第5条，对从事不公平或欺骗性行为及实践的公司享有广泛的执法权力。依据第5条，联邦贸易委员会可对未能履行保护消费者个人信息承诺的组织提起法律诉讼，例如未能遵守其发布的隐私政策，而该政策载有与公平信息实践原则相关的数据隐私声明，如通知、同意或控制权（例如，选择退出第三方数据共享的权利）。

除联邦贸易委员会执法先例外，对本公司最为适用的联邦法律是《联邦儿童在线隐私保护法》的潜在适用，该法为严格责任法规，每次违规最高可处罚款53,000美元。COPPA要求面向13岁以下儿童提供在线服务的公司提供特定通知并获得父母同意，同时限制公司处理儿童数据的方式。即便是并非面向儿童的在线服务，也有义务以中立方式向用户询问年龄信息，若其实际知悉已在未获父母同意的情况下收集了13岁以下儿童的个人数据，则必须采取措施立即从其系统中删除该信息。联邦贸易委员会已就面向年龄较大或更广泛受众的公司因未能维持中立年龄验证机制而提起多项执法行动。

数据安全法律

美国目前尚无联邦网络安全法，取而代之的是54部独立的州数据泄露法（包括哥伦比亚特区、关岛、波多黎各及美属维尔京群岛）。这些法律规范遭遇数据泄露的实体应如何应对。每部数据泄露通知法对个人信息的定义、豁免情形以及向受影响客户、司法部长及州监管机构发出通知的义务均有不同规定。这些数据泄露通知法通常将个人信息的定义限定为个人的名字或名字首字母与姓氏的组合，并结合以下一项或多项信息：社会安全号码、驾驶执照号码或州身份证号码，或账号、信用卡或借记卡号码与安全码、访问码或密码的组合。约三分之二的州将个人信息的定义扩展至包含更多要素，约三分之一的州赋予因其个人信息遭披露而受损的个人私人诉权。此外，加利福尼亚州的一般隐私法（《加州消费者隐私法》）将数据泄露的私人诉权纳入其中，该泄露依据州数据泄露法加以定义。各州泄露法规定的处罚幅度差异较大，从每次违规500美元至5,000美元不等。

– 199 –

REGULATORY OVERVIEW

在联邦层面，联邦贸易委员会有权对发生数据泄露的公司提起执法行动，其理论依据为：发生泄露的公司可能以不公平方式处理了消费者数据，从而导致泄露事件的发生。在泄露调查中，联邦贸易委员会通常会评估公司为保护个人信息而采取的政策和程序是否合理。

除泄露报告法律外，美国约20个州制定了数据安全法。这些法律对公司维持"合理"安全实践规定了不同要求，包括根据个人信息的敏感程度采取适当的技术、组织及物理控制措施，以及维护书面信息安全政策的要求。相关主管机构并不规定具体的数据安全标准，而仅发布数据安全"最佳实践"或建议达到或超越相关行业公认标准，如NIST 27003或SOC 2。主管机构可对相对于同行业其他实体而言未能实施合理数据安全措施的实体启动执法程序——相关标准依据市场对安全实践的采纳程度及技术变化而发展演进。数据安全违规行为可因以下情形而产生，例如未能履行与数据相关的安全声明，或未能根据所存储数据的性质实施合理的商业安全程序。

在联邦层面，COPPA、HIPAA及GLBA项下亦存在一般合理安全义务，联邦贸易委员会同样可就上述义务予以执行。总体而言，涉及儿童数据、金融数据和/或健康数据的数据泄露事件往往会导致最高额的财务处罚。

– 200 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE OVERVIEW Our Company was incorporated in the Cayman Islands as an exempted company with limited liability on June 30, 2021. It is common for pure-play foundation AI model companies to spend several years on research and development before generating model-based revenue. From the start of January 2022, we have been in operation with R&D activities commenced and R&D as well as administrative expenses incurred and recorded. The trading activities at the very beginning of the Track Record Period primarily consisted of R&D activities and administration of the Group such as procuring technical services from vendors regarding data processing and conducting collaborative research and development projects. For example, with over ten R&D staffs, we spent R&D expenses amounting to more than US$80 thousand and administrative expenses amounting to more than US$18 thousand for the month ended January 31, 2022. In April 2022, our first text model abab 1 was launched for internal use of our own products, placing it among the earliest global leaders in accessible AI. Building on this foundation, it expanded into speech capabilities with the launch of Speech-01 in November 2023 and further into video generation capabilities with Video-01 in August 2024. In June 2022, we completed independent R&D and training of the abab 2 model. The model demonstrated dialogue and question-answering capabilities, achieving fluent and coherent intelligent conversations. It also exhibited logical reasoning abilities, establishing it as a technologically advanced and effective text-based large model at the time. Shortly after launching abab 2, we increased the parameter count for the abab 3 model and completed its training in October 2022. The abab 3 model significantly outperformed its predecessor in key capabilities, including language comprehension, multi-turn dialogue, and content generation. Building on this foundation, we launched the abab 4 model, which optimized reasoning speed, computational resource utilization, and deployment costs. In October 2022, we also launched an experimental intelligent dialogue product that enables users to engage in multi-round conversations with relevant AI themes incorporating emotional interaction. By the end of 2022, this product had reached a user base of approximately 100,000. The aforementioned work conducted through abab 1, abab 2, abab 3 and abab 4 models served as the foundation of our further development and operations through the accumulation of data, algorithm and infrastructure knowledge, in particular with respect to our text models, which is one of the three major categories of our core models for now. During its years of development, the Company underwent rounds of Pre-IPO Investments as elaborated in the section headed "— Pre-IPO Investments" and continuously achieved milestones as described in the section headed "— Our Key Milestones" below. Under the leadership of Dr. Yan and Ms. Yun, our founders and WVR beneficiaries, we have become a global AI foundation model company after years of development. Founded in 2022 by a group of engineers, we are committed to advancing AI towards performing the full range of human intellectual tasks, from learning and reasoning to planning and generalizing knowledge across diverse domains.

– 201 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE OUR KEY MILESTONES The following is a summary of our Group's key business development milestones:

| Year | Month | Milestone | |------|-------|-----------| | 2022 | April | Our first text model abab1 was launched. | | 2023 | May | We entered into agreement with our first API customer. Our text model abab5.5 was launched. | | 2024 | June | Our first AI-native multi-modal entertainment platform Talkie was launched. | | | September | Another AI-powered multi-modal entertainment platform Xingye was launched. | | | November | Our speech model MiniMax-Speech-01 was launched. | | | January | Our MoE text model abab6 was launched. | | | April | Our MAU surpassed 10 million. | | | August | Our visual generation platform Hailuo AI and video generation model Hailuo-01 were launched. Our music model Music-01 was launched. | | 2025 | January | Our open-source text model MiniMax-Text-01 with proprietary "Linear Attention" mechanism was launched. Our audio generation tool MiniMax Audio was launched. | | | April | Our multilingual speech model Speech-02 was launched. MiniMax MCP was launched. |

– 202 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

| Year | Month | Milestone | |------|-------|-----------| | | June | MiniMax-M1, an open-source, large-scale hybrid-attention reasoning model was launched. Our video generation model Hailuo-02 was launched. Our intelligent agent application MiniMax was launched. | | | October | MiniMax-M2 was launched. The latest generation of our music synthesis model, MiniMax Music 2.0, was launched. |

OUR MAJOR SUBSIDIARIES As of the Latest Practicable Date, we had six major operating subsidiaries which had made material contributions to our financial results during the Track Record Period. All of the major operating subsidiaries are wholly owned by our Company and the corporate details of these subsidiaries are set forth as below:

| Name of subsidiary | Place of incorporation | Date of incorporation | Principal business | |--------------------|----------------------|-----------------------|--------------------| | Shanghai Jizhi (上海积志) | PRC | November 3, 2021 | Research and development of foundation models and products | | Beijing Jizhi (北京积志) | PRC | November 18, 2021 | Research and development of foundation models and products | | SUBSUP PTE. LTD. | Singapore | September 14, 2022 | Operation of AI-native products | | Shanghai MiniMax (上海MiniMax) | PRC | January 28, 2023 | Operation of Open Platform and AI-native products | | NanoNoble PTE. LTD. | Singapore | March 19, 2024 | Operation of Open Platform and AI-native products | | MiniMax HongKong | Hong Kong | April 10, 2025 | Operation of Open Platform |

– 203 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

For capital changes of our major subsidiaries with respect to the Reorganization and during the two years immediately preceding the date of this Prospectus, please refer to "— Corporate Reorganization" in this section and "Statutory and General Information — A. Further Information about Our Group — 3. Changes in the Share Capital of Our Subsidiaries" in Appendix IV to this Prospectus, respectively. Save as disclosed above, there were no capital changes in our major subsidiaries during the Track Record Period and up to the Latest Practicable Date.

MAJOR SHAREHOLDING CHANGES OF OUR COMPANY

(1) Incorporation of our Company

Our Company was incorporated on June 30, 2021 in the Cayman Islands as an exempted company with limited liability with an authorized share capital of US$50,000 divided into 50,000 shares with a par value of US$1. At the time of the incorporation of our Company, we were beneficially owned by Dr. Yan as to 98.5% and a former employee who is our Independent Third Party as to 1.5%.

(2) Pre-IPO Investments

Following the incorporation of our Company, we conducted several rounds of pre-IPO investments. For details, please refer to the sub-sections headed "Pre-IPO investments" below in this section.

(3) Our WVR structure

Dr. Yan and Ms. Yun, our executive Directors and WVR beneficiaries, were beneficially interested in the issued share capital of the Company as to approximately 46% and 5%, respectively, before we adopted the WVR structure in July 2023. Upon establishment of our WVR structure, our Company's issued shares comprise Class A Ordinary Shares, Class B Ordinary Shares and preferred Shares. Each of the Class B Ordinary Shares entitles the holders thereof to exercise ten votes and each of the Class A Ordinary Shares and preferred Shares entitles the holders thereof to exercise one vote, on any resolution tabled at our Company's general meetings, except for certain resolutions that require voting on a one vote per share basis pursuant to then existing articles of association.

Before the shareholding structure adjustment as described below, (i) 99,650,075 Class B Ordinary Shares are beneficially owned and controlled by Dr. Yan, through his controlled entities, namely MiniMax Limited, MiniMax Matrix, Alpha EXP, MiniMax Awakening, and MiniMax Gene (collectively, the "Intermediary Companies of Dr. Yan"); and (ii) 7,000,000 Class B Ordinary Shares are beneficially owned and ultimately controlled by Ms. Yun through her controlled entity, Floating Sky (the "Intermediary Company of Ms. Yun"). Dr. Yan was also interested in 343,195 Class A Ordinary Shares through his controlled entities. The preferred shares are held by our Pre-IPO Investors.

– 204 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

(4) Adjustment of our WVR Beneficiaries' Shareholding Structure

To comply with Chapter 8A of the Listing Rules, and to streamline its shareholding structure, between April 2025 and June 2025, our Company has undergone the following shareholding structure adjustment, including, among others:

(i) the shareholding structure of the Intermediary Companies of Dr. Yan was adjusted to the effect that they are ultimately beneficially owned as to 100% by Dr. Yan (other than Minimax Matrix);

(ii) the shareholding structure of the Intermediary Company of Ms. Yun was adjusted to the effect that it is ultimately beneficially owned as to 100% by Ms. Yun;

(iii) Dr. Yan holds an aggregate of 99,993,270 Class B Ordinary Shares through the Intermediary Companies of Dr. Yan; and

(iv) Ms. Yun holds an aggregate of 7,000,000 Class B Ordinary Shares through the Intermediary Company of Ms. Yun.

(5) Establishment of our employee shareholding platform

To attract, retain and incentivize selected employees, directors, and consultants of the Company and to further promote the success of the Company's business, we adopted the Pre-IPO Equity Incentive Plan. As of the date of the Latest Practicable Date, options representing an aggregate of 20,890,736 Class A Ordinary Shares were granted. No awards or options will be further granted upon or after the Listing under the Pre-IPO Equity Incentive Plan. See "Appendix IV — Statutory and General information — D. Share Incentive Plans" for details.

In September 2025, 6,509,339 Class B Ordinary Shares held by MiniMax Gene were transferred to MiniMax Awakening and MiniMax Gene became the employee shareholding platform holding a total of 20,890,736 Class A Ordinary Shares. MiniMax Gene was held by MiniMax Gene Alpha Limited, which was in turn held by MiniMax Gene Trust with Futu Trust Limited, an independent professional trust company, as its trustee and the Company as its settlor. In October 2025, MiniMax Gene transferred all shares it held in MiniMax Limited to MiniMax Awakening.

– 205 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

Upon completion of the aforesaid shareholding structure adjustment and establishment of our employee shareholding platform, the following chart sets out the shareholding structure and beneficial interests of our WVR Beneficiaries in the Company:

``` Ms. Yun Dr. Yan |100% |100% | | Apricity MiniMax Investment Awakening Limited | |100% 4.11% | Local Linearity | |67.10% 32.90% | | Floating Sky(2) MiniMax Matrix(1) 2.50% 1.79% | | Alpha EXP(2) | 0.00001% | | MiniMax Limited | 22.35% | | MiniMax Gene | 7.46% | |_______________________________________________| | The Company | Other Pre-IPO Investors 61.80% ```

Notes:

(1) Minimax Matrix was established at the early stage of the Company's development in order to hold relevant beneficial interests in the Company. There is no side arrangement in terms of MiniMax Matrix's control between Dr. Yan and Ms. Yun.

(2) For estate planning purpose and by November 2025, all the Shares controlled by Dr. Yan in Alpha EXP and by Ms. Yun in Floating Sky were held under the trusts of Dr. Yan and Ms. Yun, respectively. For details of the trusts, see note 3 to note 12 in "— Capitalization" in this section.

MAJOR ACQUISITIONS, DISPOSALS AND MERGERS We have not conducted any material acquisitions, disposals or mergers during the Track Record Period and up to the date of this Prospectus.

CORPORATE REORGANIZATION

Before the Reorganization, our Group engaged in businesses operating certain applications and websites that required the Value-added Telecommunication License for Internet Information Service (中华人民共和国增值电信业务经营许可证) ("ICP License"). According to the Negative List and the Telecommunications Regulations of the PRC (《中华人民共和国电信条例》), the provision of value-added telecommunications services falls within the restricted industries and the percentage of foreign ownership cannot exceed 50% (except for e-commerce, domestic multi-party communications, store-and-forward and call centers). In order to comply with the PRC laws and regulations and maintain an effective control over the operation of such businesses, Shanghai MiniMax, one of our current wholly owned subsidiaries, was controlled by one of our subsidiaries through contractual arrangements. Dr. Yan and Ms. Pan Lin were the registered shareholders of Shanghai MiniMax. Ms. Pan Lin is our employee and an Independent Third Party. In preparation for the Listing and to streamline our business and shareholding structure, we underwent the Reorganization involving the following steps:

– 206 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

(1) Establishment of Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng

On April 18, 2025, Shanghai Jizhi Wujie was established in the PRC with a registered share capital of RMB1 million. It is held by Dr. Yan as to 99% and Ms. Pan Lin as to 1%.

On April 23, 2025, Shanghai Jizhi Zongheng was established in the PRC as a wholly-owned subsidiary of Shanghai Jizhi Wujie, with registered share capital of RMB1 million.

Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng were established to provide services to the Group as described below with focus on different applications related to the Company's businesses. For their scale of operations in the future, please refer to the section headed "Connected Transactions — Non-exempt Continuing Connected Transaction — 2. The Business Cooperation Agreement — Annual Cap and Basis of Cap".

(2) Capital increase on Shanghai MiniMax by an independent third party

In May 2025, Dr. Yan, Ms. Pan Lin, an Independent Third Party (the "Foreign Investor") and Shanghai MiniMax entered into a capital increase agreement, pursuant to which the Foreign Investor injected RMB20,303 into the registered share capital of Shanghai MiniMax. Upon completion of the capital increase, Shanghai MiniMax was owned as to 98.5075% by Dr. Yan, 0.4925% by Ms. Pan Lin and 1% by the Foreign Investor. The amount of capital injected was determined with reference to the equity valuation of Shanghai MiniMax appraised by an independent valuer.

(3) Termination of the contractual arrangements

In May 2025, in order to streamline our shareholding and corporate structure, Dr. Yan, Ms. Pan Lin, the Foreign Investor and Shanghai Jizhi entered into equity transfer agreements, pursuant to which the entire equity interest in Shanghai MiniMax were transferred to Shanghai Jizhi at a consideration of RMB9.9021 million based on valuation from an independent appraiser. Upon completion of the transfer, Shanghai MiniMax was wholly owned by Shanghai Jizhi. In June 2025, the contractual arrangements between Shanghai Jizhi, Shanghai MiniMax, Dr. Yan and Ms. Pan Lin were terminated. In June 2025, assets including two domain names (xingyeai.com and hailuoai.com) and two software copyrights were transferred from Shanghai MiniMax to Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng at a total consideration of RMB4.36 million so as to fulfil their daily business operations and functions. After the aforementioned transfer of assets, Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng became responsible for holding the ICP Licences required for the Group's applications and websites, previously held by Shanghai MiniMax subsequent to June 2025, and they possessed all the requisite licences and qualifications in this regard in June 2025. Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng have been providing supportive technical services including but not limited to the operation and management of operational support, routine updates and maintenance, contents uploading, promotion and marketing on the abovementioned applications and websites, while the Group is primarily responsible for the operation and

– 207 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

management of the underlying and basic technology development, function development, designing and technology upgrades of the applications and websites. The Group does not rely on the supportive services as an essential part of its operations, given their availability in the market. These domain names and software copyrights are necessary for Shanghai Jizhi Wujie (上海极智无界) and Shanghai Jizhi Zongheng (上海极智纵横) to provide relevant services to the Group as well as to enable Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng to obtain the ICP License required for operation. There are no major changes in the businesses of Shanghai MiniMax prior or subsequent to June 2025. The Group jointly owned these software copyrights with Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng currently and ceased to use these domain names as relevant services can be obtained from Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng, and Shanghai MiniMax therefore also ceased to hold the ICP License in June 2025. As such, the Company was allowed to become the sole indirect shareholder of Shanghai MiniMax under applicable laws. After such transfer of assets, Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng were able to commence their daily business operations by obtaining the ICP Licenses and providing services to the Group, and the Group was also able to have a streamlined shareholding structure without contractual arrangements or restrictions on foreign investment imposed from applicable PRC laws and regulations. Despite that the domain names are necessary for Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng to hold the ICP Licenses and provide their service to the Group, their service are readily available in the market and the Company will be able to obtain similar services from independent third parties in the market on terms no less favorable than the terms from Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng if they cease to provide such services to the Group. In addition, the two domain names are not fundamental or irreplaceable for the Group's own operations. As such, the cease of cooperation from Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng, if any, will not have any material adverse impact to the Group. Therefore, the Company believes that such transfer of assets to Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng is in the best interest of the Company's shareholders. As confirmed by our PRC Legal Advisor, the Group's businesses were not subject to any regulatory restrictions on foreign investment in the PRC during the Track Record Period and up to the Latest Practicable Date, save for the operation of certain applications and websites of the Group by Shanghai MiniMax prior to the termination of the historical contractual arrangements in June 2025. Taking into consideration of the PRC Legal Advisor's view above, and based on the review of the relevant transaction documents, nothing has come to the attention of the Joint Sponsors which would reasonably cause them to disagree with the Company's view above.

For details of the services provided by Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng to the Group, please refer to the section headed "Connected Transactions — Non-exempt Continuing Connected Transaction — 2. The Business Cooperation Agreement".

Our PRC Legal Advisor has confirmed that all the equity transfers of our PRC subsidiaries as described above have been legally completed, and our Group has obtained all necessary regulatory approvals and permits and completed all necessary filings in respect of such transfers that our Group had to obtain from PRC regulatory authorities.

– 208 –

Overview

PRE-IPO INVESTMENTS

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

1.

We have received several rounds of Pre-IPO Investments since our inception. The following table summarizes the key terms of the Pre-IPO Investments to our Company made by the Pre-IPO Investors:

| | Series Angel | Series Pre-A | Series A | Series A+ | Series Pre-B | Series Pre-B+ | Series Pre-B++ | |---|---|---|---|---|---|---|---| | Date of the last share purchase agreement | Dec 2, 2021 | Mar 29, 2022 | May 4, 2023 | Jul 6, 2023 | Mar 15, 2024 | Dec 4, 2024 | August 16, 2025 | | Date of last payment of consideration | Dec 27, 2021 | Apr 11, 2022 | Dec 14, 2023 | Jul 13, 2023 | Feb 19, 2025 | Jun 20, 2025 | August 19, 2025 | | Cost per Share (US$) | $1.69 | $4.23 | $6.91 | $8.81 | $10.46 | $12.28 | $15.14 | | Discount to the Offer Price(1) | 91.7% | 79.2% | 66.0% | 56.6% | 48.5% | 39.5% | 25.4% | | Total consideration received by our Company (US$ million) | 31.0 | 50.0 | 257.0 | 50.0 | 654.0 | 123.5 | 390.4 | | Implied pre-money valuation (US$ million) | 169.0 | 500.0 | 900.0 | 1,550.0 | 1,900.0 | 3,000.0 | 3,850.0 | | Implied post-money valuation (US$ million) | 200.0 | 550.0 | 1,157.0 | 1,600.0 | 2,554.0 | 3,123.5 | 4,240.4 | | Use of proceeds from the Pre-IPO Investments | As of the Latest Practicable Date, approximately 30% of the funds raised from the Pre-IPO Investments had been utilized. Such proceeds were utilized for the research and development, capital expenditures and general working capital needs of our Group. The Company plans to utilise the remaining proceeds from the Pre-IPO Investments for cloud services procurement related to training and inferencing, human resources matters and marketing activities. ||||||||

– 209 –

– 210 –

For lock-up period of our key persons and Pathfinder SIIs pursuant to Rule 18C.14 of the Listing Rules, see the section headed "— Lock-up Periods" below. For lock-up period of our other existing Shareholders (including all the other Pre-IPO Investors), see the section headed "Underwriting — Underwriting Arrangements and Expenses — Hong Kong Public Offering — Undertaking by the other existing shareholders".

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

**2. Special rights of the Pre-IPO Investors**

The Pre-IPO Investors have been granted certain special rights in relation to our Company, including but not limited to redemption rights, the pre-emptive rights, right of co-sale, liquidation preferences, rights of first refusal, information rights and director appointment rights. Pursuant to a shareholders' agreement dated June 23, 2025, the redemption rights have been suspended immediately prior to the first filing of the listing application and all special rights (including the redemption rights) will only be terminated upon Listing.

**3. Compliance with the Guide for New Listing Applicants**

On the basis that (i) the consideration for the last Pre-IPO Investment was irrevocably settled on a date, which is more than 120 clear days before the Listing Date, and (ii) the special rights granted to the Pre-IPO Investors will be suspended immediately prior to the first filing of a listing application and/or shall cease to be effective and be discontinued upon Listing, the Joint Sponsors confirm that the Pre-IPO Investments are in compliance with Chapter 4.2 of the Guide for New Listing Applicants issued by the Stock Exchange.

**4.**

---

**Pre-IPO Investment**

**HISTORY, REORGANIZATION AND CORPORATE STRUCTURE**

– 211 –

**Reasons for fluctuations in valuation as compared to the immediate previous round of pre-IPO Investment and the Global Offering**

| Series | Reason | |---|---| | Series Angel | (a) the increase of our valuation from series angel financing to series pre-A financing was primarily due to the potential launch of our new products such as text model abab1; | | Series Pre-A | (b) the increase of our valuation from series pre-A financing to series A financing was primarily due to our business development such as the entering of agreement with our first API customer and the launch of our text model abab5.5; | | Series A | (c) the increase of our valuation from series A financing to series A+ financing was primarily due to the launch of our new platform such as AI-native multi-modal entertainment platform Talkie; | | Series A+ | (d) the increase of our valuation from series A+ financing to series pre-B financing was primarily due to our further business breakthrough including the launch of our AI-powered multi-modal entertainment platform Xingye, speech model MiniMax-Speech-01 and MoE text model abab6; | | Series Pre-B | (e) the increase of our valuation from series pre-B financing to series pre-B+ financing was primarily due to the launch of our visual generation platform Hailuo AI and video generation model Hailuo-01, our music model Music-01, and that our MAU surpassed 10 million; | | Series Pre-B+ | (f) the increase of our valuation from series pre-B+ financing to series pre-B++ financing was primarily due to our business development such as the launch of our open-source reasoning model MiniMax-M1 with proprietary "Linear Attention" mechanism, our video generation model Hailuo-02, our multilingual speech model Speech-02, our audio generation tool MiniMax Audio and our intelligent agent application MiniMax; | | Series Pre-B++ | (g) the increase of our valuation from series pre-B++ financing to the Global Offering was primarily due to revenue growth in the year of 2025 and our business prospects as a result of our rapid business development. |

Note: The discount to the Offer Price is calculated based on the assumption that the Offer Price is HK$158 per Offer Share, being the mid-point of the indicative Offer Price range and the exchange rates as disclosed in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion".

---

**Pre-IPO Investment**

**Lock-up period**

Sophisticated investors (including our Pathfinder SIIs) under Chapter 2.2 of the Guide for New Listing Applicants are expected to retain at least an aggregate of 50% of their investment at the time of Listing for a period of at least six months following the Listing, in accordance with paragraph 6 under Chapter 2.2 of the Guide for New Listing Applicants.

**Basis of determining the consideration paid**

The consideration for the Pre-IPO Investments was determined based on arm's length negotiations between our Company and the Pre-IPO Investors after taking into consideration various factors including but not limited to, (i) status of milestones and prospects of commercialization of our specialist technology products; (ii) our expansion capacity and R&D management system; (iii) strategic layout, execution efficiency and other factors of our Company, and (iv) the timing of the investments, the market condition, and the prospects of our business.

**Strategic benefits the Pre-IPO Investments brought to our Company**

At the time of the Pre-IPO Investments, our Directors were of the view that our Company would benefit from the additional capital provided by the Pre-IPO Investors' investments in our Company and their knowledge and experience.

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

Information relating to our key Pre-IPO Investors

Our Sophisticated Independent Investors and Pathfinder SIIs

Set out below is a description of our Sophisticated Independent Investors (as defined in Chapter 2.5 of the Guide for New Listing Applicants issued by the Stock Exchange). We have four Sophisticated Independent Shareholders, namely Alisoft China (as defined below), miHoYo SIIs (as defined below), IDG SIIs (as defined below) and Image Frame (as defined below), and two of which, namely IDG SIIs and miHoYo SIIs, are our Pathfinder SIIs. Save for being a shareholder of our Company and as disclosed otherwise, each of our Sophisticated Independent Investors and their ultimate beneficial owners is independent from and not connected with any Director, chief executive or other substantial shareholders of our Company, its subsidiaries or any of their respective associates (within the meaning of the Listing Rules). Each of the Pre-IPO Investors and their ultimate controller and ultimate beneficial owners who is interested in it as to more than 30% is independent from other Pre-IPO Investors and their ultimate controller and ultimate beneficial owners who is interested in it as to more than 30%.

Alisoft China

Alisoft China Holding Limited ("Alisoft China") is a limited liability company incorporated in Hong Kong and an indirect wholly-owned subsidiary of Alibaba Group Holding Limited ("Alibaba Group"). Alisoft China is the holding company of certain PRC subsidiaries of Alibaba Group primarily involved in the operation of cloud computing business. Alibaba Group is a company incorporated in the Cayman Islands, with its American depositary shares, each representing eight ordinary shares, listed on the New York Stock Exchange (symbol: BABA), and its ordinary shares listed on the Stock Exchange (stock code: 9988). Alibaba Group's mission is to make it easy to do business anywhere. Alibaba Group aims to build the future infrastructure of commerce and envisions that its customers will meet, work and live at Alibaba, and that it aspires to be a good company that will last for 102 years. Alibaba Group's core businesses are comprised of e-commerce and cloud computing. According to CIC, Alibaba ranked as the largest player in China's cloud computing market in 2024, with a market share of approximately 22%.

Alisoft China was interested in approximately 15.66% of our Company as of the date of 12 months prior to the date of the listing application and its shareholding was decreased to 15.04% of our Company as of the date of our listing application. Alisoft China is expected to be our substantial shareholder under the Listing Rules upon Listing. Despite that Alisoft China or its close associates maintains business relationship with the Company as disclosed in the section headed "Connected Transactions", having considered, among others, (i) our business relationship with Alisoft China commenced prior to its investments in our Company, (ii) Alisoft China has no involvement in our daily operations and management, and (iii) all of the transactions contemplated thereunder are conducted under normal commercial terms and in the ordinary course of our business, the Company is of the view that the existence of such business relationship will not affect the independence of Alisoft China as one of our sophisticated independent investors.

miHoYo SIIs

miHoYo Limited and Shanghai Mihoyo Argo Technology Co., Ltd (上海米哈游阿爾戈科技有限公司) ("miHoYo SIIs") collectively held 7.34% beneficial interests in the Company as of the date of 12 months prior to the date of the listing application and their beneficial interests were decreased to 7.05% in the Company as of the date of our listing application, which is in compliance with Rule 18C.05 of the Listing Rules. miHoYo Limited is indirectly wholly owned by Mr. Luo Yuhao and Shanghai Mihoyo Argo Technology Co., Ltd is collectively owned by Mr. Cai Haoyu, Mr. Liu Wei and Mr. Luo Yuhao. The aforesaid companies are part of the private enterprise groups founded by Mr. Cai Haoyu, Mr. Liu Wei and Mr. Luo Yuhao. Mr. Cai Haoyu, Mr. Liu Wei and Mr. Luo Yuhao are responsible for all investment decisions in such private enterprise groups and there had been no investment in an entity without the unanimous agreement of Mr. Cai Haoyu, Mr. Liu Wei and Mr. Luo Yuhao for all of the investments made by its investment department, which has been conducting investments to more than 27 companies or partnerships with focus on artificial intelligence, the metaverse, nuclear fusion, mobile games and entertainment industry since establishment. Mr. Cai Haoyu, Mr. Liu Wei and Mr. Luo Yuhao jointly own, make decisions and control such private enterprise groups and they are also empowered to decide on the composition of the investment committee of the investment department. As such, miHoYo Limited and Shanghai Mihoyo Argo Technology Co., Ltd shall be aggregated as one Pathfinder SII pursuant to Chapter 2.5 of the Guide for New Listing Applicants issued by the Stock Exchange.

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

此类民营企业集团专注于视频游戏的开发与发行，其核心视频游戏产品包括《原神》（Genshin Impact）——根据领先的数字市场洞察平台Sensor Tower的数据，该游戏在2021年至2023年间海外收入位居中国游戏之首，Sensor Tower亦是全球最大品牌及应用发行商在移动应用、数字广告、零售媒体及受众洞察领域的领先信息来源。根据CIC的数据，按收入计算，该集团在2021年中国移动游戏公司中排名第五，市场份额为3%。截至2022年12月31日止年度，其国内市场收入近人民币300亿元，净利润逾人民币160亿元。截至2022年12月31日，其在中国境内的总资产逾人民币370亿元。迄今，其业务已拓展至逾200个国家和地区。经CIC确认，此类民营企业集团是下游游戏行业的重要参与者，具有相当的市场份额与规模，根据Sensor Tower的数据，其在2024年按收入计算位居中国移动游戏公司第三位。经CIC确认，其2024年在中国移动游戏公司中按收入计算的市场份额为6%。该集团亦为本集团的下游客户，于2023年在日常经营中使用了本集团的模型。在签署最终协议日期前不超过六个月，以及在提交上市申请日期前不超过六个月，该集团具备被认定为成熟投资者所需的相关投资经验、知识及专业技能。米哈游有限公司（miHoYo Limited）及上海米哈游网络科技股份有限公司（Shanghai Mihoyo Argo Technology Co., Ltd）已于2024年9月3日不可撤销地全额结清其对本公司的投资。根据米哈游集团提供的信息、对公开可查阅资料的审查，以及与本公司香港法律顾问的讨论，联席保荐人认为上述相关米哈游实体满足专业投资者的适用要求。

刘伟先生于2023年4月在米哈游专业投资者于2021年首次投资本公司后获委任为本公司非执行董事。由于刘伟先生并不持有米哈游有限公司的任何股份，且其在上海米哈游网络科技股份有限公司的实益权益不超过30%，根据《上市规则》，米哈游专业投资者并非刘伟先生的紧密联系人，因此，刘伟先生在本公司的董事职务不会影响米哈游专业投资者的独立性。

IDG专业投资者

Cosmic Station Limited（"Cosmic Station"）及Seasonal Charm Limited（"Seasonal Charm"，与Cosmic Station合称"IDG专业投资者"）均为根据英属维尔京群岛法律注册成立的投资控股公司。Cosmic Station为IDG中国创业投资基金VI有限合伙（IDG China Venture Capital Fund VI L.P.，"IDG China VC VI"）的全资子公司。Seasonal Charm为IDG China VI Investors L.P.（"IDG China VI Investors"）的全资子公司。IDG China VC VI及IDG China VI Investors均为根据开曼群岛法律设立的豁免有限合伙企业，由IDG Capital Fund Management Ltd.负责管理，该公司为根据开曼群岛法律注册成立的豁免公司，负责上述基金业务及事务的整体管理与运营。IDG Capital Fund Management Ltd.由IDG资本高级管理层控制。Cosmic Station及Seasonal Charm

– 214 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

collectively held approximately 3.21% beneficial interests in the Company as of the date of 12 months prior to the date of the listing application and their beneficial interests were decreased to 3.08% in the Company as of the listing application. Relevant considerations were irrevocably and fully settled on May 30, 2023.

IDG China VC VI and IDG China VI Investors are venture capital funds with a primary purpose of making equity investments, mainly in seed and growth stage companies in China, focusing on companies in the information technology, media, healthcare, energy, clean technology and non-technology consumer businesses and services related industries, including, but not limited to, companies engaged in software, internet, telecom, media and managed healthcare business. None of the ultimate beneficial owners in each of IDG China VC VI or IDG China VI Investors is interested in it as to more than 30%. There are more than 50 ultimate beneficial owners in IDG China VC VI and IDG China VI Investors which mainly include well-known overseas companies, pension funds and fund of funds. Both IDG China VC VI and IDG China VI Investors are ultimately controlled by Mr. Ho Chi Sing and Mr. Zhou Quan, both being Independent Third Parties.

As at a date which is no more than six months prior to the date of signing of the definitive agreement for their investment in the Company (being October 28, 2023) and as at a date which is no more than six months prior to the date of the Company's listing application (being June 26, 2025), the assets under management of IDG Capital Fund Management Ltd. (the fund manager of IDG China VC VI and IDG China VI Investors) were over HK$30 billion and HK$30 billion, respectively. Both IDG China VC VI and IDG China VI Investors are operated on a discretionary basis in accordance with the relevant partnership agreements. In compliance with Rule 18C.05 of the Listing Rules, IDG SIIs held approximately 3.08% and 3.21% of the total issued share capital of our Company, as of the date of submission of the Company's first listing application and the commencement date of the pre-application 12-month period, respectively.

Tencent

Image Frame Investment (HK) Limited ("Image Frame") is a company incorporated in Hong Kong and is a wholly owned subsidiary of Tencent Holdings Limited ("Tencent"), a company listed on the Hong Kong Stock Exchange (stock code: 00700.HK). Tencent is a world-leading internet and technology company that develops innovative products and services to improve the quality of life of people around the world, including communications and social networks, games, digital content, advertising, fintech and cloud services. Each of Image Frame and its ultimate beneficial owners is an Independent Third Party. Image Frame held approximately 2.96% beneficial interests in the Company as of the date of 12 months prior to the date of the listing application and its beneficial interests were decreased to 2.84% in the Company as of the date of the listing application. According to CIC, Tencent ranked as the third player in China's cloud computing market in both 2023 and 2024, with a market share of approximately 13% in both years.

– 215 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

Other key Pre-IPO Investors

We set out below descriptions of our other key Pre-IPO Investors which are of strategic importance and provided long-term support to our Group in the issued share capital of the Company. All of the Pre-IPO Investors whose background are disclosed in the section headed "— Pre-IPO Investments — 4. Information relating to our key Pre-IPO Investors" (the "Key Pre-IPO Investors") are not required to be aggregated with the other Pre-IPO investors on the basis that the other Pre-IPO investors are not under common control of the Key Pre-IPO Investors. Save as Key Pre-IPO Investors, none of the Pre-IPO Investors has a shareholding in the Company of more than 1.90% as of the date of this Prospectus.

MNM Holdings Limited and XAM Holdings Limited

Each of MNM Holdings Limited ("MNM") and XAM Holdings Limited ("XAM") is an exempted company with limited liability incorporated under the laws of the Cayman Islands.

MNM is a subsidiary of BXA Holdings, L.P. (a limited liability partnership established in the Cayman Islands, "BXA"), and the general partner of BXA is BXA Holdings II GP Limited (an exempted company with limited liability incorporated in the Cayman Islands, "BXA GP".)

XAM is a subsidiary of NVMB IV Holdings Limited (an exempted company with limited liability incorporated in the Cayman Islands) which is wholly-owned by BXA Holdings II, L.P. (a limited liability partnership established in the Cayman Islands, "BXA II"), and the general partner of BXA II is JNR Holdings GP Limited (an exempted company with limited liability incorporated in the Cayman Islands, "JNR GP"). Each of BXA GP and JNR GP is wholly-owned by Mr. Colm O'Connell, an Independent Third Party. There is no individual who directly or indirectly holds an interest of 30% or more in BXA and BXA II.

Miheng Holdings Limited

Miheng Holdings Limited is an exempted company with limited liability incorporated under the laws of Cayman Islands, which is wholly controlled by Beijing Miheng Enterprise Management Consulting Partnership (Limited Partnership) (北京覓恒企業管理諮詢合夥企業(有限合夥)) ("Beijing Miheng"), which is controlled by Zhuhai Gao Ling Private Fund Management Co., Ltd. The general partner of Beijing Miheng is Wuxi Ningjun Enterprise Management Co., Ltd. (無錫寧袀企業管理有限公司), which is controlled by Hillhouse Capital. The limited partners of Beijing Miheng are five private equity funds that are record-filed with Asset Management Association of China. There is no individual who directly or indirectly holds an interest of 30% or more in Beijing Miheng.

– 216 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

HSG

HSG Growth VII Holdco E, Ltd. and Himalia Holding Limited are companies incorporated in the Cayman Islands with limited liability. The sole shareholder of HSG Growth VII Holdco E, Ltd. is HongShan Capital Growth Fund VII, L.P. ("HSG GVII Fund"), whose general partner is HSG Growth VII Management, L.P. The sole shareholder of Himalia Holding Limited is HongShan Capital Growth Fund VI, L.P. ("HSG GVI Fund"), whose general partner is HSG Growth VI Management, L.P. HSG GVII Fund and HSG GVI Fund are investment funds whose primary purpose is to make equity investments in private companies. The general partner of each of HSG Growth VII Management, L.P. and HSG Growth VI Management, L.P. is HSG Holding Limited, which is a wholly-owned subsidiary of SNP China Enterprises Limited. Neil Nanpeng Shen is the sole shareholder of SNP China Enterprises Limited.

MPC VII Pte. Ltd.

MPC VII Pte. Ltd. ("MPC VII") is a limited company incorporated and domiciled in Singapore, which is owned as to 93.97% and 6.03% by MPC VII L.P. and MPC VII-A L.P., respectively. The general partner of both MPC VII L.P. and MPC VII-A L.P., each an exempted limited partnership incorporated under the laws of the Cayman Islands, is MPC Management VII L.P.. The general partner of MPC Management VII L.P. is MPC GPGP VII Ltd. David Su is the controlling shareholder of MPC GPGP VII Ltd.. No single limited partner holds 30% or more interests in MP VII L.P. or in MPC VII-A L.P..

Astrend Entities

Astrend Opportunity IV Beta Limited, Astrend X Fund, L.P., Astrend X-2 Limited, and Golden Horizon Limited (collectively "Astrend Entities") are entities under common control.

Astrend Opportunity IV Beta Limited is a company incorporated under the laws of the British Virgin Islands, which is wholly owned by Shunwei China Internet Opportunity Fund IV, L.P.. The general partner of Shunwei China Internet Opportunity Fund IV, L.P. is Shunwei Capital Partners V GP, L.P., and the general partner of Shunwei Capital Partners V GP, L.P. is Shunwei Capital Partners V GP Limited. Silver Unicorn Ventures Limited holds more than 50% of the issued and outstanding shares of Shunwei Capital Partners V GP Limited, and Mr. Koh Tuck Lye, an Independent Third Party, is the sole shareholder of Silver Unicorn Ventures Limited.

Astrend X Fund, L.P. is an exempted limited partnership incorporated under the laws of the Cayman Islands. The general partner of Astrend X Fund, L.P. is Astrend X Partners GP, L.P., and the general partner of Astrend X Partners GP, L.P. is Astrend X Partners GP Limited. Silver Unicorn Ventures Limited holds more than 50% of the issued and outstanding shares of Astrend X Partners GP Limited, and Mr. Koh Tuck Lye, an Independent Third Party, is the sole shareholder of Silver Unicorn Ventures Limited.

– 217 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

Astrend X-2 Limited is a company incorporated under the laws of the British Virgin Islands, which is wholly owned by Astrend X Fund, L.P.

Golden Horizon Limited is a company incorporated under the laws of the British Virgin Islands, which is ultimately controlled by Mr. Koh Tuck Lye.

Pacific Century Group

Bravo Ideas Investments Limited is an investment holding company incorporated in the Cayman Islands ultimately controlled by Mr. Li Tzar Kai, Richard ("Mr. Li"). Mr. Li is the founder, chairman and chief executive of Pacific Century Group, an Asia-based private investment group founded in 1993.

Future Capital

Each of Future Capital Discovery Fund IV, L.P. and Ideafication Holdings L.P. is a limited partnership whose general partner is Golden Equinox Ltd. and there are no limited partners who are interested in Future Capital Discovery Fund IV, L.P. as to more than 30%. The fund is ultimately controlled by Huang Mingming, an Independent Third Party and the controller of Golden Equinox Ltd. Each partnership is organized for the primary purposes of identifying, analyzing, investing in, managing, otherwise dealing with and realizing investments directly or indirectly in equity and equity-linked securities of privately-held seed and early-stage high-growth companies.

Meaningful investment from Sophisticated Independent Investors

We have received investments from two Pathfinder SIIs, namely IDG SIIs and miHoYo SIIs, each having invested in the Group for at least 12 months prior to the first submission of our listing application to the Stock Exchange for the purpose of the Global Offering. In accordance with Chapter 2.5 of the Guide for New Listing Applicants issued by the Stock Exchange, each of IDG SIIs and miHoYo SIIs holds more than 3%, and in aggregate held approximately 10.13% of the Company's total issued share capital as at the date of the first listing application throughout the period from June 27, 2024 (being the commencement date of the pre-application 12-month period) to June 26, 2025 (being the date of submission of the Company's first listing application). For details of the ownership percentage of shareholding in our Company's share capital of each of the Sophisticated Independent Investors, see "— Capitalization of Our Company".

As of the Latest Practicable Date, our Sophisticated Independent Investors (as identified above) held, in aggregate, approximately 25.44% in the total issued share capital of our Company and approximately 23.32% upon Listing assuming the Offer Size Adjustment Option and the over-allotment option are not exercised. At Listing, our expected market capitalization at the time of Listing will exceed HK$30 billion based on the indicative Offer Price range and such Sophisticated Independent Investors will hold, in aggregate, no less than 15% in the total issued share capital of our Company.

– 218 –

| | Class B Ordinary Shares | Preferred Shares |

– 219 –

| MiniMax Gene | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў |

Our Employee Shareholding Platform

| Floating Sky(12) | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў | ў |

| 20,890,736 | – | – | 5,000,000 | – | – |

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

CAPITALIZATION OF OUR COMPANY

The following table sets out our shareholding structure (a) as of the Latest Practicable Date and (b) immediately upon the completion of the Global Offering (assuming that (i) the Offer Size Adjustment Option and the Over-allotment Option are not exercised, (ii) all Preferred Shares have been converted into Shares on a one-to-one basis immediately upon the completion of the Global Offering, and (iii) without taking into account any Shares that may further be issued under the Post-IPO Share Incentive Plan).

**Entities Controlled by Our WVR Beneficiary(10)**

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | As of the Latest Practicable Date | As of the Latest Practicable Date | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | MiniMax Limited | – | 62,593,180 | – | 62,593,180 | 22.35% | 59.21% | 20.49% | 60.45% | | MiniMax Matrix(2) | – | 11,509,339 | – | 11,509,339 | 4.11% | 10.94% | 3.77% | 11.12% | | MiniMax Awakening | – | 5,000,000 | – | 5,000,000 | 1.79% | 4.75% | 1.64% | 0.48% | | Alpha EXP(3) | 20,890,736 | 15 | – | 20,890,751 | 7.46% | 1.99% | 6.84% | 2.02% | | Sub-total | 20,890,736 | 79,102,534 | – | 99,993,270 | – | – | – | – |

**Our Controlling Shareholders and Entities Controlled by Our WVR Beneficiary(9)**

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | As of the Latest Practicable Date | As of the Latest Practicable Date | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | Cosmic Station Limited | – | – | 7,301,687 | 7,301,687 | 2.61% | 0.69% | 2.39% | 0.71% | | Seasonal Charm Limited | – | – | 535,263 | 535,263 | 0.19% | 0.05% | 0.18% | 0.05% | | Sub-total(4) | – | – | 7,836,950 | 7,836,950 | 2.80% | 0.75% | 2.57% | 0.76% |

– 220 –

**Our Pathfinder SIIs**

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | As of the Latest Practicable Date | As of the Latest Practicable Date | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | miHoYo Limited (米哈遊有限公司) | – | – | 16,015,779 | 16,015,779 | 5.72% | 1.52% | 5.24% | 1.55% | | Shanghai Mihoyo Argo Technology Co., Ltd (上海米哈游阿爾戈科技有限公司) | – | – | 1,912,399 | 1,912,399 | 0.68% | 0.18% | 0.63% | 0.18% | | Sub-total(4) | – | – | 17,928,178 | 17,928,178 | 6.40% | 1.70% | 5.87% | 1.73% |

**Our Other SIIs**

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | As of the Latest Practicable Date | As of the Latest Practicable Date | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | Alisoft China Holding Limited(11) | – | – | 7,000,000 | 7,000,000 | 2.50% | 0.69% (4) | 2.29% | 0.70% | | Image Frame Investment (HK) Limited | – | – | 38,247,987 | 38,247,987 | 13.66% | 3.64% | 12.52% | 3.69% | | Sub-total(4) | – | – | 7,232,084 | 7,232,084 | 2.58% | 0.69% | 2.37% | 0.70% |

– 221 –

**Our Other Key Pre-IPO Investors**

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | As of the Latest Practicable Date | As of the Latest Practicable Date | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | XAM Holdings Limited | – | – | 14,201,184 | 14,201,184 | 5.07% | 1.35% | 4.65% | 1.37% | | MNM Holdings Limited | – | – | 2,343,196 | 2,343,196 | 0.84% | 0.22% | 0.77% | 0.23% | | Sub-total(4) | – | – | 16,544,380 | 16,544,380 | 5.91% | 1.57% | 5.42% | 1.60% | | Himalia Holding Limited | – | – | 3,442,472 | 3,442,472 | 1.23% | 0.33% | 1.13% | 0.33% | | Sub-total(4) | – | – | 3,442,472 | 3,442,472 | 1.23% | 0.33% | 1.13% | 0.33% | | Miheng Holdings Limited | – | – | 1,656,805 | 1,656,805 | 0.59% | 0.16% | 0.54% | 0.16% | | HSG Growth VII Holdco E, Ltd. | – | – | 9,011,235 | 9,011,235 | 3.22% | 0.86% | 2.95% | 0.87% | | Sub-total(4) | – | – | 10,668,040 | 10,668,040 | 3.81% | 1.01% | 3.49% | 1.03% | | MPC VII Pte. Ltd | – | – | 4,932,039 | 4,932,039 | 1.76% | 0.47% | 1.61% | 0.48% | | Astrend Opportunity IV Beta Limited | – | – | 2,260,471 | 2,260,471 | 0.81% | 0.21% | 0.74% | 0.22% | | Astrend X Fund, L.P. | – | – | 1,446,417 | 1,446,417 | 0.52% | 0.14% | 0.47% | 0.14% | | Astrend X-2 Limited | – | – | 814,054 | 814,054 | 0.29% | 0.08% | 0.27% | 0.08% | | Golden Horizon Limited | – | – | 411,097 | 411,097 | 0.15% | 0.04% | 0.13% | 0.04% | | Sub-total(4) | – | – | 7,772,332 | 7,772,332 | 2.78% | 0.74% | 2.54% | 0.75% |

– 222 –

**Our Other Pre-IPO Investors**

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | As of the Latest Practicable Date | As of the Latest Practicable Date | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | Bravo Ideas Investments Limited | – | – | 3,633,558 | 3,633,558 | 1.30% | 0.35% | 1.19% | 0.35% | | Lingham Beauty Limited | – | – | 2,519,330 | 2,519,330 | 0.90% | 0.24% | 0.82% | 0.24% | | Forever Gain Limited | – | – | 1,111,903 | 1,111,903 | 0.40% | 0.11% | 0.36% | 0.11% | | Sub-total(8) | – | – | 3,631,233 | 3,631,233 | 1.30% | 0.35% | 1.19% | 0.35% | | Future Capital Discovery Fund IV, L.P. | – | – | 4,817,351 | 4,817,351 | 1.72% | 0.46% | 1.58% | 0.47% | | Ideafication Holdings L.P. | – | – | 478,100 | 478,100 | 0.17% | 0.05% | 0.16% | 0.05% | | Sub-total(4) | – | – | 5,295,451 | 5,295,451 | 1.89% | 0.50% | 1.73% | 0.51% | | Sub-total | – | – | – | – | – | – | – | – |

– 223 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | | Planetree PARTNERS HARVEST I, L.P. | – | – | 2,885,562 | 2,885,562 | 1.03% | 0.27% | 0.94% | 0.28% | | Planetree Partners III, L.P. | – | – | 478,100 | 478,100 | 0.17% | 0.05% | 0.16% | 0.05% | | Planetree Partners III-A, L.P. | – | – | 2,154,046 | 2,154,046 | 0.77% | 0.20% | 0.71% | 0.21% | | | – | – | 253,416 | 253,416 | 0.09% | 0.02% | 0.08% | 0.02% | | Sub-total(6) | – | – | 3,155,812 | 3,155,812 | 1.13% | 0.30% | 1.03% | 0.30% | | China Life (Shenzhen) Technology Innovation Private Equity Investment Fund Partnership (Limited Partnership) (國壽(深圳)科技創新私募股權投資基金合夥企業(有限合夥)) | – | – | 330,021 | 330,021 | 0.12% | 0.03% | 0.11% | 0.03% | | Hefei China Life Carbon Peak and Carbon Neutrality Phase I Equity Investment Fund Partnership (Limited Partnership) (合肥國壽碳峰碳中一期股權投資基金合夥企業(有限合夥)) | – | – | 2,825,791 | 2,825,791 | 1.01% | 0.27% | 0.93% | 0.27% |

– 224 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | | Star Bairui Holdings Limited | – | – | 217,000 | 217,000 | 0.08% | 0.02% | 0.07% | 0.02% | | Vitalbridge Fund II, L.P | – | – | 1,434,300 | 1,434,300 | 0.51% | 0.14% | 0.47% | 0.14% | | Beijing Shunjin Shunying Enterprise Management Partnership (Limited Partnership) (北京順金順贏企業管理合夥企業(有限合夥)) | – | – | 2,260,471 | 2,260,471 | 0.81% | 0.21% | 0.74% | 0.22% | | Xinnuo Yuheng Ltd. | – | – | 1,912,399 | 1,912,399 | 0.68% | 0.18% | 0.63% | 0.18% | | GW Investment Group Ltd. | – | – | 1,651,111 | 1,651,111 | 0.59% | 0.16% | 0.54% | 0.16% | | Sidsi Holding Limited | – | – | 396,266 | 396,266 | 0.14% | 0.04% | 0.13% | 0.04% | | Trend Xpand Limited | – | – | 1,446,417 | 1,446,417 | 0.52% | 0.14% | 0.47% | 0.14% | | Shanghai Lianxin Technology Equity Investment Center (Limited Partnership) (上海聯新科技股權投資中心(有限合夥)) | – | – | 2,438,309 | 2,438,309 | 0.87% | 0.23% | 0.80% | 0.24% | | Anhui Transportation Holding CICC Industrial Development Fund Partnership (Limited Partnership) (安徽交控中金產業發展基金合夥企業(有限合夥)) | – | – | 2,280,734 | 2,280,734 | 0.81% | 0.22% | 0.75% | 0.22% |

– 225 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | | Sub-total | – | – | – | – | – | – | – | – | | Suzhou CICC SAIC Emerging Industry Equity Investment Fund Partnership (Limited Partnership) (蘇州中金上汽新興產業股權投資基金合夥企業(有限合夥)) | – | – | 552,394 | 552,394 | 0.20% | 0.05% | 0.18% | 0.05% | | Sub-total(7) | – | – | 1,259,854 | 1,259,854 | 0.45% | 0.12% | 0.41% | 0.12% | | Shanghai Fortera FOF Investment Fund (Limited Partnership) (上海國孚領航投資合夥企業(有限合夥)) | – | – | 445,800 | 445,800 | 0.16% | 0.04% | 0.15% | 0.04% | | Shanghai Modou Venture Capital Partnership (Limited Partnership) (上海魔豆創業投資合夥企業(有限合夥)) | – | – | 814,054 | 814,054 | 0.29% | 0.08% | 0.27% | 0.08% | | Sub-total(8) | – | – | 1,031,052 | 1,031,052 | 0.37% | 0.10% | 0.34% | 0.10% | | Shenzhen Pengyuan Cornerstone Private Equity Investment Fund Partnership (Limited Partnership) (深圳市鵬遠基石私募股權投資基金合夥企業(有限合夥)) | – | – | 814,052 | 814,052 | 0.29% | 0.08% | 0.27% | 0.08% |

– 226 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | | Nanshan Alauda Limited | – | – | 891,599 | 891,599 | 0.32% | 0.08% | 0.29% | 0.09% | | JointForce Fund I LP | – | – | 384,392 | 384,392 | 0.14% | 0.04% | 0.13% | 0.04% | | Shanghai Guangqihuichan Phase I Private Equity Investment Fund Partnership (Limited Partnership) (上海光啟匯產一期私募投資基金合夥企業(有限合夥)) | – | – | 334,670 | 334,670 | 0.12% | 0.03% | 0.11% | 0.03% | | Shanghai Guofang Kapa Enterprise Management Partnership (Limited Partnership) (上海國方卡帕企業管理合夥企業(有限合夥)) | – | – | 1,298,626 | 1,298,626 | 0.46% | 0.12% | 0.43% | 0.13% | | Cloud Maximus Limited | – | – | 723,208 | 723,208 | 0.26% | 0.07% | 0.24% | 0.07% | | Shanghai SSCI Leading Artificial Intelligence Private Equity Investment Fund Partnership (Limited Partnership) (上海國投先導人工智能私募投資基金合夥企業(有限合夥)) | – | – | 407,027 | 407,027 | 0.15% | 0.04% | 0.13% | 0.04% | | Sub-total(8) | – | – | 773,352 | 773,352 | 0.28% | 0.07% | 0.25% | 0.07% | | Nanjing Lingyi Cornerstone Equity Investment Partnership (Limited Partnership) (南京領益基石股權投資合夥企業(有限合夥)) | – | – | 220,958 | 220,958 | 0.08% | 0.02% | 0.07% | 0.02% |

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

– 227 –

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | | Sub-total | | | | | | | | | | AIH Global Pte. Ltd. | – | – | 792,533 | 792,533 | 0.28% | 0.08% | 0.26% | 0.08% | | China Orient Enhanced Income Fund | – | – | 198,133 | 198,133 | 0.07% | 0.02% | 0.06% | 0.02% | | Alliance Winford Limited | – | – | 1,981,333 | 1,981,333 | 0.71% | 0.19% | 0.65% | 0.19% | | Jupiter Global Master Fund Ltd. | – | – | 330,222 | 330,222 | 0.12% | 0.03% | 0.11% | 0.03% | | CoreView Master Fund Limited | – | – | 330,222 | 330,222 | 0.12% | 0.03% | 0.11% | 0.03% | | LI FAMILY HOLDINGS PTE. LTD | – | – | 1,320,888 | 1,320,888 | 0.47% | 0.13% | 0.43% | 0.13% | | CloudAlpha Master Fund | – | – | 1,320,888 | 1,320,888 | 0.47% | 0.13% | 0.43% | 0.13% | | Yang family Investments Limited | – | – | 541,564 | 541,564 | 0.19% | 0.05% | 0.18% | 0.05% | | Charoen Pokphand Robot Limited | – | – | 990,666 | 990,666 | 0.35% | 0.09% | 0.32% | 0.10% | | Futron Capital Limited | – | – | 330,222 | 330,222 | 0.12% | 0.03% | 0.11% | 0.03% | | CMG Media Convergence Industry Investment Fund (Limited Partnership) (央視融媒體產業投資基金(有限合夥)) | – | – | 2,377,599 | 2,377,599 | 0.85% | 0.23% | 0.78% | 0.23% | | Elephant Vision Technologies Limited | – | – | 332,893 | 332,893 | 0.12% | 0.03% | 0.11% | 0.03% | | Mentor Group Limited | – | (8) | 2,044,706 | 2,044,706 | 0.73% | 0.19% | 0.67% | 0.20% | | XEP-1 Holdings Limited | – | – | 891,599 | 891,599 | 0.32% | 0.08% | 0.29% | 0.09% | | TAL China Focus Master Fund | – | – | 330,222 | 330,222 | 0.12% | 0.03% | 0.11% | 0.03% | | Janchor Partners Pan-Asian Master Fund | – | – | 924,622 | 924,622 | 0.33% | 0.09% | 0.30% | 0.09% | | Janchor Partners Opportunities Master Fund III | – | – | 2,971,999 | 2,971,999 | 1.06% | 0.28% | 0.97% | 0.29% | | | – | – | 1,320,888 | 1,320,888 | 0.47% | 0.13% | 0.43% | 0.13% |

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

| Shareholders | Class A Ordinary Shares | Class B Ordinary Shares | Preferred Shares | Aggregate number of Shares | Aggregate ownership percentage | Voting power in our Company | Aggregate beneficiary interest | Aggregate Voting Power percentage(1) | |---|---|---|---|---|---|---|---|---| | | | | | | As of the Latest Practicable Date | | Upon Completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised) | | | Nexus Vector Limited | – | – | 1,981,333 | 1,981,333 | 0.71% | 0.19% | 0.65% | 0.19% | | Subtotal | 27,547,541 | 81,102,534 | 171,407,993 | 280,058,068 | 100% | 100% | 91.69% | 97.55% | | Public shareholders | 25,389,220 | – | – | 25,389,220 | – | – | 8.31% | 2.45% | | Total | 52,936,761 | 81,102,534 | 171,407,993 | 305,447,288 | 100.0% | 100.0% | 100.00% | 100.00% |

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

– 228 –

Notes:

(1) On the basis that each Class A Ordinary Share and Preferred Share entitles the Shareholder to one vote per Share and each Class B Ordinary Share entitles the Shareholder to ten votes per Share.

(2) As of the Latest Practicable Date, 5,000,000 Class B Ordinary Shares were held by MiniMax Matrix, which is owned as to approximately 67.1% and 32.9% by Dr. Yan and Ms. Yun, respectively. Upon completion of the Global Offering, all the 5,000,000 Class B Ordinary Shares will be converted into Class A Ordinary Shares.

(3) As of the Latest Practicable Date, among the 62,593,180 Shares held by Alpha EXP, 343,195 were Class A Ordinary Shares and 62,249,985 were Class B Ordinary Shares. Upon completion of the Global Offering, the 343,195 Class A Ordinary Shares will be converted into Class B Ordinary Shares. Alpha EXP is held by Scaling EXP Limited as to 99% and Local Linearity as to 1%. Local Linearity is wholly-owned by Dr. Yan. Scaling EXP Limited is wholly-owned by Trident Trust Company (Hong Kong) Limited, which acts as the trustee of Alpha EXP Trust. Alpha EXP Trust is a trust established by Dr. Yan (as settlor) for the benefit of himself.

(4) For further details, please refer to "4. Information relating to our key Pre-IPO Investors" above in this section.

(5) All the entities are ultimately controlled by China Life Insurance (Group) Company (中國人壽保險(集團)公司).

(6) All the entities are under Planetree Partners.

(7) The executive partners of both entities are subsidiaries of China International Capital Corporation Limited (中國國際金融股份有限公司).

(8) These entities are under common control.

(9) Being our Controlling Shareholders and all controlled by one of our WVR Beneficiaries, Dr. Yan, and will not be counted towards public float.

(10) Controlled by one of our WVR Beneficiaries, Ms. Yun, and will not be counted towards public float.

(11) Will not be counted towards public float.

(12) Floating Sky is held by Floating Cloud Limited as to 99% and Apricity Investment Limited as to 1%. Apricity Investment Limited is wholly-owned by Ms. Yun. Floating Cloud Limited is wholly-owned by Trident Trust Company (Hong Kong) Limited, which acts as the trustee of Floating Sky Trust. Floating Sky Trust is a trust established by Ms. Yun (as settlor) for the benefit of herself.

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

– 229 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

PUBLIC FLOAT

Upon completion of the Global Offering, the Shares held by (i) entities controlled by Dr. Yan and Ms. Yun, namely MiniMax Matrix, MiniMax Limited, MiniMax Awakening, Alpha EXP, and Floating Sky, being close associates of our Directors, and (ii) Alisoft China, being our substantial shareholder, which are our core connected persons, will not be counted towards the public float.

Save as disclosed above, upon the completion of the Global Offering, assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised, 179,550,967 Class A Ordinary Shares (without taking into account the indicative allocation to Alisoft China as set out in the section headed "Cornerstone Investors"), representing approximately 80.0% of the total number of issued Class A Ordinary Shares of our Company, will be counted towards the public float, which is higher than 13.3%, 12.7% and 12.2%, the prescribed percentage of Class A Ordinary Shares required to be held in public hands based on the low-end, mid point or high-end of the indicative Offer Price Range, respectively. The prescribed percentage of Class A Ordinary Shares required to be held in public hands is the higher of (i) the percentage that would result in the expected market value of such securities in public hands to be HK$4.5 billion at the time of Listing, and (ii) 10%, under Rule 8.08(1) (based on the low-end, mid point or high-end of the indicative Offer Price Range). Therefore, our Company will be able to meet the minimum public float requirements under Rules 8.08 of the Listing Rules.

FREE FLOAT

Further, under Rule 8.08A of the Listing Rules, the Company must ensure that a portion of the total number of its issued shares listed on the Stock Exchange with a market capitalization of at least HK$600,000,000 are not subject to any disposal restrictions (whether under contract, the Listing Rules, applicable laws or otherwise) at the time of listing. Our Company will be able to meet the minimum free float requirement under Rule 8.08A of the Listing Rules based on the market capitalization of the Shares listed on the Stock Exchange that are not subject to any disposal restrictions at the time of Listing.

– 230 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

LOCK-UP PERIODS

The table below sets out the list of persons who are, together with their respective close associates, subject to lock-up requirements pursuant to Rule 18C.14 of the Listing Rules:

| Name | Capacity | Aggregate number of Shares held immediately following the completion of the Global Offering(1) | Aggregate ownership percentage of shareholding in the total issued share capital of our Company following the completion of the Global Offering(1) | Lock-up period for a Pre-Commercial Company(3) | |---|---|---|---|---| | Dr. Yan(2) | | | | | | Alpha EXP | } Dr. Yan's close associates | 62,593,180 | 20.49% | The period commencing on the date by reference to which disclosure of its shareholding is made in this Prospectus and ending on the date which is 24 months from the Listing Date. | | MiniMax Awakening | | 11,509,339 | 3.77% | | | MiniMax Matrix | | 5,000,000 | 1.64% | | | MiniMax Limited | | 15 | 0.000005% | | | Total | | 79,102,534 | 25.90% | | | Ms. Yun(2) | | | | | | Floating Sky | Ms. Yun's close associate | 7,000,000 | 2.29% | The period commencing on the date by reference to which disclosure of its shareholding is made in this Prospectus and ending on the date which is 24 months from the Listing Date. | | The miHoYo SIIs | | | | | | Shanghai Mihoyo Argo Technology Co., Ltd | } Pathfinder SII | 1,912,399 | 0.63% | The period commencing on the date by reference to which disclosure of its shareholding is made in this Prospectus and ending on the date which is 12 months from the Listing Date. | | miHoYo Limited | | 16,015,779 | 5.24% | | | Total | | 17,928,178 | 5.87% | | | The IDG SIIs | | | | | | Cosmic Station Limited | } Pathfinder SII | 7,301,687 | 2.39% | The period commencing on the date by reference to which disclosure of its shareholding is made in this Prospectus and ending on the date which is 12 months from the Listing Date. | | Seasonal Charm Limited | | 535,263 | 0.18% | | | Total | | 7,836,950 | 2.57% | |

Notes:

(1) Assuming that (i) the Offer Size Adjustment Option and the Over-allotment Option are not exercised, (ii) all Preferred Shares have been converted into Shares on a one-to-one basis immediately upon the completion of the Global Offering, and (iii) without taking into account any Shares that may further be issued under the Post-IPO Share Incentive Plan.

(2) Dr. Yan and Ms. Yun are our founders, WVR beneficiaries, executive Directors and senior management and Dr. Yan is also our key personnel responsible for our technical operations and/or the research and development of our Specialist Technology Products, who are subject to lock-up requirements pursuant to Rule 18C.14 of the Listing Rules.

(3) The lock-up period pursuant to Rule 18C.14 of the Listing Rules may be shortened if the Company's revenue exceeds HK$250 million by 2025 and is no longer regarded as a Pre-Commercial Company after the Listing, which will be subject to the application by the Company and approval of the Stock Exchange. The lock-up period will not be changed automatically. In the event that upon the notification by the Stock Exchange that our Company will no longer be regarded as a Pre-Commercial Company after the Listing, the lock-up period will expire on the later of: (i) the date on which such lock-up periods would have ended if the Company had applied for listing as a Commercial Company; and (ii) the date falling on the 30th day after the announcement on the removal of designation as a Pre-Commercial Company as required under Rule 18C.24 of the Listing Rules.

– 231 –

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

In addition, outstanding options granted to Ms. Yun, Mr. Pengyu Zhao and Mr. Yucong Zhou under the Pre-IPO Share Incentive Plan as disclosed under the section headed "Statutory and General Information — D. Share Incentive Plans — 1. Pre-IPO Share Incentive Plan — Outstanding Options and Awards — (a) Options" are subject to disposal restrictions for the period commencing on the date of this Prospectus and ending on the date which is 24 months from the Listing Date, subject to Rule 18C.23 of the Listing Rules, in view of their roles as key personnel responsible for the Company's technical operations and/or R&D activities.

– 232 –

– 233 –

1.79%

MiniMax Matrix

Beijing Jizhi

4.11%

MiniMax Awakening

100%

Shanghai Jizhi(2)

100%

Shanghai MiniMax

22.35%

Alpha EXP(4)

MiniMax HongKong(2)

7.46%

(3)

MiniMax Gene

NanoNoble PTE. LTD(2)

100%

The Company

SUBSUP PTE. LTD(2)

0.00001%

MiniMax Limited

Other subsidiaries(1)

61.80%

Other Pre-IPO Investors

Notes: (1) including other subsidiaries wholly owned by the Company. (2) all being indirectly wholly owned by the Company. (3) being managed by an independent professional trustee company for the benefits of the eligible employees under the Company's share incentive plans. (4) for details, please refer to note 3 and note 12 in "— Capitalization" in this section.

2.50%

Floating Sky(4)

67.10%

Local Linearity

32.90%

100%

100%

Dr. Yan

Apricity Investment Limited

Ms. Yun

The following diagram illustrates the shareholding structure of our Company immediately prior to the completion of the Global Offering:

CORPORATE STRUCTURE OF OUR GROUP IMMEDIATELY UPON COMPLETION OF THE REORGANIZATION

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

– 234 –

1.64%

MiniMax Matrix

Beijing Jizhi

100%

Shanghai Jizhi(2)

3.77%

MiniMax Awakening

100%

Shanghai MiniMax

20.49%

Alpha EXP(4)

MiniMax HongKong(2)

6.84%

MiniMax Gene(3)

NanoNoble PTE. LTD(2)

100%

The Company

SUBSUP PTE. LTD(2)

0.000005%

MiniMax Limited

Notes: Please see "— Corporate Structure of our Group Immediately Upon Completion of the Reorganization" above.

2.29%

Floating Sky(4)

67.10%

Local Linearity

32.90%

100%

100%

Dr. Yan

Apricity Investment Limited

Ms. Yun

8.31%

Other Public Shareholders

Other subsidiaries(1)

56.66%

Other Pre-IPO Investors

The following diagram illustrates the shareholding structure of our Company immediately after the Global Offering assuming the Offer Size Adjustment Option and the Over-Allotment Option are not exercised:

CORPORATE STRUCTURE OF OUR GROUP IMMEDIATELY UPON COMPLETION OF THE GLOBAL OFFERING

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE

HISTORY, REORGANIZATION AND CORPORATE STRUCTURE PRC LEGAL COMPLIANCE M&A RULES

According to the Regulations on Merger with and Acquisition of Domestic Enterprises by Foreign Investors (《关于外国投资者并购境内企业的规定》) (the "M&A Rules") jointly issued by the MOFCOM, the State-owned Assets Supervision and Administration Commission of the State Council, the SAT, the CSRC, SAIC and the SAFE on August 8, 2006, effective as of September 8, 2006 and amended on June 22, 2009, merger and acquisition of domestic enterprises by foreign investors means (1) acquiring the equity of a domestic enterprise so as to convert the domestic enterprise into a foreign-invested enterprise; (2) subscribing the increased capital of a domestic enterprise so as to convert the domestic enterprise into a foreign-invested enterprise; (3) establishing a foreign-invested enterprise through which it purchases the assets of a domestic enterprise and operates these assets; or (4) purchasing the assets of a domestic enterprise, and then investing such assets to establish a foreign-invested enterprise (collectively the "Regulated Activities"). The M&A Rules, among other things, further purport to require that an offshore special purpose vehicle, formed for purposes of overseas listing of equity interests in PRC companies and controlled directly or indirectly by PRC companies or individuals, shall obtain the approval of the CSRC prior to the listing and trading of such special purpose vehicle's securities on an overseas stock exchange.

Our PRC Legal Advisor is of the opinion that, based on its understanding of the current PRC laws and regulations, each of the prior CSRC approval for the Global Offering and MOFCOM approval under M&A Rule is not required because our subsidiaries in the PRC were established or acquired by us without involving any Regulated Activities as defined under the M&A Rules.

SAFE REGISTRATION

Pursuant to the SAFE Circular 37, promulgated by SAFE and became effective on July 4, 2014, a PRC resident must register with the local SAFE branch in connection with their contribution of legitimate offshore or domestic assets or equity interests in an overseas special purpose vehicle (the "Overseas SPV") that is directly established or indirectly controlled by the PRC resident for the purpose of conducting overseas investment or financing. Pursuant to SAFE Circular 37, failure to comply with these registration procedures may result in penalties. In addition, due to such failure to comply with the registration procedures, the PRC subsidiaries of that Overseas SPV may be prohibited from distributing their profits and dividends to their offshore parent company or from carrying out other subsequent cross-border foreign exchange activities, and the Overseas SPV and its offshore subsidiary may be restricted in their ability to contribute additional capital to their PRC subsidiaries.

Pursuant to the Notice on Further Simplifying and Improving Foreign Exchange Administration Policy on Direct Investment (《国家外汇管理局关于进一步简化和改进直接投资外汇管理政策的通知》), promulgated by SAFE and effective on June 1, 2015, the power to accept SAFE registration was delegated from local SAFE to qualified banks.

As advised by our PRC Legal Advisor, Dr. Yan and Ms. Yun who are PRC residents have completed the registration as required by SAFE Circular 37.

– 235 –

BUSINESS

OVERVIEW

MiniMax is a global AI foundation model company. Founded by a group of forward-thinking engineers, we are committed to driving AI innovation towards performing the full range of human intellectual tasks, from learning and reasoning to planning and generalizing knowledge across diverse domains.

The foundation model market is expanding at an unprecedented pace, rapidly reshaping human society. The global foundation model market is projected to exceed US$300 billion by 2030. IDC estimates that AI will cumulatively contribute US$19.9 trillion to the global economy through 2030 and drive 3.5% of global GDP in 2030. We believe we have established a solid foundation to capture this market potential and have already made meaningful progress.

Our Journey

Our journey has been guided by a clear vision since inception centered on two key areas: developing advanced foundation models and creating AI-native products that enhance productivity and enrich life. Recognizing that real-world human interaction is inherently multi-modal, we stand out as one of the few foundation model developers who are committed to developing multi-modal models from day one. We take a cost-efficient approach in pursuing AI advancement, delivering high performance while ensuring our technological breakthroughs remain accessible and affordable to users globally. We adopted the Mixture-of-Experts (MoE) architecture and hybrid attention mechanism at an early stage, which significantly reduced computation resources while maintaining globally recognized performance.

| Category | Model Versions | |---|---| | Large Language Model | abab 1 → abab 5.5 → abab 6.0 (MoE) → Text-01 → M1 → M2 | | Video Generation Model | Hailuo-01 → Hailuo-02 | | Audio Model | Speech-01 → Speech-02 / Music-01 → Music-02 |

| | OPEN Platform | Product Launching | |---|---|---| | | MiniMax (with Agent) | Talkie/Xingye | | | | MiniMax Audio |

Timeline: 2022 → 2023 → 2024 → 2025

– 236 –

BUSINESS

We have been consistently iterating our models to higher intelligence levels. Today, our proprietary foundation model suite, led by MiniMax-M2, Hailuo-02, and Speech-02, has long context processing capacity and can understand, generate, and integrate a wide range of modalities, including text, video, and audio. These models power our major AI-native products — including MiniMax, Hailuo AI, MiniMax Audio, Talkie/Xingye, and our enterprise and developer-facing Open Platform, delivering intelligent and dynamic experiences to users globally.

As of September 30, 2025, our AI-native products had cumulatively served over 200 million individual users across over 200 countries and regions, and more than 100 thousand enterprises and developers across over 100 countries and regions.

Scalability

We believe scalability is pivotal to our long-term goal. To build one of the most scalable AI businesses globally, we focus on three core competencies — original research, a sustainable business model, and organizational efficiency. These pillars support both continuous model advancement and product commercialization at scale. Together, the three core competencies enable an elevated level of intelligence for everyone—powering productivity and enriching life.

Original Research

• Multi-modal Focus. We are among the first globally to pursue a multi-modal technology strategy, and we commenced the development of models across multiple modalities in 2022. Our models consistently rank at the top across text, video, and speech benchmarks, reflecting a systematic advantage in multi-modal architecture that empowers the development of more scalable models and AI-native products.

• Enhanced AI Infrastructure. We have prioritized and significantly enhanced AI infrastructure efficiency, achieving persistent improvement in training performance and significantly reducing overall inference costs. Our proprietary AI infrastructure dynamically allocates computing resources, ensuring service availability and supporting sustainable large-scale delivery of high-performance foundation models.

Sustainable Business Model

• Technology as Product. We focus on developing scalable AI systems optimized for real-world use cases. This strategy has enabled us to develop and commercialize high-performing models across multiple modalities. Based on these foundation models, we have developed a suite of AI-native products that serve both individual users, developers and enterprise customers across a broad range of application scenarios. Our multi-pronged monetization enables a self-reinforcing cycle of innovation and commercial value, which further allows us to continuously reinvest in original research and development.

– 237 –

业务 •

全球运营。从第一天起，我们便在国际市场上同步推出所有基础模型和产品，目标只有一个：以极具吸引力的价值主张，使下一代人工智能技术真正广泛普及。我们在多个国际市场的同步增长，证明了我们全球化战略的有效性以及我们技术护城河的深厚实力。

组织效率 •

我们扁平、灵活的组织架构支持模型快速迭代、研究与产品的深度融合，以及模型与产品开发的规模化扩张。我们在首席执行官以下设置不超过三个管理层级，并围绕基于项目的使命而非僵化的部门壁垒来组建团队。这种动态架构赋予员工早期主人翁意识，促进人才快速成长，推动技术、产品与业务职能之间的深度协作，并有助于将研究创新成果转化为现实世界的落地与影响。

我们的模型与产品供应 我们的基础模型套件 我们充分发挥研发能力，构建了一套全面的基础模型矩阵，并在多种模态上保持竞争力。我们的基础模型套件涵盖大语言模型、视频生成模型，以及语音和音乐生成模型。

大语言模型：MiniMax MiniMax M系列由MiniMax-M1和MiniMax-M2组成，是我们旗舰级大语言模型系列。MiniMax-M1于2025年6月发布，是一款开源的大规模混合注意力推理模型。该模型采用混合MoE（混合专家）架构与闪电注意力机制相结合的设计，支持最长达100万tokens的超长上下文处理，并为构建能力更强的AI智能体提供支撑。

MiniMax-M2是我们最新的大语言模型，专为在编程和智能体任务中实现卓越性能而设计。MiniMax-M2采用精心设计的数据高效MoE架构与激活参数设计，在保持模型智能、响应速度与成本效益优化的同时，相较于MiniMax-M1实现了更高性能及显著更快的推理速度。

– 238 –

业务

视频生成模型：Hailuo-02

Hailuo-02系列模型能够从多种形式的信息输入中生成高质量视频内容。该模型于发布时即实现大规模商业化，并在全球基准测试中取得具有竞争力的成绩。Hailuo-02提供电影级视频质量、出色的提示词遵循能力、流畅的动态效果以及多样化的风格。凭借用户友好的界面及美学优化能力，它帮助内容创作者和广告商从简单的提示词中制作出引人入胜的视频。

语音生成模型：Speech-02

Speech-02系列模型旨在从文本输入中生成自然、高质量的语音。我们的Speech-02模型于2025年4月发布，在全球范围内广受认可，被誉为顶级语音模型之一，能够跨多种语言提供超逼真的个性化语音合成。

我们的AI原生产品

依托我们的多模态基础模型套件，我们为全球个人用户、开发者及企业客户提供AI原生产品和服务，充分释放AI的力量，创造广泛价值。我们AI原生产品的演进植根于其底层基础模型的进步。通过对基础模型的持续迭代与升级以及新模型的开发，我们能够设计和打造具有更强生产力与用户体验的AI原生产品。

MiniMax：智能体应用

MiniMax是我们的智能AI智能体应用，旨在通过自然语言指令自主执行各类任务。依托我们的基础模型，MiniMax Agent能够在统一工作空间内进行规划、推理并执行复杂操作，包括编程、研究、文档起草和演示文稿创作。

Hailuo AI：旗舰视觉生成平台

Hailuo AI完整集成了我们的Hailuo-02模型，凭借用户的自发采用，已迅速跻身全球最受欢迎的AI图像与视频创作平台之列。该平台提供网页端和应用程序两种形式，专为实时、高质量的图像和视频生成而设计。

MiniMax Audio：高级音频生成工具

MiniMax Audio旨在为用户提供高保真音频生成能力。用户可通过网页平台访问，MiniMax Audio集成了公司的Speech-02模型，支持交互式音频合成，并能从文本输入中生成自然、高质量的语音。

— 239 —

BUSINESS

Talkie/星野：多模态娱乐平台

Talkie（面向国际市场）/ 星野（面向中国国内市场）是全球公认的AI原生多模态娱乐平台。Talkie/星野的用户可与由本公司自研AI模型驱动的、具有情感响应能力的AI主题或虚拟角色进行互动。

MiniMax开放平台

截至2025年9月30日，我们的开放平台已向100多个国家和地区的企业客户及开发者提供可扩展、可配置的AI服务。通过公开API及相关服务，企业及开发者客户可访问本公司的基础模型，并将文本、视频及音频模型能力集成至其自有产品与服务中。我们的开放平台支持在智能设备、医疗健康、文化旅游、金融及互联网服务等重点行业领域快速部署业务，以平均日均Token用量计算，已成为全球面向企业及开发者规模最大的开放平台之一，彰显了其广泛的市场采用度。

核心运营数据

我们的AI原生产品矩阵已吸引大量用户，平均月活跃用户数（MAU）从2023年的310万增长至2024年的1,910万，增幅超过六倍，并进一步增长至截至2025年9月30日止九个月的2,760万。截至2025年9月30日，我们AI原生产品的累计用户数已超过2.12亿。面向消费者产品用户数量的持续增长提供了宝贵的用户反馈，有助于推动产品的快速迭代与优化。

过去三个季度，我们实现了月活跃用户数的环比持续增长，这一趋势主要归因于Talkie/星野App出色的用户体验与积极的用户评价，以及2024年第三季度上线的旗舰视觉生成平台海螺AI（Hailuo AI）的推动。Talkie/星野作为我们已商业化的AI原生产品组合中具备多模态能力的产品，我们持续整合新基础模型技术，引入一系列沉浸式体验与视觉美学增强功能，从而全面提升用户交互质量。与此同时，海螺AI底层的Hailuo-01及Hailuo-02系列模型已在全球范围内取得极具竞争力的性能表现，使应用内用户能够在与其他市场产品的对比中，明显感受到其生成视频在整体画面质量及电影级镜头运动效果方面的差异化优势。

我们AI原生产品的付费用户数从2023年的约119,700人增长至2024年的约650,300人，并进一步增长至截至2025年9月30日止九个月的约1,771,600人。

— 240 —

BUSINESS

Complementing our growing individual paying user base, we have also cultivated a portfolio of enterprise customers and developers. Enterprise customers and developers access our core AI models via our Open Platform, which supports growing business needs across key industry sectors. Our Open Platform demonstrates solid monetization capabilities for the foundation models offered. We have consistently observed a rapid increase in paying customers on the Open Platform. Our number of paying users on the Open Platform, defined as users who have individually consumed no less than US$50 worth of API calls in a given period, expanded from around 100 in 2023 to around 700 in 2024, and further to approximately 2,500 in the nine months ended September 30, 2025.

The following chart sets forth the number of users and customers within each period of the Track Record Period 1,2:

| | As of December 31, | | | As of September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 | 2025 | | | ('000 users) | | | | | | AI-native products | – | 11,131 | 115,378 | 76,571 | 212,247 | | MiniMax | – | 686 | 13,541 | 10,969 | 19,057 | | Hailuo AI | – | – | 5,735 | 36 | 42,348 | | MiniMax Audio | – | – | 47 | – | 3,742 | | Talkie/Xingye | – | 10,445 | 96,055 | 65,566 | 147,100 | | Open Platform | – | 13 | 42 | 34 | 132 | | Total | – | 11,144 | 115,420 | 76,605 | 212,379 |

Notes:

1. Number of users comprise all registered users for our web-based AI-native products and all activated devices for our app-based AI-native products. As some users may have multiple accounts, we cannot guarantee that each user is a unique individual.

2. Number of customers of our Open Platform comprise all registered customers who have made API calls on our Open Platform. Customers who have registered but not made API calls are not included. Our Open Platform is designed as a technology access platform for a broad range of developers, including both developers and enterprise customers. Our Open Platform is managed on a developer account basis. During registration and subsequent use, we only require users to provide basic contact information (such as email address, mobile number and account nickname) and do not require them to upload business licences or identity documents, nor do we use such information as a mandatory classification standard. As a result, we are currently not able to reliably distinguish whether an Open Platform customer is a developer or an enterprise customer.

– 241 –

BUSINESS

The following chart sets forth the number of paying users within each period of the Track Record Period¹:

| | For the year ended December 31, | For the year ended December 31, | For the year ended December 31, | For the nine months ended September 30, | For the nine months ended September 30, | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 | 2025 | | ('000 users) | | | | | | | AI-native products | – | 119.7 | 650.3 | 489.1 | 1,771.6 | | MiniMax | – | – | – | – | 10.3 | | Hailuo AI | – | – | 64.8 | – | 311.1 | | MiniMax Audio | – | – | – | – | 59.8 | | Talkie/Xingye | – | 119.7 | 585.5 | 489.1 | 1,390.4 | | Open Platform | – | 0.1 | 0.7 | 0.4 | 2.5 | | **Total** | **–** | **119.8** | **651.0** | **489.5** | **1,774.1** |

Note: 1. A paying user for AI-native products is defined as a user who has made at least one monetary transaction in a given period. A paying user for our Open Platform is defined as a user who has individually consumed no less than US$50 worth of API calls in a given period.

The following chart sets forth the number of average monthly active users ("MAU") within each period of the Track Record Period¹˒²:

| | For the year ended December 31, | For the year ended December 31, | For the year ended December 31, | For the nine months ended September 30, | For the nine months ended September 30, | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 | 2025 | | ('000 users) | | | | | | | AI-native products | – | 3,144 | 19,106 | 14,601 | 27,622 | | MiniMax | – | 239 | 2,195 | 2,166 | 1,429 | | Hailuo AI | – | – | 2,172 | 36 | 5,648 | | MiniMax Audio | – | – | 47 | – | 494 | | Talkie/Xingye | – | 2,905 | 14,692 | 12,399 | 20,051 | | Open Platform | – | 4 | 5 | 4 | 16 | | **Total** | **–** | **3,148** | **19,111** | **14,605** | **27,638** |

Notes: 1. MAUs comprise all unique devices that performed at least one action on our AI-native apps and all registered user accounts that logged into our web platforms at least once during a given month, including both paying and non-paying users. As some users may have multiple accounts, we cannot guarantee that each user is a unique individual.

2. The average monthly active customers for Open Platform comprise all registered customers who have made API calls during a given month on our Open Platform including both paying and non-paying customers. Customers who have registered but not made API calls are not included.

– 242 –

BUSINESS

Our suite of AI-native products has attracted a broad user base, with average MAU rising more than six times from 3.1 million in 2023 to 19.1 million in 2024 and further to 27.6 million in the nine months ended September 30, 2025. Specifically, the average MAU of Talkie/Xingye increased from 2,905,000 in 2023 to 14,692,000 in 2024, and further to 20,051,000 in the nine months ended September 30, 2025; the average MAU of Hailuo AI increased from 2,172,000 in 2024 to 5,648,000 in the nine months ended September 30, 2025; and the average MAU of MiniMax Audio increased from 47,000 in 2024 to 494,000 in the nine months ended September 30, 2025. Our average MAU of MiniMax decreased from the nine months ended September 30, 2024 to the same period in 2025. This decrease was not the result of weak user retention, but was primarily driven by our strategic product shift from broad chat use cases to agent-based capabilities targeting professional, higher-value users. During this period, we introduced usage-tiered paid features and reduced free consumption, resulting in lower activity from non-core users while core user stickiness remained stable. We also reduced paid marketing and promotional spending by approximately 90% in the nine months ended September 30, 2025 as we transitioned to an organic growth strategy. Despite this substantial cut in advertising spending, the MAU of MiniMax did not decline proportionately, reflecting MiniMax's market recognition and positive user feedback, which further indicate the success of organic user acquisition driven by model intelligence. We prioritize organic acquisition as paid traffic relies heavily on external channels and rising market bidding costs, whereas organically acquired users convert based on model capability, product value and user experience. This builds a more durable user base, drives word-of-mouth growth and improves unit economics. By focusing on model performance, agentic capabilities and user experience, we aim to develop a self-reinforcing growth flywheel with a more controllable cost structure and improved profitability.

The following chart sets forth the number of new users within each period of the Track Record Period 1,2:

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 | 2025 | | | ('000 users) | | | | | | AI-native products | | | | | | | MiniMax | – | 686 | 12,855 | 10,283 | 5,516 | | Hailuo AI | – | – | 5,735 | 36 | 36,613 | | MiniMax Audio | – | – | 47 | – | 3,695 | | Talkie/Xingye | – | 10,445 | 85,610 | 55,121 | 51,045 | | Open Platform | – | 13 | 29 | 21 | 90 | | **Total** | **–** | **11,144** | **104,276** | **65,461** | **96,959** |

Notes:

1. New users comprise all newly registered users for our web-based AI-native products and all newly activated devices for our app-based AI-native products.

2. New customers for Open Platform comprise all newly registered customers who have made API calls. Customers who have registered but not made API calls are not included.

– 243 –

BUSINESS

我们 AI 原生产品的新用户数量从 2023 年的 1,110 万增长至 2024 年的 1.042 亿，增幅超过九倍；同期，截至 2024 年 9 月 30 日止九个月的新用户数量为 6,540 万，截至 2025 年同期则增至 9,690 万。MiniMax 的新用户数量从截至 2024 年 9 月 30 日止九个月至 2025 年同期有所下降，Talkie/星野的新用户数量在同期亦有所减少。上述下降主要归因于我们的战略调整——将资源从广泛的用户获取转向 MiniMax 的变现举措，包括对付费功能的优化。作为此次转型的一部分，我们削减了整个产品组合的整体市场营销及推广支出，导致 MiniMax 和 Talkie/星野的营销活动减少。尽管营销支出有所降低，新用户的减少幅度明显小于相关营销预算的削减幅度，这反映出用户留存率的提升、品牌认知度的增强，以及市场对 MiniMax 和 Talkie/星野接受度的持续改善。

下表列示了业绩记录期内各期间每位付费客户的平均消费额¹：

| | 截至 12 月 31 日止年度 | | | 截至 9 月 30 日止九个月 | | |---|---|---|---|---|---| | | 2022 年 | 2023 年 | 2024 年 | 2024 年 | 2025 年 | | | （美元） | | | | | | AI 原生产品 | – | 6 | 11 | 7 | 15 | | MiniMax | – | – | – | – | 73 | | Hailuo AI | – | – | 36 | – | 56 | | MiniMax Audio | – | – | – | – | 18 | | Talkie/星野 | – | 6 | 8 | 7 | 5 | | 开放平台 | – | 27,020 | 12,454 | 14,813 | 6,167 |

注： 1. 对于 AI 原生产品，每位付费客户的平均消费额按各产品在特定期间内来自应用内充值及订阅所产生的收入除以付费用户数计算。对于开放平台，每位付费客户的平均消费额按开放平台在特定期间内产生的总收入除以开放平台的付费用户数计算。

我们 AI 原生产品每位付费客户的平均消费额从 2023 年的 6 美元增至 2024 年的 11 美元，并进一步增至截至 2025 年 9 月 30 日止九个月的 15 美元。其中，Hailuo AI 每位付费客户的平均消费额从 2024 年的 36 美元增至截至 2025 年 9 月 30 日止九个月的 56 美元。Talkie/星野每位付费客户的平均消费额从截至 2024 年 9 月 30 日止九个月至 2025 年同期有所下降。这主要是由于 Talkie/星野在产品持续快速扩张的过程中，加速渗透至更广泛的用户群体，尤其是消费水平较低的用户群体。用户结构的变化导致每位付费客户的平均消费水平下降。尽管如此，用户基础的拓宽反映出该产品市场覆盖范围的持续扩大及其对更广泛用户群体的吸引力不断增强，预计将为长期变现提供更为可持续的基础。

– 244 –

BUSINESS

The average spending per paying customer of Open Platform decreased from 2023 to 2024 and from the nine months ended September 30, 2024 to the same period in 2025, primarily attributable to the rapid expansion of the user base following the official launch of our Open Platform to overseas customers in the fourth quarter of 2024, which significantly broadened the platform's user reach beginning in the fourth quarter of 2024 and continuing through 2025. As the proportion of new users with relatively smaller transaction volumes increased, the overall average spending per paying customer was subject to a structural dilution effect, reflecting a more pronounced long-tail contribution as total paying customers grew.

The number of new users of our AI-native products increased more than nine-fold from 11.1 million in 2023 to 104.2 million in 2024, and also grew from 65.4 million for the nine months ended September 30, 2024 to 96.9 million for the same period in 2025. The number of new users of MiniMax decreased from the nine months ended September 30, 2024 to the corresponding period in 2025, while the number of new users of Talkie/Xingye decreased during the same periods. The decrease was primarily attributable to our strategic adjustment to reduce marketing and promotional spending, with an increased emphasis on organic user acquisition and user quality as we focused on the monetization phase of MiniMax. Notwithstanding the lower level of marketing spending, the decline in new users was significantly smaller than the reduction in the related marketing budget, reflecting enhanced user retention, brand recognition, and improving market acceptance of MiniMax and Talkie/Xingye.

OUR STRENGTHS

Our core strength resides in the scalability embedded across our key operational pillars. This includes our algorithms, model training and inference infrastructure, commercialization roadmaps, and organizational structure, all of which are structured to facilitate the long-term scalability of our company.

A Prospective Innovation Roadmap for Scalable Model Capabilities

MiniMax is built on a prospective technological vision and a focus on scalability. From day one, we focus on expanding multi-modal capabilities and pursuing model algorithm innovation.

Multi-modal Capabilities:

• From the outset, we prioritize developing multi-modal models, recognizing that the content people engage with daily in real world extends beyond text to include multi-modal formats such as video and audio. As one of the earliest adopters of a comprehensive approach to foundation model development, we offer competitive commercial-scale foundation models across text, video and audio.

– 245 –

BUSINESS

•

Our multi-modal models have achieved both notable commercial success and recognition from independent model benchmarking providers. These models have consistently delivered top-tier results across text, video and speech benchmarks, in both model performance and cost efficiency, underscoring a systematic advantage inherent in our multi-modal architecture. Leveraging the capabilities of our multi-modal models, the Company's diverse suite of AI-native products has successfully attracted a growing user base. Average monthly active users have surged nearly nine-fold, from 3.1 million in 2023 to 27.6 million in the nine months ended September 30, 2025.

Model Algorithm Innovation:

•

MoE Architecture: Having commenced research in 2023 and released our initial model featuring the MoE architecture by 2024, our early adoption of MoE architecture has notably delivered a significant reduction in inference latency, underscoring our technological foresight. We are the first in Asia and one of the first globally to commercialize the MoE foundation model architecture, addressing limitations of traditional "dense" models. This structural advantage provides an enhancement in scalability and efficiency, translating directly into reduced computational demands and lower inference costs.

•

Linear Attention: Recognizing the limitations of conventional model architectures in handling large-scale inputs, we developed a proprietary "Linear Attention" mechanism to overcome these constraints. This innovation allows our models to excel in long-context processing — further enhancing their efficiency and scalability and facilitating the development of more powerful AI agents.

Scalable AI Infrastructure Delivering Efficiency and Performance

We view the high cost of AI model training and inference as a key barrier to widespread adoption of AI technologies. We believe that model scaling depends on the volume of available computing power, utilization efficiency of computing resources, and associated costs. To scale up at a lower cost, we have prioritized the development of proprietary AI infrastructure. This involved building an in-house infrastructure team and independently creating a high-performance training framework suitable for large-scale computing clusters. Our AI infrastructure design takes a holistic approach, from operator-level to strategic cross-cluster load balancing, to optimally facilitate scaling. The notable aspects of our AI infrastructure include:

Advanced AI Training and Inference Framework: Our advanced AI training and inference framework provides a flexible, scalable solution for the development and inference of foundation models. At the operator level, which is the atomic computational building blocks that form the model training and inference algorithm, we have engineered deep optimization to significantly enhance computational efficiency, reduce latency, and improve utilization of computing resources. Our training and inference framework features automated functions such as scaling, descaling, and dynamic parallel strategy adjustment, with deep integration into

– 246 –

BUSINESS

underlying hardware and networks. Together, these measures yield a stable and high-performance model training and inference environment tailored for our foundation models. We have achieved more than 75% inference Model Flop Utilization (MFU), a key measure of how efficiently our foundation models use computing power to conduct inference activities, significantly higher than the industry average of approximately 40% to 50%. A higher inference MFU signifies a greater and more effective utilization of our available computing resources, which in turn translates to lower inference costs, faster inference performance, and enhanced scalability.

Unified Training and Inference Computing Resources: We believe that efficiently utilizing all available computing resources is crucial, and a key approach is to share computing resources between model training and inference activities. However, handling both activities across the same large computing clusters presents significant scheduling challenges. By implementing an intelligent scheduling system, we are able to successfully allocate computing resources across various types of assignments, prioritizing time-sensitive tasks. During periods of lower demand, resources typically used for time-sensitive training and inference workloads can be redirected to less time-sensitive offline tasks like data processing. This dynamic approach effectively fills idle gaps, allowing us to maximize computing resource utilization without affecting the overall performance of our computing clusters.

Cross-Cluster Load Balancing: To address diverse AI-native product demands and meet high computing needs, our AI infrastructure employs multiple independent computing clusters that operate as a unified system, orchestrated by our intelligent cross-cluster load balancer. This proprietary technology significantly enhances computing efficiency: it acts as a smart traffic controller, evaluating tasks in real-time and routing them to the most suitable computing cluster based on workload, availability, and hardware compatibility. Such technology ensures the consistent performance of our AI infrastructure, even during peak user demand, by dynamically monitoring and rerouting tasks from overloaded to underutilized clusters, thereby effectively preventing computing resource bottlenecks and maximizing utilization.

Scalable Commercialization Approach with Global Adoption

From inception, we have purposefully selected and developed the most scalable products, characterized by model intelligence-driven product iteration, organic traffic-focused customer acquisition strategies, and ease of standardization. We have also cultivated diversified monetization channels, including subscription services, token-based in-app purchases, online marketing services, and usage-based enterprise APIs. Our Hailuo AI has quickly become one of the world's most popular AI video creation platforms through organic user adoption. According to CIC, Talkie/Xingye is the second-largest AI-native entertainment platform globally in terms of average MAU in the nine months ended September 30, 2025.

We have embraced a global approach, launching all models and products across international markets to compete at scale from day one. The feedback received from our customers at a global scale has tremendously aided in the enhancement of our products and technology, enabling us to develop competitive product offerings. We are currently one of the

– 247 –

BUSINESS

two companies founded in the Asia-Pacific region to achieve global-scale commercialization of multiple AI native products. Our offerings serve both individual users and enterprise customers across the globe. Our unified model architecture and organic user acquisition strategy have facilitated rapid expansion across more than 200 countries and regions as of September 30, 2025.

Flat and Nimble Organizational Structure Enabling Innovation and Execution

Our unique organizational structure drives our scalability, enabling rapid and highly iterative R&D progress. From day one, our colleagues are immersed in a culture that empowers talent at every level. Led by our visionary founder Dr. Junjie Yan, we operate at the cutting edge in our field, advancing AI from research to implementation.

Our organizational structure is intentionally flat and nimble, with no more than three layers beneath the CEO, enabling faster decision-making which facilitates iteration of intelligent foundation models. We operate cross-functional teams assembled through a project-based model. Breaking down traditional silos between tech, product and business, all teams are aligned on one goal: elevating the intelligence level of our models and making them accessible to everyone.

We promote research and creativity through an inclusive environment that values diverse contributions and continuous improvement. We empower our colleagues to take early ownership opportunities and entrust them with real authority as they demonstrate their capabilities: many individuals hired directly from universities now lead key research and development initiatives. We do not place strict limits on job scope, actively encouraging individuals to embrace more responsibilities beyond their defined roles. This is exemplified in our research and development team leaders, who are typically less than 30 years old. Our incentive mechanisms are flexible and performance-driven, including constant salary adjustments and project-based bonuses.

OUR STRATEGIES

We are committed to scaling our capabilities to make AI universally accessible. We plan to implement the following three core strategies:

Advance AI through R&D Leadership

We will remain heavily invested in core R&D capabilities to maintain and continuously expand our leadership in foundation model development. Specifically, we aim to pursue innovations that enhance productivity and accessibility by reducing model inference costs. We will continue advancing model architectural breakthroughs such as dynamic multi-modal model integration. Sustained investment based on technological insights will allow us to push multi-modality integration and real-world application.

– 248 –

BUSINESS

Deliver "Technology as Products" with Commercial Potential

We will continue to develop and deliver our "technology as products" approach to provide positive experiences to both individual users and enterprise customers. By leveraging our expertise in model research and product commercialization, we will broaden and deepen monetization venues across AI-native products. Our goal is to elevate the intelligence levels of our AI models while lowering associated costs for our individual users and enterprise customers, enabling our AI-powered products to reach a broader user base and unlock new application scenarios. We will continue to enhance the performance of our existing AI-native products and Open Platform. As model intelligence unlocks new application scenarios, we will iterate on product interfaces, delivery formats and monetization models to capture commercialization opportunities. We aim to achieve a scalable and sustainable economic model. We plan to continue investing in the development and refinement of our AI-native products and Open Platform, with a focus on improving engagement, multi-modal capabilities and enterprise adoption. We will expand R&D to enhance existing products and to launch new AI-native applications that leverage ongoing advancement of our models. To support commercialisation, we intend to expand our product development and commercialization organization, as well as our international sales and marketing team, by hiring approximately 70 additional specialists over the next five years. These hires will enable us to improve user experience as well as customer relationship management, and scale adoption in overseas markets.

Evolve our Organization and Expand Talent Pool

We will continue to adapt our organizational structure to maintain competitiveness in the rapidly evolving AI industry. We are committed to maintaining a collaborative and inclusive culture, empowering team members with early ownership and authority as they demonstrate their capabilities. By championing cross-functional innovation, we will continue to foster a sustainable R&D organizational culture. Regarding talent acquisition, we will keep building a deep talent pipeline for both foundation model and AI-native product development to secure long-term technological leadership. Specifically, we plan to continue to recruit 150 top-tier foundation model and AI infrastructure researchers, engineers, and scientists globally over the next five years, while also investing in the growth and development of our in-house talent.

OUR AI MODELS

We focus on developing foundation models and AI-native products to meet the evolving needs of individual users, developers and enterprise customers. Our "technology-as-product" offerings are built on proprietary, self-developed technologies, ensuring advanced quality, scalability, and cost-efficiency. Recognizing that real-world human interaction is inherently multi-modal, we strive to enhance user engagement and content accessibility by delivering multi-modal foundation model solutions that maximize reach and impact.

– 249 –

BUSINESS

Our foundation models are artificial intelligence systems designed to interpret human inputs and generate high-quality outputs across multiple formats, including text, video, and audio. These models are built using advanced deep learning techniques. Their primary function is to learn and recognize patterns, structures, and nuances within the training data, allowing them to generate contextually relevant and highly accurate outputs.

We possess globally recognized foundation models and are among the few AI companies that excel across various modalities. Our key model offerings include: (i) MiniMax-M2 and MiniMax-M1, our proprietary large language model series; (ii) Hailuo-02, our video generation model covering text-to-video and image-to-video; and (iii) Speech-02, our multilingual speech generation model. These models support a wide range of our proprietary AI-native products, including MiniMax, Hailuo AI, MiniMax Audio, Talkie/Xingye, and our Open Platform. Our models are designed to deliver high scalability, cost efficiency, and advanced multi-modal capabilities.

The following chart sets forth our core models matrix as of the Latest Practicable Date:

| Category | Latest Core Models | Description | Launch Dates | Development Timespan from the Previous Version | |---|---|---|---|---| | Text | MiniMax-M1 | Reasoning Model | June 2025 | 5 Months from MiniMax-Text-01 | | Text | MiniMax-M2 | Reasoning Model | October 2025 | 4 Months from MiniMax-M1 | | Video | Hailuo-02 | Video Generation | June 2025 | 8 Months from Hailuo-01 | | Audio | Speech-02 | Multilingual Voice Generation | April 2025 | 16 Months from Speech-01 |

MiniMax M Series: Large Language Model

We are one of the earliest companies globally to start developing large language models, having begun exploring such technology since 2022. Our large language model has undergone multiple iterations, evolving from the abab1 launched in 2022 to the MiniMax-M2, released in October 2025.

MiniMax-M1

The MiniMax-M1 is an open-source, large-scale hybrid-attention reasoning model released in June 2025. As a general matter, a reasoning model is a type of foundation model designed to process inputs, analyze information, and generate logical, structured outputs by simulating human-like reasoning processes. These models are optimized for tasks that require critical thinking, problem-solving, and the ability to connect disparate pieces of information to produce actionable or explanatory results.

– 250 –

BUSINESS

MiniMax-M1 Highlights

Upon its release, MiniMax-M1 outperforms several prominent reasoning models in independent benchmark testing, including benchmarks on mathematical reasoning, code generation and general reasoning and knowledge. The MiniMax-M1's leadership is particularly evident in high-value functions requiring complex reasoning, such as automated software engineering, dynamic tool utilisation, and long-context data analysis. The model's design ensures it is not only powerful but also commercially viable for large-scale deployment.

• Software Engineering: Leveraging a training process that incorporates execution-based software engineering environments, our MiniMax-M1 model has achieved high scores on SWE-Bench. This benchmark, which evaluates large language models on real-world software engineering capabilities, confirms that MiniMax-M1 significantly surpasses the performance of most competing open-source models available in the market upon its release.

• Long Context Understanding: Empowered by its 1 million tokens context window, MiniMax-M1 outperforms all other open-source models in long-context understanding. When evaluated on benchmarks such as OpenAI-MRCR and LongBench at the time of its release, MiniMax-M1 ranks second globally, trailing only the top-ranked reasoning model by a narrow margin.

• Agentic Tool Use: In agentic tool-use scenarios, we benchmark MiniMax-M1 against TAU-Bench, which simulates dynamic conversations where agents must leverage API tools while adhering to domain-specific policy guidelines. Tested under such benchmark, MiniMax-M1 outperforms all other open-source models available upon its release. This achievement underscores MiniMax-M1's technological leadership and commercial potential in general AI agent application scenarios.

Upon its release, our MiniMax-M1 model achieved recognized performance. It secured the second position globally among open-source reasoning models and a top-ten ranking among all foundation models globally on the Artificial Analysis Intelligence Index, a third-party leaderboard assessing the intelligence level of foundation models.

The MiniMax-M1 is powered by a hybrid Mixture-of-Experts (MoE) architecture combined with a "Lightning Attention" mechanism, a variation of Linear Attention that we believe can deliver ideal model performance. The MiniMax-M1 model has one of the world's longest context window upon its release, supporting an input context length of up to 1 million tokens and output length of up to 80k tokens. Furthermore, the "lightning attention" mechanism in MiniMax-M1 significantly reduces the computation resources required during model inference stage. These attributes make MiniMax-M1 particularly well-suited for complex tasks that require processing long inputs and deep reasoning.

– 251 –

BUSINESS

除了在综合评测中的优异表现外，我们还通过两项关键创新，大幅提升了MiniMax-M1训练过程的成本效益。我们的第一项突破是CISPO（即截断重要性采样权重策略优化，Clipped IS-weight Policy Optimization），这是一种新颖算法，能够显著提升强化学习的效率，而强化学习是基础模型训练的关键组成部分。对比测试表明，CISPO在性能上优于业内同类强化学习算法。此外，MiniMax-M1的"混合注意力"设计是一种处理训练数据的独特方式，天然有助于强化学习的规模化扩展。得益于上述技术进步，我们仅用三周时间便完成了MiniMax-M1的完整强化训练，所实现的成本显著低于业内大多数同行。

MiniMax-M1已在我们的开放平台上线提供。

MiniMax-M2

2025年10月，我们发布并开源了MiniMax-M2——这是我们专为智能体及代码相关应用场景设计的最新模型。MiniMax-M2通过进一步优化的MoE架构，以显著低于MiniMax-M1的成本和更快的推理速度提供高性能能力，从而推动我们愿景的实现。基于经过优化的激活参数设计，MiniMax-M2在智能性、速度与成本效益之间实现了更优的平衡，为专业和消费级用户提供高度响应迅速、经济高效的AI基础模型。MiniMax-M2在发布时即跻身全球开源模型Artificial Analysis人工智能指数排名第一。

交错思考：提升可靠的智能体推理能力

MiniMax-M2集成了一套先进的推理框架，称为"交错思考"（Interleaved Thinking）。该框架将结构化推理与工具执行融合为"规划—行动—反思"的持续循环。模型不再孤立地完成推理后再行动，而是在推理与行动之间交替进行，使每次工具调用的结果能够指导下一步操作。这一过程使模型能够保持上下文连贯、实时调整策略，并在自主智能体操作、迭代代码开发等复杂多阶段任务中取得更可靠、更具可解释性的结果。通过在交互过程中保持一致的推理状态，"交错思考"增强了MiniMax-M2的稳定性、自我纠错能力与透明度，并支持其实现可靠智能体行为这一更广泛的目标。

– 252 –

BUSINESS Artificial Analysis Intelligence Index¹

68

68 65

GPT-5 Codex GPT-5 (high) (high)

Grok 4

63

61

61

60

Claude 4.5 MiniMax- Gpt-oss-120B Grok 4 Sonnet M2 (high) Fast

60

59

57

57

Gemini 2.5 Claude 4.1 Qwen3 235B DeepSeek Pro Opus A22B 2507 V3.2 Exp

Notes: 1.

As of November 7, 2025, shortly after the release of MiniMax-M2, and compared against the peers' latest publicly released models as of the same date.

2.

According to CIC, Artificial Analysis is a suite of authoritative and independent AI benchmarks widely recognized in the foundation model industry for evaluating models from the perspective of large-model users.

3.

The Artificial Analysis Intelligence Index integrates a comprehensive suite of evaluation datasets to assess language model capabilities across reasoning, knowledge, mathematics and programming, and provides a holistic analysis of overall model capability, including general knowledge, mathematical reasoning and coding ability. According to CIC, the benchmark suite is designed with an emphasis on fairness and real-world applicability, and independently evaluates numerous models using standardized implementations of the constituent evaluation datasets.

MiniMax-M2 Highlights •

Intelligence Performance: According to benchmark results published by Artificial Analysis, MiniMax-M2 demonstrates highly competitive performance across mathematics, science, instruction following, coding and agentic tool use. Upon its release, it ranked first in the Artificial Analysis Intelligence Index among all open-source models globally, reflecting its strong overall cognitive and reasoning capability.

•

Advanced Coding Capabilities: Engineered for end-to-end developer workflows, MiniMax-M2 assists developers in writing, revising and improving code through iterative testing and debugging. The model demonstrates high practical performance across real-world programming environments and multiple programming languages, enabling more reliable software development.

– 253 –

BUSINESS

•

智能体能力：MiniMax-M2展现出增强的智能体能力，能够规划并执行涉及shell、浏览器、搜索及代码执行的复杂、长链路工具调用。在实际应用中，该模型能够持续从复杂数据源中识别相关信息，保持可追溯的推理步骤，并能从不完整或不一致的输入中平稳恢复。上述特性支撑了其在自主及半自主智能体系统中的适用性。

•

效率与成本优化：MiniMax-M2采用先进架构，激活参数约100亿（总参数量2,300亿），旨在以大幅降低的成本实现高性能表现。其效率优化结构可实现更快的推理速度、更低的延迟及更少的算力消耗，同时保持高准确率和推理能力。通过对模型架构及推理流程的持续优化，MiniMax-M2在性能、速度与成本之间实现最优平衡，支持在多元应用场景中的规模化部署。其API定价约为每百万输入token 0.30美元、每百万输出token 1.20美元，约为国际领先模型价格的8%，进一步凸显了其成本优势及面向大规模商业化应用的可及性。

MiniMax-M2现已在MiniMax平台及我们的开放平台（Open Platform）上正式提供服务。

海螺-02：视频生成模型

我们的视频生成模型是多模态产品矩阵的核心组成部分。通过分析和解读多种形式的信息输入，我们的视频生成模型能够依据给定描述生成相应视频。这一转化过程通过融合计算机视觉技术的算法实现，将文本、图像等静态信息转换为动态视觉内容。我们将视频生成模型的研发置于优先地位，这使我们有别于那些主要专注于文本到文本基础模型的竞争对手。

我们于2024年8月首次推出自研的海螺-01（又称Video-01）系列视频生成模型，并于2025年6月推出最新的海螺-02视频生成模型。海螺-02模型发布后，即跻身市场上"同类最佳"视频生成模型之列，这一判断有独立第三方模型排行榜对视频生成模型性能评估结果为证。例如，海螺-02发布后，在Artificial Analysis图生视频基准测试中位列全球第二。

– 254 –

BUSINESS

Arena ELO1 (Image to Video)

1,355

1,331 1,244 1,195 1,144

1,120

1,118 1,064

Seedance 1.0

Hailuo-02 (0616)

Veo 3 Preview (No Audio)

Kling 2.0

Kling 1.6 (Pro)

Runway Gen 4

Veo2

LTX Video v0.9.7 (13B)

1,047

I2V-01Director

1,005

Runway Gen 3 Alpha Turbo

Notes: 1.

As of June 22, 2025, shortly after the release of Hailuo-02, and compared against the peers' latest publicly released video generation models as of the same date.

2.

According to CIC, Artificial Analysis is a suite of authoritative and independent AI benchmarks widely recognized in the foundation model industry for evaluating models from the perspective of large-model users.

3.

The Arena ELO Score is derived from blind user evaluations conducted through the Artificial Analysis Video Arena, in which participants compare outputs generated by different video generation models without knowledge of the model responsible for each output. According to CIC, this benchmarking methodology is widely adopted for providing an unbiased and comparative assessment of video generation model performance to the greatest extent possible.

Hailuo-02 Highlights

We developed Hailuo-02 with a goal to help global creators to fully unleash their imagination, lower creative barriers and increased accessibility. To this end, our team led the development of a proprietary video generation architecture: Noise-aware Compute Redistribution (NCR). This model architecture increases training and inference efficiency by 2.5 times at an equivalent parameter scale, enabling us to increase the model's complexity without raising end-user costs. Consequently, we scaled the Hailuo-02 model to three times the total parameters of its predecessor. Guided by user feedback from our initial Hailuo-01 model series, we also quadrupled the training data volume while significantly enhancing its quality and diversity.

– 255 –

BUSINESS

这一架构的协同效应、扩展的参数规模以及丰富的数据，使模型能力得到显著提升，具备高度的物理准确性及生成高度复杂视频场景的能力：

•

原生1080p及增强视觉质量：Hailuo-02可生成分辨率为1080p、帧率流畅达24fps的高清视频。

•

精准指令遵循：凭借我们在基础模型研发领域的深厚专业积累，Hailuo-02模型在准确遵循用户指令方面同样表现卓越。尤其在处理包含复杂面部表情渲染、镜头运动及动态主体运动等要素的长篇用户指令时，展现出出色的性能，确保生成结果高质量且具有强烈视觉吸引力。为提升用户体验，我们开发了"AI润色"功能，可对用户的初始提示词输入进行优化，并提升所生成视频输出的整体美学质量。

•

复杂物理性能：Hailuo-02已在创建极其复杂的物理交互方面展现出经过验证的能力，涵盖体操、杂技及跳板跳水等运动中极具挑战性的动作。与其他在快速复杂运动中表现欠佳、常导致画面失真（即"体操难题"）的模型不同，Hailuo-02确保人物在完成复杂动作的全程中保持不失真且动作流畅，呈现出卓越的平滑度与准确性。

最新Hailuo-02系列模型：Hailuo 2.3

2025年10月，我们发布了Hailuo 2.3——我们专有视频生成模型系列的重大升级版本，在Hailuo-02的基础上进一步提升了表现力与视觉效果。新模型在人体动作渲染、风格化视觉表达及精细面部动态方面实现了显著增强，同时在响应复杂运动指令时达到了更高的精准度。

•

增强物理真实感与运动控制：通过精细化的物理仿真及更强的指令遵循能力，Hailuo 2.3能够以更优的流畅性与连贯性再现复杂的肢体动作和镜头运动。即便在动态镜头切换的情况下，Hailuo 2.3亦能实现接近照片级真实感的光照、色调及色彩效果。

— 256 —

BUSINESS

•

Broader Artistic and Stylization Capabilities. The model expands support for diverse visual styles, including animation, illustration, ink painting graphics, offering more vivid and stable performance across use cases from entertainment to creative design.

•

Refined Facial and Expression Modeling. Hailuo 2.3 captures subtle human micro-expressions and emotional nuances, producing more natural and expressive facial performances for storytelling, advertising and digital-character creation.

•

Commercial Efficiency and Ecosystem Integration. While substantially enhancing output quality, Hailuo 2.3 maintains the same pricing level as Hailuo-02, redefining industry cost-performance standards. A lighter variant, Hailuo 2.3 Fast, enables faster generation at up to approximately 50% lower batch-production cost.

Hailuo-02 and with its predecessors Hailuo-01 series, power our Hailuo AI, as well as Xingye/Talkie. They are also available on our Open Platform.

Speech-02: Advancing High-Fidelity, AI-Powered Voice Generation

Since our inception, we have been committed to the development of proprietary text-to-speech (TTS) technology, designed to generate natural, high-quality speech from text input. We launched our first Speech-01 model in November 2023, followed by the release of our latest Speech-02 model series in April 2025, with a significant leap in AI-driven voice synthesis capabilities. Our Speech-02 model series includes two models, Speech-02-HD and Speech-02-Turbo, optimized for high-fidelity and rapid real-time processing capabilities respectively.

Following their launch, our Speech-02 model series has established as the "best-in-class" Text-to-Speech (TTS) solutions in the market, as evidenced by independent third-party model rankings that evaluate performance of speech models. To benchmark its performance, for example, the Speech-02 model was submitted to the Artificial Analysis Speech Arena, a public leaderboard that ranks TTS models based on ELO scores, metrics applied to evaluate and rank speech models. Upon its release and as of the Latest Practical Date, our Speech-02-HD model has achieved a competitive position globally, distinguished by its outstanding voice quality.

– 257 –

BUSINESS

Arena ELO1 (Text to Speech)

1,174 1,146 1,132 1,114 1,108 1,103 1,078 1,056 1,056 1,039

Speech-02-HD | TTS-1 HD | TTS-1 | Multilingual v2 | Flash v2.5 | Sonic English | Kokoro 82M | Azure Neural | Polly (Oct '24) | v1.0 Long-Form Studio

Notes: 1. As of June 22, 2025, shortly after the release of Speech-02, and compared against the peers' latest publicly released text-to-speech generation models as of the same date.

2. According to CIC, Artificial Analysis is a suite of authoritative and independent AI benchmarks widely recognized in the foundation model industry for evaluating models from the perspective of large-model users.

3. The Arena ELO Score is derived from user evaluations conducted through the Artificial Analysis Speech Arena, where participants compare the audio outputs of different text-to-speech generation models without knowledge of the model responsible for each output. According to CIC, this benchmarking methodology is widely adopted for providing an unbiased and comparative assessment of text-to-speech model performance to the greatest extent possible.

Speech-02 Highlights

Unlike conventional TTS systems that rely on pronunciation dictionaries, where predefined parameters for sound quality limit the ability to capture contextual nuances, Speech-02 has been trained on high-quality audio data. This extensive training allows the model to autonomously recognize and reproduce subtle distinctions in accent, speech habits, connected speech patterns, syllable elision, and tonal variations, resulting in enhanced fluency and naturalness.

A key innovation in our Speech-02 model is our learnable speaker encoder, which extracts timbre features from a reference audio without requiring its transcription. This enables MiniMax-Speech to produce highly expressive speech with timbre consistent with the reference voice. In addition, the overall quality of the synthesized audio is enhanced through Flow-VAE, a technology that enhances overall audio synthesis quality. Through the learnable speaker encoder and Flow-VAE, Speech-02 refines voice synthesis to closely mimic human

– 258 –

BUSINESS

speech, making it functionally comparable to a native speaker immersed in a real-world linguistic environment. It deeply understands content and context, dynamically adjusts emotional tone, and accurately replicates specialized speech patterns and accents.

To ensure an intuitive and highly adaptive speech synthesis experience, Speech-02 provides extensive customization options, including:

• Word-Level Tone and Pause Control: Achieves precise articulation, allowing users to fine-tune speech nuances, whether handling tongue twisters or multi-pronunciation words.

• Sentence-Level Speed Adjustment: Enables users to modify speech tempo to suit their preferences, whether delivering content at a rapid or measured pace.

• Diverse Voice Selection: Offers over 100 voice templates, allowing users to select or mix tones and create a wide range of styles, ensuring optimal flexibility across use cases.

Our Speech-02 model features multilingual capabilities, fully supporting a wide range of globally spoken languages. Our mission is to break down language barriers and build a truly universal AI model. Supported languages include Chinese, Cantonese, English, Spanish, French, German, Portuguese, Italian, Japanese, Korean and other languages.

Our speech model powers MiniMax Audio, as well as Talkie/Xingye, providing users across our app ecosystem with next-generation text-to-speech capabilities for a wide range of creative and professional applications. Speech-02 is also available on our Open Platform.

Latest Speech-02 Model: Speech 2.6

Building upon the Speech-02 model, we successively launched Speech-2.6 in October 2025, marking upgrades in multilingual capability, timbre realism, and response latency. The enhanced Speech-02 model delivers more natural and expressive speech across over 40 languages, with improved accuracy in pronunciation, rhythm, and emotional tone. Through architectural optimization and streaming inference design, latency has been reduced to near-real-time levels, enabling fluid interaction for intelligent voice agents and real-time communication scenarios. The model also supports professional-grade audio formats and broader deployment flexibility, allowing developers and enterprises to create lifelike, context-aware voice experiences for virtual assistants, audiobooks, and creative applications worldwide.

– 259 –

BUSINESS

Other Models: Music Generation and Image Generation

Music-02

Launched in October 2025, MiniMax Music 2.0 is the latest generation of our music synthesis model designed for music composition, accompaniment, and background audio generation.

Music 2.0 enables users to specify musical style, vocal tone, and instrument settings through natural-language prompts. It supports multiple singing styles and genres such as pop, jazz, rock, and folk, and allows basic control of instrumental layers for richer arrangements. The model also improves overall audio quality, with more natural vocal texture and better spatial balance between instruments.

Music 2.0 is available through our MiniMax Audio platform and Open Platform, providing users, developers, and enterprises with accessible AI-based music creation tools.

Image-01

Complementing our video generation models, our Image-01 model enables cinematic-quality image generation from text prompts. It supports character styling and other creative compositions. Leveraging technologies developed alongside our other models, Image-01 offers high prompt-to-image fidelity, minimizing distortion and preserving artistic intent. We released our Image-01 in April 2023.

Our Image-01 model is currently integrated into Hailuo AI and Talkie/Xingye and available on our Open Platform.

我们的人工智能原生产品

OUR AI-NATIVE PRODUCT OFFERINGS

Leveraging our multi-modal foundation model suite, we deliver AI-native products and services that use the power of AI to benefit both individual users and businesses around the world. Our business model spans consumer subscriptions, token-based in-app purchases, online marketing service and API monetization.

The evolution of our AI-native products is rooted in advancements in the underlying foundation models. Through continuous upgrades to our existing foundation models and the development of new ones, we are able to design and create AI-native products with enhanced user experience.

According to CIC, we are currently distinguished in the Asia-Pacific region as the only pure play foundation model company offering a diverse portfolio of AI-native products with global reach. This recognition underscores our expertise in delivering foundation models across multiple platforms, enabling us to provide high quality, scalable AI tools to users worldwide.

– 260 –

BUSINESS The table below summarizes our major product offerings:

| Product Name | Target Users | Key Underlying Models | PRC Release Dates | Global Release Dates | Primary Use Cases | Monetization Model | |---|---|---|---|---|---|---| | MiniMax | Individual (B2C) | Text, Video, Speech and Music¹ | MiniMax Chat App: March 2025 MiniMax Chat Web: March 2025 MiniMax Agent Web: June 2025 | MiniMax Chat App: March 2025 MiniMax Chat Web: March 2025 MiniMax Agent Web: June 2025 | Intelligent Agent Application | Freemium, subscriptions, token-based in-app purchases | | HailuoAI | Individual (B2C) | Video | Web: August 2024 App: March 2025 | Web: August 2024 App: March 2025 | Flagship Visual Generation Platform | Freemium, subscriptions, token-based in-app purchases | | MiniMax Audio | Individual (B2C) | Speech and Music | Web: May 2025 | Web: March 2025 | Audio Generation Tool | Freemium, subscriptions, token-based in-app purchases | | Talkie/Xingye | Individual (B2C) | Text, Video, Speech and Music | App: September 2023 (As Xingye) | App: June 2023 Web: December 2023 (As Talkie) | Multi-modal Entertainment Platform | Freemium, subscriptions, online marketing service, in-app purchases | | Open Platform | Enterprise and Developer (B2B) | Text, Video, Speech and Music | May 2023 (First PRC customer signed) | September 2024 (First overseas customer signed) | API Platform | Freemium, token-based API billing, subscription (coding plan), enterprise licensing |

Note: 1 Within our MiniMax App, MiniMax Agent orchestrates a spectrum of tools, including our own foundation models and third-party models and resources, to achieve user defined goals.

MiniMax: Intelligent Agent Application

MiniMax is our intelligent AI agent application, which is designed to autonomously perform a wide range of tasks through natural language instructions. Supported by our foundation models, MiniMax Agent can plan, reason, and execute complex actions such as coding, research, document drafting, and presentation creation within a unified workspace.

Within our MiniMax application, users can invoke MiniMax Agent, our proprietary general-purpose agent officially launched in June 2025. Leveraging multiple foundation models, MiniMax Agent is designed to handle a wide range of complex, long-horizon tasks. MiniMax Agent is capable of multi-step planning to devise intelligent solutions, flexibly decomposing task requirements, and executing multiple sub-tasks to deliver integrated end results in multiple formats. MiniMax Agent performs multi-step operations by automatically selecting and coordinating tools such as document retrieval, web-based map queries, and code execution. Tasks are executed in an organized sequence and rendered in a results pane, allowing users to review structured outputs with embedded text, code, and images.

– 261 –

BUSINESS MiniMax – Main Page

Input task requirement

Recommended tasks for starter

MiniMax application offers (i) a Lightning Mode for basic conversational, search and lightweight coding tasks and (ii) a Pro Mode optimised for complex, long-horizon tasks such as development, research, report generation and web design. Users interact with MiniMax Agent through a web-based interface by assigning tasks, such as creating a travel plan. Then our MiniMax Agent will initiate a self-directed reasoning process in which it automatically breaks down the objective into structured steps, selects and coordinates appropriate tools — such as web search, maps, or code execution — and presents interim and final outputs in a task pane. MiniMax Agent also supports integrated tool invocation through the input interface, where users can access a range of built-in and third-party services directly while formulating their tasks. Built-in integrations include Google Maps (for location queries) and the MiniMax MCP protocol, which extends MiniMax Agent's capabilities through a plug-in marketplace. Via the MCP Market, users can activate third-party tools such as Slack, Notion, GitHub, Figma, and MySQL Server. Once activated, these services can be used within ongoing tasks to support project tracking, API calls, or external data ingestion — enabling collaboration and cross-platform functionality. This end-to-end capability allows users not only to automate research or coding tasks, but also to deploy and preview interactive websites — such as fan sites, learning tools, or dashboards — without writing code manually or configuring hosting. Task responses are organized into step-by-step updates, providing clear traceability across planning, execution, and output.

– 262 –

BUSINESS MiniMax – Task Page

Agent working process Results display window

MiniMax Agent supports a broad set of use cases across industries and personal productivity settings, including research and competitive analysis, software and web development, travel and route planning, code execution and data processing.

MiniMax Agent includes built-in safeguards to promote responsible use and protect users. It is programmed to actively refuse assistance with tasks that involve illegal activity, fraud, or harmful behavior. When such prompts are submitted, the Agent explicitly declines to respond and provides a clear explanation that the request cannot be fulfilled.

In addition, all tool usage and output generation are presented with transparent execution traces. For sensitive or ambiguous content, the system displays refusal messages and halts task execution, reinforcing alignment with compliance standards. Each task is handled independently with visible file references, decision steps, and outputs, supporting safe experimentation and traceable workflows.

MiniMax Agent employs a tiered monthly subscription model, featuring the "Basic" plan at US$19.0 per month and the "Pro" plan at US$69.0 per month. These premium subscriptions include exclusive benefits such as peak-hour priority access and early access to beta features. Additionally, users may purchase task-execution credits at US$39.0 per 5,000 credits, which can be applied to execute supplementary tasks via the MiniMax Agent. In addition, MiniMax Agent offers a team plan under which users are charged US$15 per seat per month, providing each team with a shared monthly credit pool and enhanced features.

– 263 –

BUSINESS

Hailuo AI: 旗舰视觉生成平台

Hailuo AI是我们的旗舰视觉生成平台，提供网页端和移动应用程序两种格式，专为实时高清视频和图像合成而设计。Hailuo AI服务于内容创作者、广告商及普通用户，通过文字和图像输入，提供直观的工具，帮助用户创作视觉效果震撼的内容和电影级画面序列。

Hailuo Video – 主页面

集成内容创作中心

精选内容

Hailuo AI的核心在于将我们的专有视频生成模型整合至统一界面，支持用户通过多种输入方式生成视频。Hailuo AI可生成时长6至10秒的电影级质量视频，订阅用户最高可享受1080p分辨率输出。

Hailuo AI的用户可借助我们的视频生成模型，制作具有出色视觉保真度、高质量输出及精准响应性的内容。针对专业级项目，Hailuo AI模型提供高级镜头运动控制功能，可模拟电影导演技术。此外，Hailuo AI还能以增强的角色表现力对上传的图像进行动态化处理，专为将静态图像制作成动画的技术而特别设计。

Hailuo AI为寻求富有表现力的多媒体叙事工具的内容创作者、广告商及普通用户提供了一个创作沙盒。平台界面允许用户输入自然语言提示词，并从预设主题中进行选择。用户可通过调整指令来优化输出效果，包括帧比例、镜头处理方式、运动动态及光照条件。此外，通过主体参考功能，用户可从此前生成的画面中选取某一角色，以在后续场景中保持一致的外观形象和人物性格。

– 264 –

BUSINESS Hailuo AI – Image to Video Creation

Prompt input bar

History creation

Video creation setting

Users may re-render or recreate selected scenes, enabling iterative refinement. The platform also provides a Discover feature with selected content. Hailuo AI employs a tiered subscription model, with plans starting at US$9.99 per month for the "Standard" tier and scaling to US$199.99 per month for the "Max" tier. Premium subscriptions unlock more functionality including 1080p resolution outputs, watermark-free outputs, and priority access to new features. Complementing subscription tiers, users may purchase top-up credits within our Hailuo AI. MiniMax Audio: Advanced Audio Generation Tool MiniMax Audio is the Company's audio generation tool, engineered to provide users with high-fidelity speech and music generation capabilities. Accessible via web platform, MiniMax Audio integrates the Company's Speech-02 model to support interactive audio synthesis. Within the MiniMax Audio platform, users can also generate speech using the Speech-02 model by typing custom text across more than one languages. They can select voices from a curated library of presets with different emotions and styles, adjust pitch and speed, and manage their own voice profiles, and preview/export voice outputs.

– 265 –

BUSINESS MiniMax Audio – Main Page

Text to speech/ music generation bar

AI tools

Selected voices

Harnessing the speech synthesis technology of the Speech-02 model, MiniMax Audio enables users to convert text input into remarkably lifelike speech across multiple languages. The platform offers enhanced accessibility to audio content by allowing users to upload local files or paste web URLs, thereby transforming various text sources, including documents, websites, and ebooks, into their preferred vocalizations.

Leveraging our advancement in model architecture, a key feature of MiniMax Audio is its capability in handling long-form text. The system supports asynchronous speech synthesis for inputs of up to 10,000,000 characters or documents as large as 50MB in a single instance. This significantly streamlines the creation of audiobooks and podcasts by obviating the need for manual segmentation of lengthy texts prior to synthesis.

The monetization strategy for MiniMax Audio incorporates a credit-based consumption model, priced at US$50.0 per million credits. Additionally, users may choose from tiered monthly subscription plans from "Starter" plan priced at US$5.0 per month to "Pro" plan priced at US$99.0 per month. These premium subscriptions offer benefits such as accelerated speech generation, the ability to generate speech with specified emotions and languages, and access to an expanded library of more than 100 distinct voices. Furthermore, monthly subscribers to MiniMax Audio are granted a license for the commercial use of the generated audio.

Talkie/Xingye: AI-powered Multi-modal Entertainment Platform

Talkie (for international markets) and Xingye (星野) (for Chinese domestic market) are emotionally intelligent AI-native multi-modal entertainment platforms designed for real-time human-AI interaction experience. Talkie/Xingye enables users to co-create, customize, and interact with virtual themes and characters that exhibit memory, emotion, and dynamic

– 266 –

BUSINESS

personality. According to CIC, Talkie/星野 ranked top five respectively among global foundation model apps, in terms of average daily usage time in the nine months ended September 30, 2025, with users spending an average of more than 70 minutes per day on these apps.

Chat

Creation Center

Trending

Each user of Talkie/星野 can design their own AI characters by uploading reference photos or inputting textual descriptions, customizing visual appearance, personality traits, and voice styles. The characters retain memory across sessions, enabling continuity and personal bond formation. Users engage with emotionally responsive AI characters powered by our proprietary foundation models. These characters can hold realistic conversations, express emotion through tone, and remember previous interactions. The platform enables extensive personalization through multilingual voice support and customizable avatars. In addition, Talkie/星野 enables "character remixing", where users may generate derivative characters based on existing templates or personas. This has resulted in a highly active creative platform, with over tens of millions AI characters created by users.

Users engage with AI characters through continuous, multi-turn conversations conducted via text or voice. The application supports real-time switching between modalities and incorporates full voice interaction, allowing users to speak directly with characters and hear synthesized responses in natural, emotionally nuanced speech. In addition to free-form chat, users may engage with characters in scripted or thematic "stories", where the dialogue follows a semi-guided narrative arc. These stories can be authored by users or derived from popular templates, supporting structured storytelling alongside open-ended dialogue.

The platforms include a range of safeguards designed to promote responsible use and comply with platform regulations across the jurisdictions in which they are used. The system displays periodic reminder prompts during extended sessions to help users remain aware that they are interacting with a virtual character rather than a human.

– 267 –

BUSINESS

为进一步符合数字环境中青少年保护的监管要求，Talkie和星野均设有青少年模式，对未成年用户实施更严格的管控。该模式启用后，应用程序将在每晚10时至次日早上6时期间无法访问，与针对未成年用户的宵禁政策相一致。在青少年模式下，部分创作功能亦被禁用，用户将被限制搜索、创建、编辑或分享AI角色。上述限制旨在减少屏幕使用时间，体现了我们对负责任AI使用及国家合规标准的承诺。

我们的Talkie/星野平台变现模式多元化，包括在线营销服务、订阅服务及应用内购买。用户可购买应用内积分，用于兑换应用内道具。此外，月度会员订阅提供增强系统功能，例如更多对话回复重新生成次数、增强的对话记忆及更快的响应速度。

此外，我们通过Talkie产生在线营销服务收入，该平台提供整合营销及推广服务。与其他主流移动应用开发商类似，我们采用广告模式，向免费层级用户在使用应用期间展示广告。这些广告可能以应用内横幅、插屏页面或短视频形式呈现，并可将用户重定向至第三方应用程序、网站或产品页面。我们的广告客户主要为第三方程序化广告平台及其开发商，旨在向Talkie用户推广其应用程序、游戏或数字产品。就此类在线营销服务而言，我们主要以效果为基础向第三方广告平台收费，收费标准以有效千次曝光成本（eCPM）计算，即第三方广告平台每千次广告展示所产生的费用。适用费率通过第三方平台广告主之间的实时竞价机制确定，因此具有动态性，eCPM一般约为2美元至10美元。广告活动通过我们的内部广告管理界面集成至Talkie应用中，使我们能够控制广告投放频率和形式，以平衡变现效果与用户体验。与其他移动应用开发商的普遍做法一致，广告仅向免费服务层级用户展示，而高级订阅用户则享有无广告体验。这一方式使我们能够通过用户流量实现变现，同时不对用户参与度或留存率产生实质性影响。广告客户可通过我们的活动管理系统自定义广告投放位置及定向设置。

对于我们运营的其他AI原生产品，由于广告变现模式可能对用户体验和参与度产生不利影响，我们尚未大规模推行基于广告的变现模式。我们仅在用户体验、参与度及收入贡献方面的成本效益分析明确呈现正向结果的情况下，才采用应用内广告。

在线营销服务产生的收入于2024年及截至2025年9月30日止九个月分别为1,460万美元及1,100万美元。该变现策略已形成多元化收入来源，同时保留了免费增值访问模式，维持了较高的月度活跃度。

－268－

BUSINESS

Open Platform

Our Open Platform offers scalable, customizable AI services to global enterprise customers. Through public APIs and services and cloud SDKs, enterprise and developer customers can access the Company's foundation model suite and integrate our foundation model into their own products and services without the need for independent foundation model development. It is one of the world's largest open platforms for enterprises in terms of average daily token volume, and has reached more than 100 thousand enterprise customers and developers, including a range of well-known enterprise customers. International adoption has accelerated rapidly, driven by the Open Platform's competitive price-to-performance ratio and multi-modal offerings. Each day, it processes billions of tokens and powers mission-critical use cases across smart devices, cultural tourism, healthcare, finance, and internet services.

应用场景

互联网

医疗健康

智能设备

文化出版

游戏娱乐

教育

企业服务

在线营销

其他

场景

互动娱乐

生产力工具

增强搜索

智能语音助手

IP角色扮演

增长营销

有声读物

视频创作

……

API接口

文本

模型

大语言模型与视觉语言模型

基础设施

大规模实验平台 自研训练框架

语音

视频

训练阶段

图像

MCP

生成式媒体模型

大规模推理平台

推理阶段

Applications

Internet

Healthcare

Smart Devices

Cultural Publishing

Gaming Entertainment

Education

Enterprise Services

Online Marketing

Others

Scenarios

Interactive Entertainment

Productivity Tools

Enhanced Search

Intelligent Voice Assistants

IP Roleplaying

Growth Marketing

Audiobooks

Video Creation

…

APIs

Text

Models

Large Language Model and Visual Language Model

Infra

Large-scale Experimentation Platform Self-developed Training Framework

Speech

Video

Training Stage

Image

MCP

Generative Media Model

Large-scale Inference Platform

Inference Stage

Our Open Platform provides multiple productized APIs, covering all of our foundation models with additional features:

• Text API: Supports chat completion and batches of text API requests for asynchronous processing. The Text API accesses our language models including MiniMax-M1 and MiniMax-M2, supporting real-time, high-speed responses and long context processing capabilities.

• Speech API: Provides functions including text to speech, voice design and deletion of specified voice, powered by our Speech-02 series.

• Video API: Converts text or images into AI-generated videos with up to 1080p resolution with subject-aware cinematic composition and scene transitions. The Video API accesses our video generation models including the Hailuo-02 and Hailuo-01 series.

• Music Generation & Image Generation APIs: Allow users to create stylized images or music based on prompts or uploaded references, supporting personalized creative workflows.

– 269 –

BUSINESS •

MCP（模型上下文协议）API：一种标准化编排协议，使第三方工具和服务能够调用我们的专有基础模型。通过兼容MCP的插件，开发者可以创建代理任务流，结合我们的基础模型执行各种形式的任务。

开放平台的收入主要通过基于使用量的定价模式产生，按"token"（文本单位）、"视频片段"（视频单位）或"字符"（音频输入单位）收取费用。订阅套餐提供固定费用访问权限，从用于测试的入门级方案到具有增强功能的高级选项不等。为进一步扩大开发者采用范围，我们于2025年11月推出了MiniMax M2编程套餐，提供专为编程和软件开发场景设计的分级订阅方案。该套餐允许个人开发者通过统一的API接口便捷访问我们的基础模型能力（尤其是MiniMax M2），无需独立搭建基础设施。编程套餐已完全集成至我们的开放平台，旨在鼓励个人及机构开发者进行实验、原型开发和小规模部署。它与我们开放平台的企业订阅框架相辅相成，支持我们培育充满活力且全球互联的开发者生态系统的战略。我们还为企业客户提供其他基于AI的企业服务，主要包括根据企业需求定制的安排及许可交付物。对于定制化安排，我们与企业客户合作，为其量身搭建专属推理资源池，以确保稳定且可预期的模型推理性能。对于许可交付物，我们向客户授权基础模型，使其能够在自身系统中部署和运营此类模型。 我们的开放平台展示出强劲的变现能力，以及用户对所提供基础模型的高度粘性。我们持续观察到开放平台付费用户数量同比增长——付费用户定义为单独消耗API调用金额不低于50美元（或等值其他货币）的用户——从截至2024年9月30日的九个月内的约400名，增长至2025年同期的约2,500名。 为提升用户体验并改进我们的产品，用户向我们和/或我们的关联方授予有限许可，以在适用法律的限制范围内使用相关内容。该许可在必要时可包括转许可或再许可权利。它使我们能够仅将内容用于合法目的，包括：（i）产品和服务的提升与优化；以及（ii）品牌推广和合法营销活动。

– 270 –

BUSINESS OUR CORE TECHNOLOGIES Highlight of Model Algorithm Innovation MoE Architecture In response to the limitations identified within the then-current AI model architectures, we undertook a strategic initiative in 2023 to accelerate the development of models leveraging the Mixture-of-Experts ("MoE") architecture. We chose the MoE architecture due to its demonstrated ability to achieve greater scale and operational efficiency than conventional model architectures. Unlike traditional "dense" model architectures, which activate all experts (parameters) for every input, the MoE architecture features multiple sub-networks of "experts" and a gating mechanism that dynamically routes each token to a small subset of them. This approach allows the model's total parameter count to scale dramatically without a proportional increase in computational cost. As a result, MoE models can achieve the performance of a much larger dense model while maintaining significantly lower computational demands and inference costs per processed token. A significant milestone in this strategy was the launch in January 2024 of our abab 6.0 model, the first commercialized text model in Asia to adopt a Mixture-of-Experts (MoE) architecture, according to CIC.

Linear Attention With our expertise in AI model architecture, we recognized that conventional large language models face an inherent computational bottleneck in their core attention mechanism when processing long sequences inputs. The computational and memory costs of conventional mechanism scale quadratically with input length, making it prohibitively expensive and inefficient for tasks requiring long context, such as comprehensive document analysis, large-scale software development assistance, or learning from numerous examples at once. This scalability challenge is the key barrier limiting the application of current foundation models in more complex scenarios.

We have successfully addressed this challenge through the development of our proprietary "Linear Attention" mechanism, which represents a significant advancement in artificial intelligence technology, particularly for large language models. Traditional Transformer-based architectures experience performance degradation and escalating computational costs as input lengths increase. By contrast, our Linear Attention algorithm efficiently processes contexts up to four million tokens — dramatically surpassing conventional architectures, which typically manage hundreds of thousands of tokens.

The core advantage of our Linear Attention lies in its computational efficiency. By simplifying the calculation, it reduces both memory footprint and computational overhead, which in turn boosts the model's inference speed and throughput. Consequently, this improvement in efficiency means we can not only tackle extremely long-context tasks but also deliver powerful AI capabilities to our customers at a competitive price-to-performance ratio.

– 271 –

BUSINESS

Independent benchmarks validate the strength of our algorithm, with our models consistently ranking among the top global performers in widely recognized tests such as AIME, LiveCodeBench, SWE-bench, TAU-bench and OpenAI-MRCR.

Clipped IS-weight Policy Optimization (CISPO)

CISPO represents a reinforcement learning algorithm. It is designed to enhance the stability of foundation model training process and accelerate its convergence — the point where a foundation model achieves optimal performance. Traditionally, foundation models employ "clipping" techniques to stabilize model training by removing tokens that introduce instability during model training. However, this process risks inadvertently discarding valuable learning signals. CISPO innovates by "clipping" the importance sampling weights — scores indicating data importance — instead. This ensures all data, even minor elements, contribute to the core learning mechanism. This technique prevents the loss of crucial learning signals, thereby fostering a more reliable foundation model training process.

CISPO's efficiency is underscored by its ability to achieve performance comparable to established algorithms like GRPO and DAPO, while requiring approximately half the training iterations — or model learning cycles — in real-world tests. This innovation holds practical implications for developing foundation model training processes, reducing the required computational resources and time associated with such processes.

Our AI Infrastructure

From the moment of our inception, we established an in-house infrastructure team and began developing our proprietary training and inference framework. Our self-developed AI infrastructure offers a comprehensive and flexible model training and inference solution through experimentation and optimization strategies, solid parallel and scalability capabilities, and automated operational support. We not only possess technological advantages but also ensure that each training and inference task can be executed in a stable environment, thereby maximizing return on investment. Our framework not only holds advantages in the present but also provides a solid foundation for future expansions and upgrades of our foundation model offerings. Below are key highlights of our self-developed AI infrastructure:

High-Efficiency Training and Inference Framework

Our training and inference framework is the cornerstone of our technical capabilities, designed to optimize foundation models' performance at every level of the computational stack. At the operator level, the atomic computational building blocks that form the model training and inference algorithm, we have engineered deep optimization to significantly enhance computational efficiency, reduce latency, and improve utilization of computing resources. Our training and inference framework features advanced parallelism strategies, multi-level key-value caching and disaggregated expert parallelism inference architecture. Together, these measures yield a stable and high-performance model training and inference environment tailored for our foundation models.

– 272 –

BUSINESS

统一训练与推理计算资源

最大化大规模计算集群的利用率——即专为支持训练和推理工作负载而设计的大型互联计算资源——需要打破不同工作负载之间的壁垒。我们的"统一训练与推理计算资源"策略能够跨越一系列任务动态分配资源，涵盖从低优先级数据处理任务到高优先级训练及推理活动。

我们实施了一套智能调度系统，可持续监控计算集群及工作负载需求，将离线作业（如数据处理或模型评估）视为可抢占的低优先级任务。当高优先级训练或推理作业到达时，调度器可自动暂停上述离线任务，回收其资源，并实时将其重新分配给高优先级任务。这种基于抢占的动态方法确保对时间敏感、面向用户的活动始终拥有所需的计算资源，同时利用低优先级任务回填潜在的可调度空隙。这种动态方法有效填补了空闲间隙，使我们能够在不影响计算集群整体性能的情况下最大化计算资源利用率。

跨集群负载均衡

我们采用先进的方法构建和管理多个计算集群，包括有效的资源切换方法，以确保模型推理过程的平稳运行。为确保高可用性并支持大规模部署，我们的基础设施专为多集群扩展而设计。我们通过实施自适应策略体系并结合实时负载计算与反馈回路，克服了跨集群负载均衡和状态同步的固有挑战。这使得跨多个集群的智能动态切换和资源分配成为可能，确保了企业级人工智能应用所需的高可用性、容错能力及灵活的资源扩展能力。

我们的组织架构与人才

我们独特的组织架构推动了我们的可扩展性，使研发进程得以快速迭代推进。自入职第一天起，员工便沉浸于一种赋能各层级人才的企业文化之中。我们重视具备第一性原理思维、致力于通过跨领域创新实现卓越成果的人才。在我们富有远见的创始人兼首席执行官严骏驰（Dr. Junjie Yan）博士的带领下，我们在前沿阵地上运营，推动人工智能从研究走向部署。

— 273 —

BUSINESS

我们扁平灵活的组织架构

我们的组织架构刻意保持扁平灵活，首席执行官以下设置不超过三个层级，并围绕基于项目的任务而非僵化的部门壁垒来组建团队。我们相信，这种组织架构有助于加快决策速度，并加速智能基础模型的迭代。除部门负责人外，我们的团队不设正式职级。这有助于促进部门内部及跨部门之间的协作，并消除冗余的汇报层级。

我们的企业文化与招募实践

我们通过营造包容性环境来促进研究创新，重视多元化的贡献与持续改进。我们鼓励员工突破初始岗位职责的局限，积极支持他们拓展技能、主动承担更广泛、更多元的挑战。我们赋予同事早期承担责任的机会，并在其展现能力时授予其实质性权力；许多直接从高校招募的员工现已主导核心研发工作。

我们研发团队成员的平均年龄不足30岁，充分体现了我们致力于为有能力、有干劲的人才提供机会的理念，以推动团队与个人的快速成长。

基本信息

截至2025年9月30日，我们共有全职员工385名。我们的团队中绝大多数人员从事研究与开发工作。下表列示了截至2025年9月30日，按职能划分的员工人数。

| 职能 | 员工人数 | |------|--------| | 研究与开发 | 284 | | 管理及综合行政 | 60 | | 销售与市场营销 | 41 | | 合计 | 385 |

我们绝大多数员工在中国境内工作。我们与员工保持良好的劳动关系，未曾发生重大劳动争议。

根据中国法律法规的要求，我们参加了由地方市级及省级政府组织的各类员工社会保障计划，包括养老保险、生育保险、失业保险、工伤保险、医疗保险及住房公积金。

— 274 —

BUSINESS

We enter into standard employment contracts and agreements with our executives and full-time employees, covering confidentiality, non-competition, intellectual property rights, employment terms, and business ethics. These contracts generally include non-competition clauses effective during employment and for up to two years post-employment, as well as confidentiality clauses effective during and after employment.

We provide comprehensive onboarding, continuous training programs, and mentorship support to facilitate employee development. Our compensation structures, including equity incentives, are competitively designed to attract and retain top talent. We regularly organize internal knowledge-sharing sessions where employees are invited to discuss industry trends, products, and technologies, enhancing our team's professional skills and knowledge base.

RESEARCH AND DEVELOPMENT

Overview on Model Training

| Training Process | Model Pre-training | Model Post-training | |---|---|---| | | Training process of our model using processed data | Supervised Fine-Tuning (SFT) and large-scale Reinforcement Learning (RL) |

| Key Elements for Training | High-Performance Computing Cluster Construction | Curate and Pre-process Data | Model Architecture Design | |---|---|---|---| | | Robustness and Efficiency | Quality and Variety | Innovation and Efficiency |

Our approach to developing our foundation models is a multi-phase endeavor designed to maximize performance, efficiency, and scalability.

• High-Performance Computing Cluster Construction: From the outset, we have led the design of high-performance computing cluster that integrates our proprietary training framework and model training strategies. Such computing cluster enables the training of our foundation models across various modalities.

• Curate and Pre-process Data: The model training process begins with the careful curation and organisation of massive volumes of available data sourced from a wide range of domains. Prior to use, all data undergoes a rigorous pre-processing pipeline to ensure quality and variety. Key steps include cleansing erroneous data and selecting high-quality data for training. We collaborate with third-party providers for data storage and database services.

– 275 –

BUSINESS

•

模型架构设计（Model Architecture Design）：数据准备完成后，模型架构的设计成为关键重点。通过大量实验和反复迭代优化，我们构建出不仅能提供更优性能，同时在训练和推理效率方面也经过优化的架构。我们采用了MoE（混合专家）架构，该架构通过其高效的计算设计，在提升预测精度的同时，兼顾提高训练和推理效率。

•

模型预训练（Model Pre-training）：在模型预训练阶段，我们持续监控和评估模型性能，使用一套全面的指标来衡量学习进度。这包括针对独立验证集测试模型的泛化能力及各类下游任务能力。我们采用多阶段训练方法，动态调整训练策略和预处理数据的分布，直至达到预期的性能基准。该阶段计算密集，需要处理海量数据，并依托我们先进的训练基础设施予以支撑。

•

模型后训练（Model Post-training）：预训练完成后，我们利用精心构建的特定任务数据集，通过对齐技术进一步增强和塑造模型能力。通过采用监督微调（Supervised Fine-Tuning，SFT）和大规模强化学习（Reinforcement Learning，RL）等技术，我们激发模型的潜在能力，赋予其新的能力，并提升其在特定任务上的表现。这显著提升了模型在下游应用中的效能，以及与人类偏好的对齐程度。

•

模型部署（Model Deployment）：训练完成后，经过完整训练的模型被部署至我们的计算集群，根据用户输入执行一系列生成任务，包括文本、视频和音频生成。这些能力被集成到我们的自有产品中，实现跨多种格式的高质量多模态输出。

AI安全与对齐（AI Safety and Alignment）

我们采用贯穿基础模型从开发到部署整个生命周期的全面AI安全与对齐方法。这一多层次、纵深防御策略旨在增强我们产品拒绝协助执行违法或有害任务的能力。上述措施的有效性通过数据驱动的迭代和人工监督持续提升，从而增强我们基础模型及AI原生产品在拒绝协助违法或有害任务方面的能力。

– 276 –

BUSINESS

总体而言，我们的人工智能安全与对齐措施在以下阶段实施：

•

数据整理阶段：我们从一开始便将安全性直接嵌入基础模型，重点防止有害内容的摄入。用于训练活动的数据须经过标注、过滤和去毒处理，通过自动化工具结合人工审核来识别和删除有害内容。

•

模型训练阶段：在数据整理阶段之后，我们开展大量模型训练，包括微调和强化学习，以进一步提升模型安全性，并使其输出与人类价值观保持一致。本阶段包括训练基础模型识别和拒绝有害意图，所采用的技术包括以下内容：

•

o

监督微调与对齐：在经过整理的数据集上对模型进行微调，优先考虑安全响应。例如，模拟违法活动的提示词与拒绝响应相配对。对齐数据集还涵盖多样化场景，以覆盖广泛的潜在有害活动。

o

强化学习增强：采用先进的强化学习方法来强化安全且符合对齐要求的行为。具体包括：（i）我们的基础模型因生成安全、有益的输出而获得奖励，同时对有害输出予以惩罚；（ii）在训练过程中，不安全的生成内容被采样并被明确拒绝，从而强化基础模型的拒绝行为。

模型部署阶段：在模型部署阶段，我们实施运行时防护措施——即产品在使用过程中处于激活状态的主动机制——以在其人工智能原生产品的实时交互中实现现实世界的安全保障、监控和干预。本阶段重点关注动态拒绝机制及持续管理，以处理涉及有害行为的用户输入。我们已部署的基础模型利用分类器扫描输入内容中的有害意图（例如，可能涉及违法或不当内容的提示词（如涉及危害安全、色情或暴力的内容），或其他违背公序良俗的内容（如涉及自我伤害的内容）），并对其进行屏蔽或重定向。同样，特定输出内容也受到审核管理。

防止不当或有害输出及用户滥用

我们采取技术与行政措施相结合的方式，防止我们的人工智能基础模型或产品生成不当、有害或操纵性内容，并防止用户从事有害或违法行为。

— 277 —

BUSINESS

From the perspective of model safety, we maintain a high level of investment and R&D intensity and embed safety governance requirements throughout model and product development. At an early stage, we conduct risk identification and assessment, manage risks associated with inappropriate or harmful content, and introduce preventive mechanisms during the model design stage to avoid the generation of harmful outputs.

We have established a dual mechanism combining automated and manual review to detect and handle illegal or harmful information, thereby preventing the generation or dissemination of inappropriate or harmful content. In terms of user compliance, we inform users of prohibited behaviours under our community rules, specify penalties for unlawful use, and maintain clear complaint and reporting channels with defined response timelines. We promptly handle and respond to user misconduct and related reports.

Continuous Measures for AI Safety and Ethical Development

We have implemented a series of technical and administrative measures to ensure AI safety and ethical development and to control related safety risks. We aim to develop leading foundation models through sustained investment and high-intensity R&D, incorporating safety governance requirements across the full lifecycle of model development and operation.

At the model design stage, we perform graded risk assessments, particularly for high-risk areas such as mental health or self-harm, and introduce specific preventive mechanisms to avoid outputs that could cause psychological distress to users. During operation, we maintain continuous monitoring and intervention mechanisms for real-time detection and screening of potentially harmful outputs. The system automatically triggers warnings, content redirection or manual review, and escalates high-risk interactions to our safety team for timely intervention.

For content moderation, we apply a dual-layered mechanism of automated and manual review, supported by a dynamically updated database of illegal and harmful information samples to enhance moderation efficiency and accuracy. We further verify compliance through random testing. We also specify in our user agreements the penalties for unlawful use of our generative AI services and provide effective complaint and reporting channels. We have established a real-time content moderation system that continuously evaluates both user inputs and model-generated outputs against a multi-layered safety framework. "Harmfulness" is identified and benchmarked based on (i) applicable laws and regulatory requirements, and (ii) internal risk-classification guidelines informed by operational experience, user feedback, and incident review data accumulated over time. These standards define harmful content to include, for example, illegal or non-compliant information, content that endangers public safety or social stability, explicit sexual or violent content, or content involving self-harm or other acts contrary to public order and good morals.

– 278 –

BUSINESS

Based on the design of our safety architecture, content moderation, and benchmarking methodology, the Directors are of the view that the measures adopted are adequate and effective as well as consistent with industry standards, given (i) the presence of automated monitoring tools capable of interception of potentially harmful content, (ii) defined harmfulness standards derived from applicable laws, regulatory guidance and internal experience, (iii) a structured escalation pathway supported by dedicated personnel, and (iv) continuous improvement of classifiers and risk thresholds through performance evaluation and user feedback.

In addition, we have established a regular model safety assessment and improvement mechanism. Through continuous monitoring of system performance in real-world applications, we periodically analyse the rate of inappropriate outputs, user feedback and manual review results to quantitatively evaluate and enhance the effectiveness of our safeguards.

Our R&D Team

As of September 30, 2025, we have built an R&D team of 284 members, representing approximately 73.8% of our total employees. Our R&D team is structured into specialized groups focused on text, video, audio models, AI infrastructure (training and inference), and product development. Our core R&D team comprises experts formerly with global AI leaders such as Microsoft, Google, Meta, Alibaba, ByteDance and DeepSeek. The table below sets out the profiles of our core R&D team members:

| Core R&D team member | Profile | |---|---| | Dr. Junjie Yan (闫俊杰) | As the leader of our R&D team and our CEO, Dr. Junjie Yan brings more than a decade of R&D experience and has been integral to the success of the Company, materially contributing to its founding and growth. With profound technical insight and deep understanding and knowledge of general artificial intelligence technology, Dr. Yan laid the foundation for MiniMax and was critical in shaping the Group's long-term strategies for the R&D and operations of the Group over the years, in particular with respect to the adoption and advancement of technology innovations such as MoE and Linear Attention, which are crucial to the industry. |

– 279 –

BUSINESS Core R&D team member

Profile

Dr. Yan obtained a bachelor of mathematics in Southeast University (東南大學) in the PRC in June 2010. He then obtained a doctorate degree in the area of artificial intelligence in the Institute of Automation, Chinese Academy of Sciences (中國科學院自動化研究所) in July 2015 and conducted post-doctorate research at Tsinghua University. Dr. Yan has published approximately 200 academic articles on top conferences and journals with over 30,000 citations and won several awards and honors. Dr. Yan was awarded (i) the First Prize in the Guangdong Province Technology Invention Award (廣東省技術發明獎) in February 2020; (ii) the Wu Wenjun Artificial Intelligence Natural Science Award (吳文俊人工智能自然科學獎) in October 2019; (iii) Wu Wenjun Artificial Intelligence Technology Progress Award (吳文俊人工智能科技進步獎) in October 2019; (iv) 2024 Shanghai Oriental Talents Program (上海市東方英才計劃) in December 2024; and (v) Senior Professional Title (正高級職稱), a professional title generally denoting individuals who have outstanding achievements in their respective technical or professional fields, by the Shanghai Municipal Professional Title Evaluation Committee in February 2025.

Mr. Pengyu Zhao (趙鵬宇)

Our executive Director and a large language model research and engineering leader. Mr. Zhao joined our Company as a natural language processing researcher and engineer since August 2023. He is primarily responsible for the research and development of large language models.

– 280 –

BUSINESS Core R&D team member

Profile

Mr. Zhao obtained his bachelor's degree and master's degrees in computer science and technology from Peking University (北京大學) in the PRC in July 2017 and July 2020, respectively. Prior to joining our Company, Mr. Zhao served as a research software development engineer in Beijing Hulu Technology Co., Ltd. (北京葫蘆科技有限公司), a company mainly engaged in the research and development of streaming media technology, between August 2020 and July 2023, where he was primarily responsible for recommendation algorithms. Mr. Zhao has extensive experience in algorithm-related research and development and has published several papers covering fields such as neural networks and reinforcement learning.

Mr. Yucong Zhou

Our executive Director and a visual model research and engineering leader. Mr. Zhou joined our Company as a visual model researcher and engineer since March 2022. He is primarily responsible for research and development of visual models.

Mr. Zhou obtained his bachelor's degree in math and systems science and master's degree in computer science from Beihang University (北京航空航天大學) in July 2015 and March 2018, respectively. Prior to joining our Company, Mr. Zhou gained extensive research and development experience in computer vision, automated machine learning, and AI training system design. He has also authored papers in the fields of neural networks and deep learning.

Mr. Pengyu Zhao and Mr. Yucong Zhou, being members of the core R&D team, do not hold shares in the Company. The salient terms of agreements with management and R&D staff are set out below:

• Non Conflict of Interest. During employment, the employee is prohibited from engaging in employment, whether full-time or part-time, with third parties without our consent, especially in companies or roles that compete with our business.

• Proprietary information arrangement. All our proprietary information, including all business, technical and financial information that the employee learns or develops during their employment, shall be kept confidential and only used for our business purposes.

– 281 –

BUSINESS

• Confidentiality. Employees shall maintain the confidentiality of our technology secrets, trade secrets and any confidential business information both during and after employment.

• Non-Competition. We have the right to enforce a non-competition period of up to 24 months after the termination of employment. During this period, the employee is not allowed to work for competitors listed in the agreement, or start competing businesses.

We understand that our sustained ability to innovate and compete effectively in a dynamic market hinges on attracting, nurturing, and retaining talent. Recognizing that our human capital is paramount to our future success and sustainable growth, we place a strategic emphasis on both talent recruitment and retention across all levels of our organization, particularly within our R&D functions.

To cultivate a stable and high-performing R&D team, we employ a multi-faceted approach. This includes fostering a compelling shared vision that deeply aligns our key R&D management and technical staff with our overarching mission and future objectives, thereby cultivating a profound sense of purpose and connection. We are committed to offering challenging and stimulating projects that not only encourage continuous learning and professional growth but also move beyond routine tasks to embrace complex endeavors. Furthermore, we strategically implement long-term incentives, such as competitive stock-based incentive packages and performance-based bonuses, designed to directly align the financial interests of our key personnel with our sustained corporate success, thereby fostering a sense of ownership and commitment.

To proactively mitigate any potential disruption resulting from key employee departures, we continuously develop a comprehensive talent pipeline as a cornerstone of our talent management strategy, ensuring that capable individuals are consistently ready to step into critical roles. In the event of a key employee's departure, rigorous hand-over procedures are immediately initiated, guaranteeing the continuation of ongoing projects and responsibilities. Additionally, to safeguard our proprietary information and competitive advantages, we diligently execute and enforce non-compete and confidentiality arrangements, preventing former employees from leveraging company knowledge to a competitor's benefit.

During the Track Record Period and up to the Latest Practicable Date, the Company does not have any legal claims or proceedings that may have an influence on its R&D for any of the Company's products. In addition, to our best knowledge, all executive Directors and all key R&D employees of the Group have not violated any non-compete agreements with their previous employers.

– 282 –

BUSINESS 研究与开发流程

我们的研发流程旨在促进创新、确保技术卓越，并推动我们基础模型及AI原生产品的商业化。该流程以严格的实验文化和第一性原理决策为基础，通常包括以下几个关键阶段：

• 概念阶段：该初始阶段专注于对新概念的初步探索及潜在架构框架的构思。我们的研发团队开展初步可行性研究，目标是识别符合我们战略愿景和市场机遇的新型基础模型架构或AI原生产品功能。在此阶段，高层次架构设计得以概念化，为后续开发奠定必要基础。

• 规划阶段：在概念阶段所获洞见的基础上，规划阶段涉及对拟议基础模型或AI原生产品的具体研发方向进行全面记录、严格迭代和最终确定。我们的研发团队准备详细的研发文档，通常包括项目范围、技术规格、资源分配、初步时间表及成功指标。通过协作讨论和同行评审，创意得到优化，潜在研发挑战得以识别，并就项目方向达成明确一致的共识。该阶段确保在投入大量资源进行开发之前，所有相关研发团队成员均已明确路线图并保持一致。

• 实验阶段：在确认研发方向后，实验阶段专注于广泛的实践性实验。我们的研发团队成员开展一系列综合性实验，包括但不限于对不同算法进行原型设计、测试各种模型架构以及验证各类假设。这一迭代过程涉及快速性能基准测试及对实验结果的深入分析，以收集关键数据。该阶段所获得的洞见对于指导研发决策、优化模型性能及降低潜在技术风险至关重要。

• 开发与优化阶段：在开发与优化阶段，实验阶段的研究成果被用于确定最优技术方案，并推进全面的基础模型训练或产品开发。我们的研发团队对实验结果进行分析，以确定最有效的算法、模型架构或产品设计。对于基础模型，这涉及利用AI基础设施投入大量算力资源进行模型训练。对于AI原生产品，这包括全面开发和综合内部测试，以确保产品的功能性、可扩展性和性能，为最终发布做好准备。

– 283 –

BUSINESS

•

发布与商业化阶段：最后阶段涵盖从基础模型及AI原生产品定稿到市场发布及后续商业化的严格步骤。这包括全面的质量保证、安全审计、合规检查及用户验收测试，以确保产品或模型符合我们的严格标准及监管要求。在成功完成所有上市前工作后，所开发的AI原生产品或基础模型将战略性地推向市场并整合至我们的商业运营中，同时持续进行监控、上市后支持，并根据用户反馈及市场表现进行迭代改进。

我们所有基础模型及AI原生产品和服务均作为研发活动的组成部分由内部自主研发。我们对上述研发活动产生的所有知识产权拥有完整所有权。我们的基础模型及AI原生产品的研发不依赖任何重大的经许可引入的第三方技术。在业绩记录期间及截至最后实际可行日期，我们未将任何研发活动外包予第三方，亦未与第三方开展对我们研发运营具有重大意义的合作。

知识产权

知识产权是我们研究、产品开发及商业成功的核心所在。我们通过分层策略保护专有技术，该策略包括：(i) 依据中国及其他司法管辖区的专利、商标、版权、商业秘密及不正当竞争法律所提供的法定保护，以及 (ii) 合同保障措施，包括保密承诺、发明权属转让契约及许可协议。所有雇佣合同及主要商业合同均明确界定在我们业务过程中创建或使用的知识产权的归属及保护义务。在业绩记录期间，我们的核心技术已获得专利保护。该等专利的有效期通常为10至20年。

截至最后实际可行日期，我们已在中华人民共和国国家知识产权局登记注册专利75项，在中国注册商标144件，在中华人民共和国国家版权局登记注册版权73项，在中国注册战略性域名28个，并在国际范围内注册商标87件。有关重要知识产权的明细，请参阅"附录四——法定及一般资料——B. 本集团业务的进一步资料——2. 本集团的知识产权"。

除已登记注册的知识产权外，我们在基础模型的训练、微调及推理过程中广泛依赖内部开发的、未申请专利的专有技术诀窍及专有商业秘密。为保护上述技术诀窍及商业秘密，我们已实施保密规程、限制对敏感信息的访问，并与参与模型开发的员工签订保密协议。

— 284 —

BUSINESS

We accord importance to intellectual property rights and has established a rigorous IP protection and management framework designed to safeguard both our proprietary intellectual property and that of third parties. This framework encompasses the acquisition of authorised training data from reputable providers, coupled with adherence to prevailing intellectual property legislation. For open-source datasets, we review licensing terms and related documentation to verify data provenance and ensure that our use is consistent with the applicable open-source licence conditions. For commercially procured datasets, we enter into data procurement agreements with suppliers and require them to provide representations and undertakings regarding the legality and provenance of the data supplied. In addition, we conduct quality control and validation through assessment, sampling-based review and other evaluation measures to ensure data quality and compliance. Furthermore, our term of use explicitly stipulate that users are prohibited from infringing the intellectual property rights of any third party and obliged to warrant that any content submitted or uploaded by them does not infringe upon the intellectual property rights of any third party. Upon receipt of any allegations pertaining to unauthorised user-generated inputs or outputs, we will investigate and implement appropriate measures in accordance with applicable law, such measures may include, among other things, the deletion of infringing content or the suspension or termination of infringing user accounts. During the Track Record Period and up to the Latest Practicable Date, we were not involved in any IP litigation, arbitration or administrative proceedings, nor have we received any claim alleging infringement of third-party rights that would have a material adverse effect on our business, results of operations, or financial condition. Our Directors confirm that they are not aware of any material legal, arbitral or administrative proceedings of infringement of any third parties' intellectual property rights by us as of the Latest Practicable Date. We will continue to monitor the landscape and, where necessary, defend or enforce our rights vigorously.

Notwithstanding the foregoing measures, we cannot rule out the possibility of challenges to our IP or allegations of infringement against us. For example, on September 16, 2025, a group of major U.S. movie studio companies, including Disney, Universal and Warner Bros. Discovery (the "Plaintiffs"), filed a civil complaint (the "Complaint") in the United States District Court for the Central District of California, against our Group in relation to Hailuo AI, our visual generation platform. See "Business — Legal Proceedings and Compliance."

Enforcement actions may involve significant cost and management distraction. For a discussion of these and other related risks, please refer to "Risk Factors — We may not be able to adequately protect or enforce our intellectual property rights throughout the world, and our efforts to do so may be costly" and "Risk Factors — We may become subject to litigation brought by third parties claiming infringement by us of their intellectual property rights."

To mitigate risks relating to potential infringement of intellectual property rights, we have implemented and continue to enhance a series of internal control and compliance measures, including the following:

– 285 –

BUSINESS

•

Enhanced internal awareness and communication on intellectual property matters. We have strengthened internal communications to reinforce the importance of intellectual property compliance across relevant teams, including management, research and development, product, and operations, with a view to promoting awareness of applicable intellectual property laws and our internal compliance expectations.

•

Establishment of internal standard operating procedures for intellectual property complaints. We have formulated internal standard operating procedures to govern the identification, escalation, review and handling of potential complaints or allegations relating to intellectual property infringement, including coordination among relevant internal departments and, where appropriate, engagement of external legal advisers.

•

Implementation of technical filtering and screening measures. We have implemented, and continue to refine, technical measures designed to identify and restrict certain inputs and outputs on our platforms that may potentially involve or implicate third-party intellectual property rights, including keyword-based filters and other screening mechanisms, with the aim of reducing the risk of generating content that may infringe intellectual property rights.

The table below sets forth our key IP rights and their respective technological significance to us as of the Latest Practicable Date:

| No. | Name of Patent | Type | Covered Region | Registered Owners | Related Specialist Technology Product | Patent Registration Number | Date of Filing | Date of Grant | Expiry Date | |-----|----------------|------|----------------|-------------------|--------------------------------------|---------------------------|----------------|---------------|-------------| | 1 | Video generation method, apparatus, system and computer-readable storage medium | Invention Patent | Mainland China | Shanghai Jizhi | Video generation model | ZL202211231054.X | October 9, 2022 | July 21, 2023 | October 9, 2042 | | 2 | Video generation method, apparatus, system and computer-readable storage medium | Invention Patent | Mainland China | Beijing Jizhi; Shanghai Jizhi | Video generation model | ZL202211226180.6 | October 9, 2022 | August 29, 2023 | October 9, 2042 | | 3 | Timbre mixing method and apparatus; audio processing method, apparatus, electronic device and storage medium | Invention Patent | Mainland China | Shanghai Jizhi | Speech generation model | ZL202311864508.1 | December 29, 2023 | August 27, 2024 | December 29, 2043 |

– 286 –

BUSINESS

| No. | Name of Patent | Type | Related Specialist Technology Product | Patent Registration Number | Date of Filing | Date of Grant | Expiry Date | Covered Region | Registered Owners | |-----|---------------|------|--------------------------------------|---------------------------|----------------|---------------|-------------|----------------|-------------------| | 4 | Speech synthesis model training method; speech synthesis method, apparatus, electronic device and storage medium | Invention Patent | Speech generation model | ZL202311870114.7 | December 29, 2023 | December 17, 2024 | December 29, 2043 | Mainland China | Shanghai Jizhi | | 5 | Speech synthesis, speech recognition methods, training method, apparatus, electronic device and storage medium | Invention Patent | Speech generation model | ZL202311873032.8 | December 31, 2023 | December 13, 2024 | December 31, 2043 | Mainland China | Shanghai Jizhi | | 6 | Speech and singing synthesis method, training method, apparatus and model | Invention Patent | Speech generation model | ZL202410672187.3 | May 28, 2024 | November 22, 2024 | May 28, 2044 | Mainland China | Shanghai Jizhi |

We also own registered trademarks for our company name, AI-native products, and proprietary technologies in the PRC and overseas. These trademarks are important for brand identity, user recognition, and market differentiation. We continue to pursue additional registrations where commercially appropriate.

Under the user agreements of our consumer-facing AI-native products, including our MiniMax, Hailuo AI, MiniMax Audio, Talkie/Xingye and our Open Platform, the ownership and use of user-generated content are governed as follows:

• Ownership by Users: Users and/or original rights holders retain all ownership and intellectual-property rights in the content they input. The use of our services does not alter or transfer such rights. Between us and our users, all rights, title and interest in the generated content belong to the users.

• Limited Licence to the Group: To enhance user experience and improve our products, users grant us and/or our affiliates a limited licence to use such content strictly within the limits of applicable law. This licence may include rights of sub-licence or re-licence where necessary. It enables us to use the content solely for legitimate purposes, including: (i) product and service enhancement and optimisation; and (ii) brand promotion and lawful marketing activities.

– 287 –

BUSINESS

As a general matter, our legal advisors have advised that the ownership of user-generated content created by users using our AI models and products keeps evolving within the PRC, U.S., and Singapore. Therefore, determining ownership is fact-dependent, involving factors such as the level of human involvement and the terms outlined in user agreements. Given the ongoing evolution of applicable laws, we do not claim any independent IP ownership over such user-generated content beyond this limited licence, and our legal advisers have confirmed that the arrangement above complies with applicable PRC, U.S. and Singapore laws.

销售与营销

我们的销售与营销成果主要受我们基础模型的整体智能水平以及我们AI原生产品的强大价值主张所驱动。为此，我们建立了一个支持与全球用户和客户互动的销售与营销网络。这些销售与营销能力使我们能够在全球主要市场同步推出新的基础模型和产品。

定价策略

我们的定价策略主要以价值为基础，在很大程度上依赖于将我们模型的性能和价值与市场产品进行基准比较。我们建立了定价管理机制，据此，产品团队和营销策略团队密切协作，共同作出定价决策。营销策略团队对我们的产品进行全面的市场调研和基准分析，并根据实际性能和成本信息与产品团队讨论最优定价。

我们的定价流程综合考虑多种因素，以建立具有竞争力且对客户具有吸引力的产品方案。我们针对不同的产品类型、客户类别、应用场景和战略业务目标，采用多样化的定价策略。此外，我们密切监测和评估整体市场状况及竞争格局，以确定最具战略优势的市场定位。通过严格的内部评估，我们确保定价与我们的成本结构和盈利目标有效对齐。通过在相关细分市场对我们产品的性能进行基准比较，并有意将其定位为提供具有竞争力的性价比，我们的定价方式持续增强市场竞争力，并为客户提供具有吸引力的价值主张。下表列示了我们在不同变现方式下产品的价格区间和结构：

| 变现方式 | 产品 | 货币 | 价格区间 | 定价层级/套餐 | 主要定价因素 | |---|---|---|---|---|---| | 月度订阅计划 | MiniMax（MiniMax） | 美元（USD） | 19–69 | 月度订阅（2个层级及团队计划，每个席位15美元，共享积分） | 层级反映可用积分数量 |

– 288 –

BUSINESS Monetization Method

Product

Currency

Price Range

Pricing Tier/Package

Key Pricing Factors

Tiers reflect amount of credits and features available Tiers reflect usage caps and technical capability Based on feature set, user level (e.g., standard vs. premium and user's region) Based on feature set, user level (e.g. standard and premium) and subscription type Tiers reflect amount of credits and features available Tiers reflect amount of credits available Standalone top-up Standalone top-up Standalone top-up; price varies during promotional events Standalone top-up; different unit value from Talkie Tiers reflect amount of credits and RPM options available Differentiated input/output pricing based on input length

Hailuo AI

USD

9.99–199.99

Monthly Subscription (5 Tiers)

MiniMax Audio

USD

5–99

Talkie

USD

1.49–29.99

Monthly Subscription (4 Tiers) Monthly Subscription (4 Tiers)

Xingye

RMB

11–35

Monthly Subscription (4 Tiers)

Open Platform

USD

5–999

Monthly Audio Plan (5 Tiers)

USD

39

5,000 Credits

Hailuo AI MiniMax Audio Talkie

USD USD USD

5 5 1.99

500 Credits 100,000 Credits 180 Gems

Xingye

RMB

6

600 Gems

API Pack (Tiered) ў ў ў Open Platform

USD

1,000–6,000

Video API Pack (4 Tiers)

Token/Project-Based ў ў Open Platform

USD

Input price: 0.4–1.3 per 1,000,000 tokens

Token-based plan for API access to M-1

Output price: 2.2 per 1,000,000 tokens

Differentiated input/output pricing based on input length

Prepaid Credits ў ў ў ў MiniMax

USD

0.28–0.56 per clip

Clip-based plan for API access to Hailuo 02

Differentiated input/output pricing based on clip length and quality

USD

60–100 per 1,000,000 characters

Character length-based plan for API access to Speech 02

Differentiated input/output pricing based on whether 02-hd or 02-turbo is used

– 289 –

BUSINESS

注： 1.

我们所有产品均提供基于订阅制的会员计划，具有周期性自动扣费功能。所有订阅均可由用户随时自由取消。此外，海螺AI、MiniMax Audio及星野还提供非订阅制会员计划，该类计划不具有周期性自动扣费功能。

2.

所有产品的订阅均支持按月、按季度及按年订阅，但MiniMax Agent仅提供月度订阅。按季度及按年订阅计划相较于按月订阅计划，每月单价更低。

销售与营销策略

我们的客户群主要由两大细分市场构成：个人用户，以及企业与开发者客户。在服务个人用户方面，我们的销售与营销策略着重于品牌建设及技术创新的清晰传达。我们战略性地借助主要社交媒体平台，展示我们基础模型所蕴含的独特技术进步与核心创新。通过以具体演示和贴近生活的使用场景——例如我们的海螺-02所展示的高水准物理精准性——呈现基础模型能力的提升，我们得以确保高度的市场曝光度。此外，我们认为，以清晰且引人入胜的方式传达复杂的技术进步，对于在高度动态的全球基础模型行业中维持品牌影响力至关重要。

除了由基础模型智能水平提升及AI原生产品内在创新所带动的口碑推荐和用户复访外，我们还实施多项品牌推广与营销举措，以提升现有及潜在用户和广告主对我们品牌的认知度。我们主要聚焦于通过精准渠道持续提升品牌知名度并获取用户，包括在各移动应用商店（如苹果App Store及Google Play Store）投放广告，以及借助第三方服务提供商提供的推广服务。此外，我们还通过第三方服务提供商在各网站上投放针对性广告，以宣传我们的AI原生产品，确保我们产品的公众曝光度。有关我们与此类服务提供商签订的典型协议条款，请参阅"——供应链管理——与供应商协议的主要条款"部分。

此外，我们与全球AI社区内的关键意见领袖及具有影响力的社区人士开展战略合作。这些合作促成了具有参考价值且贴近实际的产品评测，并通过具有影响力的社交媒体平台（包括X、YouTube、小红书及微信）广泛传播。此类合作互动确保了不同受众群体——从专业人士到爱好者——均能切实体会到我们基础模型的实际优势与潜力。有关与上述关键意见领袖及具有影响力的社区人士所签订合同主要条款的摘要，请参阅"——我们的供应商"部分。

– 290 –

BUSINESS

我们的营销策略还包括有针对性的以社区为中心的营销活动和宣传推广。例如，我们在上海、纽约、旧金山、东京、伦敦和悉尼等主要国际科技中心举办线上和线下交流活动及研讨会。这些活动促进了我们技术团队与具有影响力的创作者之间的直接对话与交流合作，巩固了我们的社区关系，并提升了我们的市场定位。

在开拓企业客户方面，我们采用将品牌建设与行业特定优秀案例相结合的多元化策略。我们积极广泛传播详细的技术文档、成功案例以及行业特定应用实例，以提升品牌知名度和认可度。这种有针对性的方式充分展示了我们模型在各行业的相关性和经过验证的成功经验，有助于加强与企业客户的互动并提高转化率。

我们的销售策略将直接销售与通过全球主要渠道合作伙伴进行销售相结合。我们的销售和解决方案团队积极拓展并维护重点企业客户及渠道合作伙伴，为潜在客户提供直接、互动式的最新基础模型体验，从而促进客户有效入驻，并培养长久的客户满意度和忠诚度。

我们的销售及分销费用在2023年、2024年、2024年前九个月（截至2024年9月30日）及2025年前九个月（截至2025年9月30日）分别为2,280万美元、8,700万美元、5,340万美元及3,930万美元，分别占相应期间我们总收入的659.7%、285.0%、274.4%及73.6%。

2024年12月，我们Talkie应用程序的旧版本在某些司法管辖区从苹果应用商店（Apple's App Store）被临时下架，下架期间约为两个月，具体为2024年12月中旬至2025年2月中旬。尽管苹果公司未说明下架原因，且据我们所知，Talkie应用程序的临时下架并非由于任何产品缺陷或违规行为，但此次下架导致相关司法管辖区的用户无法下载Talkie应用程序旧版本，Talkie应用程序的日均下载量与下架前的平均水平相比减少了约1.68万次。然而，现有用户在此期间仍可正常使用其已下载的应用程序。

在此期间，我们对应用程序的设计和运营中的各项内容进行了有计划的修改和优化。对Talkie应用程序产品功能所做的上述调整旨在增强其风险管理能力并改善用户体验。这些优化工作的总体目标有两个方面：一是增强应用程序识别和防范潜在风险的能力，从而强化其保护机制；二是同步提升用户与应用程序整体交互的质量。上述调整旨在为所有用户打造更安全、更具吸引力的使用环境。

– 291 –

BUSINESS

自2025年2月中旬起，更新后的Talkie应用程序已可在受影响司法管辖区的苹果App Store上下载。我们认为，Talkie应用程序的临时下架对其业务并不构成重大影响，理由如下：(i) 在下架期间，其他司法管辖区用户下载Talkie应用程序，或通过Google Play Store等其他平台下载，均未受到影响；(ii) 即便在受影响的司法管辖区内，现有用户仍可继续使用其已下载的应用程序；及(iii) 更新后的Talkie应用程序自2025年2月中旬起已在苹果App Store恢复下载。

我们已实施一系列长期补救措施，旨在防止未来再次发生类似情况，并支持Talkie应用程序持续合规。首先，我们强化了内容管理及审核规程。相关工作包括审查并修订我们的内容管理指引，以提供更清晰的定义、更新的标准，以及处理不当内容（包括但不限于低俗内容及版权侵权）的既定程序。上述更新旨在使我们的指引更具可操作性，并在维护内容合法性方面更为有效。此外，我们加强了内容审核流程，现已包括用于初步检测的自动筛查，辅以针对复杂及升级案例的人工审核，以支持内容审核的准确性和一致性。其次，我们优化了用户举报系统，以助力维护积极健康的社区环境。应用程序内的用户举报流程已得到简化，为各类违规行为提供更清晰的分类选项，并为用户提供改进后的反馈机制。用户提交的举报现已在我们的审核队列中获得优先处理，旨在鼓励社区参与维护内容标准，并为用户营造相互尊重的环境。最后，我们已将App Store政策合规纳入我们的内部合规审查框架。该审查流程涉及我们的内部团队及外部顾问，他们对我们的产品功能、运营活动及内容策略进行评估。这一持续评估有助于确保Talkie应用程序始终符合相关法律、法规及平台要求。

自恢复上架以来，我们未再收到类似的下架通知。董事认为，所采取的风险管理措施，包括对Talkie应用程序功能所作的调整，足以有效防止未来再次发生类似情况。

合作伙伴生态系统

我们在人工智能开发管道的各环节与合作伙伴展开广泛协作。我们的上游生态系统包括主要云服务提供商及基础设施供应商。在下游方面，我们直接与企业、开发者及SaaS平台合作，通过API及SDK整合我们的基础模型。

我们深信开源协作的价值，视其为加速创新的强大机制，以及公开验证我们技术领先地位的重要途径。通过开放我们的技术，我们建立信任、促进透明度并广泛凝聚社区参与，从而显著提升我们的全球市场定位及开发者生态系统。

– 292 –

BUSINESS

我们积极与知名合作伙伴开展联合研发，共同推进先进人工智能解决方案的开发及共同技术标准的建立。我们还维持战略合作关系，以促进创新、互操作性及负责任的人工智能治理。

我们的客户

我们的客户涵盖多个行业及地区，范围广泛且多元化，包括中国、美国及新加坡。我们的客户遍及金融、医疗健康、智能设备及教育等主要行业。我们的销售与市场团队主要来自技术及云计算背景，能够有效推动客户参与与留存。我们的客户主要包括：(i) 企业客户，包括智能设备、文化旅游、医疗健康、金融及互联网服务等行业的企业，其通过API和SDK集成方式使用我们的人工智能模型及解决方案；(ii) 开发者，其通过访问我们的开放平台来创建和部署自己的人工智能产品；以及 (iii) 个人终端用户，主要通过我们的原生人工智能产品触达。客户获取策略多元化，涵盖直接销售推广、与云服务提供商的合作、开发者社区互动、促销活动及定向营销举措。

2022年，我们没有客户，亦未产生任何收入。在2023年、2024年及截至2025年9月30日止九个月，于各年度/期间内，我们五大客户的收入分别为210万美元、1,340万美元及1,160万美元，分别占各期间总收入的60.5%、44.1%及21.7%。在2023年、2024年及截至2025年9月30日止九个月，于各年度/期间内，我们最大客户的收入分别为130万美元、940万美元及780万美元，分别占各期间总收入的37.2%、30.9%及14.7%。

我们通常与主要客户签订框架协议，其主要条款如下：

• 期限与订阅：客户可按其订单订阅一项或多项服务，具体以实际提供的服务为准。在购买前，客户须查阅适用的服务条款，并根据自身需求判断是否适合；

• 付款条款：客户须及时支付所订购服务的费用，该等费用可能受可用性或促销限额的约束。由于持续的资源使用，服务激活后费用可能继续累计，与客户是否有进一步操作无关。优惠定价具有附条件性，若不符合相关标准则可能不适用，届时将按标准费率收取；

— 293 —

BUSINESS •

Duration: We typically do not assign a set duration to our framework agreement with our customers. Our framework agreement shall remain effective unless terminated by either party or by operation of law;

•

Privacy and Data Protection: Our customers are responsible for ensuring that all data submitted to or processed through our platform is lawfully collected and processed in full compliance with applicable data protection laws and regulations. They are required to obtain all necessary consents from data subjects and bear responsibility for any non-compliance. We are authorized to access and process customer data solely for the purpose of providing and improving our services;

•

Intellectual Properties: We retain all intellectual property rights in our products. Use of our services does not constitute any transfer or licence of our intellectual property or branding. Our customers are responsible for ensuring that all content they upload or process does not infringe any third-party rights and shall indemnify us for any resulting claims. Unauthorized use, reproduction, reverse engineering or disclosure of our technology or materials is strictly prohibited;

•

Confidentiality: Customers are required to keep our confidential information strictly confidential and use it only for purposes permitted under the agreement. Disclosure is only allowed where required by applicable laws or regulations. These confidentiality obligations shall survive the termination of the agreement;

•

Termination: We may suspend or terminate services with immediate effect if the customer breaches its obligations, becomes subject to sanctions or insolvency, or if continued performance would breach applicable laws. Upon termination, the customer must settle all outstanding fees and cease use of our services.

During the Track Record Period and up to the Latest Practicable Date, all of our five largest customers in each period during the Track Record Period were independent third parties. During the Track Record Period and as of the Latest Practicable Date, none of our Directors, their associates or any of our Shareholders (who or which to the knowledge of the Directors owned more than 5% of our issued share capital) had any interest in any of our five largest customers in each period during the Track Record Period.

– 294 –

BUSINESS

The following tables set forth details about our five largest customers in each period during the Track Record Period:

| Rank | Customers | Type of Products Purchased | Approximate Years of Business Relationship | Background | Credit Terms | Revenue (US$ in millions) | % of Our Total Revenue % | |------|-----------|---------------------------|-------------------------------------------|------------|--------------|--------------------------|--------------------------|

**For the year ended December 31, 2023**

| 1 | Customer A | Open Platform | since 2023 | A comprehensive cultural industry group headquartered in Shanghai, China, with digital reading as its foundation and IP cultivation and development at its core. | Net 15 days | 1.3 | 37.2 | |---|------------|---------------|------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------|-----|------| | 2 | Customer B | Open Platform | since 2023 | An office productivity software developer, which provides cross-platform solutions for word processing, spreadsheets, and presentations. | Net 45 days | 0.4 | 12.3 | | 3 | Customer C | Open Platform | since 2023 | A social e-commerce platform that integrates lifestyle sharing, product reviews, and online shopping. | Net 45 days | 0.2 | 6.2 | | 4 | Customer D | Open Platform | since 2023 | An online recruitment platform operator, providing comprehensive HR solutions and job-matching services. | Net 44 days | 0.1 | 2.7 | | 5 | Customer E | Open Platform | since 2023 | A digital content innovation company specializing in IP development and digital reading solutions, integrating technology with cultural creativity to deliver immersive reading experiences. | Net 20 days | 0.1 | 2.1 |

**For the year ended December 31, 2024**

| 1 | Customer F | Talkie (Supplier K) | since 2024 | A technology company based in Singapore, providing digital advertising, cloud service, and online service solutions. | Net 30 days | 9.4 | 30.9 | |---|------------|---------------------|------------|---------------------------------------------------------------------------------------------------------------------|-------------|-----|------| | 2 | Customer A | Open Platform | since 2023 | A comprehensive cultural industry group headquartered in Shanghai, China, with digital reading as its foundation and IP cultivation and development at its core. | Net 30 days | 1.1 | 3.7 |

– 295 –

BUSINESS

| Rank | Customers | Type of Products Purchased | Approximate Years of Business Relationship | Background | Credit Terms | Revenue (US$ in millions) | % of Our Total Revenue % | |------|-----------|--------------------------|-------------------------------------------|------------|--------------|--------------------------|--------------------------| | 3 | Customer B | Open Platform | since 2023 | An office productivity software developer, which provides cross-platform solutions for word processing, spreadsheets, and presentations. | Net 45 days | 1.0 | 3.4 | | 4 | Customer C | Open Platform | since 2023 | A social e-commerce platform that integrates lifestyle sharing, product reviews, and online shopping. | Net 45 days | 1.0 | 3.2 | | 5 | Customer G (Supplier I) | Talkie | since 2024 | An entity incorporated in Singapore, engaged in digital marketing and mobile app monetization services. | Net 30 days | 0.9 | 2.9 |

For the nine months ended September 30, 2025

| Rank | Customers | Type of Products Purchased | Approximate Years of Business Relationship | Background | Credit Terms | Revenue (US$ in millions) | % of Our Total Revenue % | |------|-----------|--------------------------|-------------------------------------------|------------|--------------|--------------------------|--------------------------| | 1 | Customer F (Supplier K) | Talkie | since 2024 | A technology company based in Singapore, providing digital advertising, cloud service, and online service solutions. | Net 30 days | 7.8 | 14.7 | | 2 | Customer H | Open Platform and other AI-based enterprise services | since 2025 | An AI-powered digital human and video synthesis platform that enables hyper-realistic avatar creation and multilingual video generation for global enterprises. | Net 45 days | 1.2 | 2.2 | | 3 | Customer I | Open Platform and other AI-based enterprise service | since 2024 | A developer centric platform that provides infrastructure and APIs for generative AI media (images, audio, video) and high speed model inference. | Net 14 days | 1.0 | 1.8 | | 4 | Customer J | Open Platform and other AI-based enterprise service | since 2024 | A Chinese AI and big data company specializing in intelligent customer service solutions and conversational AI platforms for enterprises. | Net 30 days | 0.9 | 1.6 | | 5 | Customer K | Talkie | since 2024 | A mobile technology company that provides end-to-end app distribution and monetization solutions for carriers, OEMs, and advertisers worldwide. | Net 60 days | 0.7 | 1.4 |

– 296 –

BUSINESS Our Suppliers We maintain stable and long-standing relationships with a select group of suppliers, principally in the areas of cloud infrastructure services. Our procurement strategy emphasizes supplier diversification, competitive bidding, and the establishment of stable, long-term contractual arrangements, thereby ensuring business continuity and the quality of our technical infrastructure. Our procurement team, consisting of specialists, is responsible for supplier identification, negotiation, and performance monitoring. Typical purchasing terms include clearly defined service level agreements (SLAs), penalty clauses for non-compliance, and standardized post-delivery payment schedules, generally ranging from 30 to 90 days following invoice issuance.

We operate under a light-asset business model, whereby critical computing infrastructure assets are owned and maintained by our suppliers, who provide computing services in accordance with our customized technical specifications. We conduct stringent quality inspections and acceptance checks on all goods and services received to ensure full compliance with our contractual requirements.

Our suppliers are primarily major technology service providers and cloud infrastructure vendors. We regularly review our supplier network to assess service reliability, cost competitiveness, and alignment with our evolving business needs.

在2022年、2023年、2024年及截至2025年9月30日止九个月内，我们于追踪记录期间每年╱期间内五大供应商的采购额分别为456万美元、4,980万美元、14,900万美元及14,360万美元，分别占同期采购总额的63.9%、63.0%、57.3%及62.5%。在2022年、2023年、2024年及截至2025年9月30日止九个月内，我们于追踪记录期间每年╱期间内最大供应商的采购额分别为160万美元、2,300万美元、7,280万美元及5,490万美元，分别占同期采购总额的22.8%、29.1%、28.0%及23.9%。

In 2022, 2023, 2024 and the nine months ended September 30, 2025, purchases from our five largest suppliers in each year/period during the Track Record Period amounted to US$4.5 million, US$49.8 million, US$149.0 million and US$143.6 million, representing 63.9%, 63.0%, 57.3% and 62.5% of our total purchases for the respective periods. In 2022, 2023, 2024 and the nine months ended September 30, 2025, purchases derived from our largest supplier in each year/period during the Track Record Period amounted to US$1.6 million, US$23.0 million, US$72.8 million and US$54.9 million, representing 22.8%, 29.1%, 28.0% and 23.9% of our total purchases for the respective periods.

As of the Latest Practicable Date, all of our five largest suppliers in each period during the Track Record Period were independent third parties, except that Supplier J comprises five subsidiaries of Alisoft China Holding Limited, which is a substantial shareholder of our company. During the Track Record Period and as of the Latest Practicable Date, none of our Directors, their associates or any of our Shareholders (who or which to the knowledge of the Directors owned more than 5% of our issued share capital) had any interest in any of our five largest suppliers in each period during the Track Record Period except Supplier J.

– 297 –

BUSINESS

The following tables set forth details about our five largest suppliers in each period during the Track Record Period:

| Rank | Suppliers | Type of Products/Services Provided | Background | Approximate Years of Business Relationship | Credit Terms | Purchase Amount (US$ in millions) | % of Our Total Purchase % | |------|-----------|-------------------------------------|------------|--------------------------------------------|--------------|-----------------------------------|--------------------------|

**For the year ended December 31, 2022**

| 1 | Supplier A | Cloud service | A cloud service provider registered in Beijing, China, primarily engaged in infrastructure and platform cloud services. | since 2022 | Net 90 days | 1.6 | 22.8 | | 2 | Supplier B | Cloud service | An enterprise cloud solutions provider based in Beijing, China, delivering cloud services. | since 2022 | Net 90 days | 1.3 | 19.1 | | 3 | Supplier C | Cloud service | A technology company incorporated in Beijing, China, specializing in cloud computing, artificial intelligence, and data analytics services. | since 2022 | Net 30 days | 1.2 | 17.1 | | 4 | Supplier D | Outsourcing service | An IT services company headquartered in Shenzhen, China, providing software development, outsourced research and development, and system integration services. | since 2022 | Net 30 days | 0.2 | 2.6 | | 5 | Supplier E | Technical service – data processing | A regional technology company registered in Shanxi Province, China, engaged in local IT services and solutions. | since 2022 | Net 30 days | 0.2 | 2.3 |

**For the year ended December 31, 2023**

| 1 | Supplier C | Cloud service | A technology company incorporated in Beijing, China, specializing in cloud computing, artificial intelligence, and data analytics services. | since 2022 | Net 30 days | 23.0 | 29.1 | | 2 | Supplier A | Cloud service | A cloud service provider registered in Beijing, China, primarily engaged in infrastructure and platform cloud services. | since 2022 | Net 90 days | 12.2 | 15.5 |

– 298 –

BUSINESS

| Rank | Suppliers | Type of Products/Services Provided | Background | Approximate Years of Business Relationship | Credit Terms | Purchase Amount (US$ in millions) | % of Our Total Purchase | |------|-----------|-----------------------------------|------------|-------------------------------------------|--------------|----------------------------------|------------------------| | 3 | Supplier F | Marketing service | A digital marketing company incorporated in Singapore, specializing in advertising, user acquisition, and brand promotion services. | since 2023 | Net 30 days | 6.9 | 8.7% | | 4 | Supplier G | Cloud service | A Beijing-based service provider specializing in the cloud services. | since 2023 | Net 90 days | 4.2 | 5.2% | | 5 | Supplier B | Cloud service | An enterprise cloud solutions provider based in Beijing, China, delivering cloud services. | since 2022 | Net 90 days | 3.5 | 4.5% |

**For the year ended December 31, 2024**

| Rank | Suppliers | Type of Products/Services Provided | Background | Approximate Years of Business Relationship | Credit Terms | Purchase Amount (US$ in millions) | % of Our Total Purchase | |------|-----------|-----------------------------------|------------|-------------------------------------------|--------------|----------------------------------|------------------------| | 1 | Supplier H | Cloud service | A technology service provider based in Zhejiang, China, offering cloud services. | since 2023 | Net 30 days | 72.8 | 28.0% | | 2 | Supplier A | Cloud service | A cloud service provider registered in Beijing, China, primarily engaged in infrastructure and platform cloud services. | since 2022 | Net 90 days | 29.7 | 11.4% | | 3 | Supplier G | Cloud service | A Beijing-based service provider specializing in the cloud services. | since 2023 | Net 90 days | 19.1 | 7.3% | | 4 | Supplier I (Customer G) | Marketing service | An entity incorporated in Singapore, engaged in digital marketing and mobile app monetization services. | since 2023 | Net 30 days | 15.3 | 5.9% | | 5 | Supplier B | Cloud service | An enterprise cloud solutions provider based in Beijing, China, delivering cloud services. | since 2022 | Net 90 days | 12.1 | 4.7% |

**For the nine months ended September 30, 2025**

| Rank | Suppliers | Type of Products/Services Provided | Background | Approximate Years of Business Relationship | Credit Terms | Purchase Amount (US$ in millions) | % of Our Total Purchase | |------|-----------|-----------------------------------|------------|-------------------------------------------|--------------|----------------------------------|------------------------| | 1 | Supplier J | Cloud service | An international cloud service provider with subsidiaries both in China and Singapore, providing cloud service to global enterprises and developers. | since 2021 | Net 90 days | 54.9 | 23.9% | | 2 | Supplier H | Cloud service | A technology service provider based in Zhejiang, China, offering cloud services. | since 2023 | Net 30 days | 53.7 | 23.3% | | 3 | Supplier A | Cloud service | A cloud service provider registered in Beijing, China, primarily engaged in infrastructure and platform cloud services. | since 2022 | Net 90 days | 13.0 | 5.7% |

– 299 –

BUSINESS

| Rank | Suppliers | Type of Products/Services Provided | Background | Approximate Years of Business Relationship | Credit Terms | Purchase Amount (US$ in millions) | % of Our Total Purchase % | |------|-----------|--------------------------------------|------------|---------------------------------------------|--------------|-----------------------------------|---------------------------| | 4 | Supplier K (Customer F) | Cloud service | A technology company based in Singapore, providing digital advertising, cloud service, and online service solutions. | since 2023 | Net 45 days | 11.4 | 5.0 | | 5 | Supplier C | Cloud service | A technology company incorporated in Beijing, China, specializing in cloud computing, artificial intelligence, and data analytics services. | since 2022 | Net 30 days | 10.6 | 4.6 |

**Overlapping of Suppliers and Customers**

During the Track Record Period and up to the Latest Practicable Date, certain parties acted as both our suppliers and our customers. In particular, Customer G/Supplier I was among our top five customers and suppliers in 2024, while Customer F/Supplier K was among our top five customers in 2024 and among our top five customers and suppliers in 2025.

These overlapping relationships arose primarily from our business interactions with global online marketing technology platforms. We provide online marketing services to Customer F/Supplier K and Customer G/Supplier I. We generate online marketing services revenue under the cost-per-mille pricing model, with reconciliation conducted monthly and payments made in the following month upon reaching the agreed payment threshold. In their capacity as suppliers, we purchase online marketing services and cloud services from Customer F/Supplier K, and we purchase online marketing services from Customer G/Supplier I. These purchases are also reconciled and settled on a monthly basis pursuant to agreed campaign terms.

Such transactions reflect the nature of our operations and the dual roles these global platforms commonly play in the digital online marketing ecosystem. From our perspective, the decision to place advertisements through these platforms is driven by their extensive user reach and solid targeting capabilities, which align with our marketing objectives. At the same time, allowing these platforms to display ads on our properties enables us to monetize our user traffic effectively.

All such transactions were conducted on an arm's length basis and at prevailing market rates. While the revenue and purchase amounts attributable to Customer F/Supplier K and Customer G/Supplier I during the Track Record Period were significant in the context of our

– 300 –

BUSINESS

前五大客户及供应商中，上述重叠关系既未导致我们对任何单一方在关键商品或服务供应方面产生依赖，亦未损害我们的经营独立性。

于2023年、2024年及截至2025年9月30日止九个月，客户F/供应商K作为客户产生的收入分别为零、940万美元及780万美元；而客户F/供应商K作为供应商的采购金额亦于截至2025年9月30日止九个月跻身我们五大供应商之列，我们向客户F/供应商K的采购金额于2022年、2023年、2024年及截至2025年9月30日止九个月分别为零、160万美元、1,140万美元及1,140万美元。同样，客户G/供应商I作为客户于2023年、2024年及截至2025年9月30日止九个月分别为我们贡献收入零、90万美元及10万美元，且亦于2024年跻身我们五大供应商之列，我们向客户G/供应商I的采购金额于2022年、2023年、2024年及截至2025年9月30日止九个月分别为零、150万美元、1,530万美元及400万美元。根据灼识咨询（CIC）的资料，我们于追踪记录期间向重叠客户及供应商的销售及采购金额合理，且符合行业惯例。

供应链管理

我们的供应链管理策略着重于通过灵活的合同安排确保充足可靠的云基础设施资源。我们以最优条款和定价利用计算基础设施及资源进行分布式训练，而不直接拥有硬件资产或订立大量长期合同安排。在我们针对硬件及计算基础设施的轻资产模式下，我们向硬件及计算基础设施服务提供商支付的费用包括：(i) 硬件租赁费及 (ii) 整体维护、维修及数据保管服务费。此轻资产模式为我们提供了灵活性、可扩展性及具有成本效益的资源利用方式。我们的采购团队定期评估供应商绩效及市场状况，并进行严格的质量检查，以优化效率并将中断风险降至最低。于追踪记录期间，我们采购了多种服务，包括：(i) 用于网络服务、数据存储及数据库运营的云基础设施及计算资源服务，通常按月计费或按数据使用量计费；(ii) 数据及内容审核技术服务，按内容审核数量计费；(iii) 外包数据标注服务，涉及原始数据的分类、标注及整理，按处理数据量计费；以及 (iv) 通过应用程序商店、网络平台及关键意见领袖（KOL）等渠道提供的营销服务。

我们向若干美国服务提供商采购服务，包括云服务及营销服务。上述两类服务均可轻易由非美国供应商替代。云服务已实现标准化，全球范围内有众多替代方案可供选择。同样，营销服务亦已成熟发展，众多非美国供应商提供具有竞争力的解决方案，能够满足类似需求，确保从美国供应商转换不会造成重大业务中断。

– 301 –

BUSINESS

与供应商协议的主要条款

我们通常与主要供应商就云基础设施及计算服务签订框架协议，其主要条款如下：

• 供应商义务：供应商须依法提供云服务及技术支持，并保持所需资质。供应商须通过多种渠道提供及时支持，并协助安装与维护。供应商保证其服务符合法律及行业标准，确保安全性与稳定性，及时处理安全事件，立即通知我们相关事件的发生，并就由此造成的任何损失对我们进行赔偿；

• 付款条款：服务费用自验收之日起计算，按实际使用量按月计费。不足一个月的费用按比例计算。款项根据最终验收及对账单结算；

• 检验与验收：供应商须按约定以单批或多批方式交付服务。相关硬件安装完毕、通电并可正常访问服务后，即视为交付完成。我们将在交付后30天内，依据服务清单所列标准进行验收。对于未通过验收的服务，供应商须在三天内予以更换，直至获得批准。验收方式及标准按项目分别确定；

• 最低采购承诺：对于特定供应商，我们的云基础设施及计算服务框架协议可能视具体情况设定最低采购承诺。此类协议可能规定在特定期限内的最低采购量，若未达到最低采购量，可能须支付差额费用。上述条款通常根据预期运营需求协商确定，符合市场惯例；

• 保密：供应商须妥善保管我们的保密信息，仅将其用于协议规定的目的，未经许可不得披露。供应商须将内部访问权限制于经授权人员，并采取适当措施保护相关数据。协议终止后，所有资料须予以归还或销毁。供应商须对任何违约行为或由此造成的损失承担责任。

我们通常与主要供应商就在线营销服务签订框架协议，此类协议的主要条款概述如下：

• 期限：一年。

– 302 –

BUSINESS •

服务范围与营销模式：供应商在指定数字平台（如移动应用商店、社交媒体平台及网站）上投放与我们产品相关的推广内容，以促进用户互动并提升产品曝光度。营销活动可采用多种计价模式开展，包括按行动付费（CPA）、按时间付费（CPT）、按下载付费（CPD）、按点击付费（CPC）、按千次展示付费（CPM）及按观看付费（CPV）等。供应商还负责广告系列设置、策略优化、数据监控及报告工作。

•

合规与知识产权保护：供应商须遵守适用法律，确保推广内容的合法性与适当性，并不得未经授权擅自修改或使用我们的产品、商标或内容。任何代表我们创作的推广素材均归我们所有，供应商须应要求转让所有相关材料及知识产权。我们确保我们的产品及推广内容符合相关司法管辖区的适用法律。

•

保密与反贿赂：供应商须对所有专有信息保密，并严格禁止在协议相关事宜中提供或收受任何不当利益。违反保密义务或商业诚信的行为可能导致协议立即终止并承担损害赔偿责任。

•

付款条款：我们就供应商提供的服务及按CPA、CPT、CPD、CPC、CPM或CPV计价模式定价的推广内容向其付款。各计价模式下的付款结构说明如下： 䡩

在CPA模式下，我们根据有效产品激活数量向供应商付款。有效产品激活是指用户通过供应商提供的推广内容成功下载、安装、在线访问或注册我们产品账户。

䡩

在CPT模式下，我们根据我们产品的推广内容在指定平台或位置的展示时长向供应商付款。

䡩

在CPD模式下，我们根据通过推广内容产生的产品下载数量向供应商付款。

䡩

在CPC模式下，我们根据推广内容的点击次数向供应商付款。

䡩

在CPM模式下，我们根据推广内容的曝光次数向供应商付款。

– 303 –

BUSINESS

•

Under the CPV model, our suppliers are paid based on the number of playbacks of the promotional video.

•

Termination and Remedies: We may terminate the agreement at any time. In cases of data fraud, material non-compliance, or unauthorized subcontracting, suppliers are subject to penalties and may be disqualified from further business with us.

We typically enter into framework agreements with multi-channel networking ("MCN") companies, and the salient terms of which are set forth below:

•

Promotion Support Services: Suppliers provide promotion support services tailored to our needs and promotion model, including but not limited to KOL selection and placement, account setup and funding, campaign planning, data monitoring and reporting, and content production;

•

Duration: The duration for MCN service agreements is typically one year and may be renewed by mutual agreement;

•

Payment Terms: Service fees are determined in accordance with the pricing terms and no additional service charges, handling fees or other payments are payable;

•

Settlement Terms: Service fees are settled in accordance with the settlement cycle specified in the agreement. After each cycle, the parties will reconcile actual service data, including service volume, fees, virtual account spending and target completion, based on our data or data approved by us. Only published and valid promotional content supported by proof, such as live links, is eligible for settlement;

•

Confidentiality: Suppliers must use all confidential information solely for the purpose of providing services and limit disclosure to employees on a need-to-know basis. Upon termination, all confidential materials must be returned or destroyed. Breach of confidentiality may result in liability for damages, and these obligations shall survive the termination of the agreement;

•

Compliance and Service Standards: Suppliers must ensure that promotional services are lawful, appropriate, and comply with the agreed scope. All KOL content must be pre-approved by us, remain accessible for at least one month post-publication, and must not be modified or removed without consent. Suppliers are fully liable for any legal violations, IP infringements, or failure to meet performance standards, and may be required to compensate us for losses;

•

Duration and Termination: The agreement may be amended or terminated by mutual consent. We may terminate it at any time for breach or with three days' notice, with unused prepaid amounts refunded within ten days.

– 304 –

BUSINESS

此外，我们通常通过KOL直接合作的方式与主要海外供应商签订框架协议，其主要条款如下：

• 社交媒体帖子：供应商须按照约定的发布时间表及法律披露要求，发布原创且真实的社交媒体内容，并确保遵守平台政策及适用法律；

• 内容时间表及审批：供应商须按照约定时间表提前提交帖子以供我们审批，我们保留监督及要求更正的权利，并在供应商不合规的情况下扣留费用或终止协议；

• 保密：供应商须对与我们及营销活动相关的所有非公开信息严格保密，仅将其用于参与营销活动，且此义务在协议终止后仍须持续履行；

• 赔偿：供应商同意就其违反本协议项下任何陈述、保证或其他义务所引起或与之相关的任何损失、责任、索赔、损害赔偿或费用（包括律师费），对我们、我们的关联公司及各自的高级管理人员、董事、雇员和代理人进行赔偿、辩护并使其免受损害；

• 付款条款：作为完成并发布所有帖子的对价，我们须向供应商支付一次性费用，并于达成约定里程碑或在指定日期分期支付；

• 终止条款：若发生重大违约且在通知后12小时内未予补救，或在发生重复违约或可能造成声誉损害的行为时，我们可立即终止协议。

商业化及业务可持续性

我们专家技术产品的商业化

自成立以来，我们的商业战略聚焦于两个核心领域：开发先进的基础模型，以及创造能够提升生产力和生活质量的AI原生产品。我们所有的基础模型和AI原生产品均被设计为《上市规则》第18C章所定义的专家技术产品。根据中国国际金融股份有限公司（CIC）的意见，我们确认所有专家技术产品均属于《上市规则》项下人工智能可接受行业范畴，且记录期间产生的所有收入均来源于上述产品的销售。我们进一步确认，所有基础模型和AI原生产品均为自主研发。有关我们主要知识产权归属的描述，请参阅"——知识产权"。以下各节就我们所有产品如何符合《上市规则》第18C章所定义的专家技术行业可接受行业范畴作出概述：

– 305 –

BUSINESS

基础模型：我们已构建了多个跨模态的综合性基础模型。我们的基础模型套件包括大语言模型、视频生成模型，以及语音和音乐生成模型。

| 专项技术产品 | 专项技术行业认可板块 | 主要功能分析 | 商业化时间表 | |---|---|---|---| | MiniMax-M1 | 人工智能（AI赋能算法编程） | MiniMax-M1是一款开放权重的大规模混合注意力推理模型。MiniMax-M1采用混合MoE架构与闪电注意力机制（一种效率更高的线性注意力变体）相结合的方式驱动，使我们的模型在长上下文处理方面表现出色——进一步提升了其效率和可扩展性，并有助于开发更强大的AI智能体。MiniMax-M1模型在发布时拥有全球最长的上下文窗口，支持最长达100万个tokens的上下文长度。上述特性使MiniMax-M1尤其适合需要处理长输入和进行深度推理的复杂任务。 | 已在我们的开放平台（Open Platform）商业部署。 | | MiniMax-M2 | 人工智能（AI赋能算法编程） | MiniMax-M2是我们最新的大语言模型，专为在编程和智能体任务中实现卓越性能而设计。凭借精心设计的数据高效MoE架构和激活参数设计，与MiniMax-M1相比，MiniMax-M2在保持模型智能、响应能力和成本效益优化的同时，以大幅更快的推理速度提供更高性能的能力。 | 已在我们的智能体产品MiniMax及开放平台（Open Platform）商业部署。 | | Hailuo-02 | 人工智能（AI赋能算法编程） | Hailuo-02系列模型可从多种形式的信息输入中生成高质量视频内容。Hailuo-02已实现大规模商业化，并在发布时于全球基准测试中取得具有竞争力的成绩，提供电影级视频质量、先进的提示词遵循能力、流畅的动态效果和多样化的风格。凭借用户友好的界面及美学优化能力，它帮助内容创作者和广告主通过简单的提示词制作出引人入胜的视频。 | 已在我们的旗舰视频生成平台海螺AI（Hailuo AI）及开放平台（Open Platform）商业部署。 |

— 306 —

BUSINESS

Specialist Technology Industry Acceptable Section

| Specialist Technology Products | | Speech-02 | |---|---|---| | | | Artificial intelligence (AI-empowered algorithm programming) |

| Timeline of Commercialization | Main Function Analysis | |---|---| | Commercially deployed in MiniMax Audio, our audio generation tool, and our Open Platform. | The Speech-02 model series is designed to generate natural, high-quality speech from text input. It is the recognized large language speech model globally upon its release, and along with its predecessors, delivers hyperrealistic, personalized voice synthesis across multiple languages. |

AI-native Products: Leveraging our multi-modal foundation model suite, we deliver AI-native products and services that use the power of AI to benefit both individual users and enterprises around the world. The evolution of our AI-native products is rooted in advancements in its foundation models. Through continuous upgrades to its existing foundation models and the development of new ones, we are able to design and create AI-native products with enhanced user experience.

Specialist Technology Industry Acceptable Section

| Specialist Technology Products | | MiniMax | | Artificial intelligence (AI solutions) | |---|---|---|---|---| | | | Hailuo AI | | Artificial intelligence (AI solutions) |

| | Main Function Analysis | Major Customer Type and Timeline of Commercialization (Commencement of Revenue Generation) | Monetization Model | |---|---|---|---| | MiniMax | MiniMax is our intelligent AI agent application powered by MiniMax Agent, which is designed to autonomously perform a wide range of tasks through natural language instructions. Supported by our foundation models, MiniMax Agent can plan, reason, and execute complex actions such as coding, research, document drafting, and presentation creation within a unified workspace. | Individual users. Commencement of revenue generation in June 2025. | Freemium, subscriptions, token-based in-app purchases. | | Hailuo AI | Hailuo AI fully integrates our Hailuo-02 model that has quickly become one of the world's most popular AI image and video creation platforms through organic user adoption. It is offered in both web and app forms, and is designed for real-time, high-quality image and video generation. | Individual users. Commencement of revenue generation in October 2024. | Freemium, subscriptions, token-based in-app purchases. |

– 307 –

BUSINESS

Specialist Technology Products

Specialist Technology Industry Acceptable Section

MiniMax Audio ✓ ✓ ✓

Artificial intelligence (AI solutions)

Talkie/Xingye ✓ ✓ ✓

Artificial intelligence (AI solutions)

Main Function Analysis

Major Customer Type and Timeline of Commercialization (Commencement of Revenue Generation)

MiniMax Audio is designed to provide users with high-fidelity audio generation capabilities. Accessible via web platform, MiniMax Audio integrates the Company's Speech-02 model to support interactive audio synthesis and generate natural, high-quality speech from text input.

Individual users.

Talkie (for international markets)/Xingye (for Chinese domestic market) is a globally recognized AI-native multi-modal entertainment platform.

Individual users. Commencement of revenue generation: Talkie – June 2023.

Commencement of revenue generation in February 2025.

Monetization Model

Freemium, subscriptions, token-based in-app purchases.

Freemium, subscriptions, online marketing service, in-app purchases.

Xingye – September 2023. Open Platform ✓ ✓ ✓

Artificial intelligence (AI solutions)

Our Open Platform offers scalable, configurable AI services to enterprise customers across more than 100 countries and regions as of September 30, 2025. Through public APIs, enterprise and developer customers can access the Company's foundation models and integrate such text, video and audio model capabilities into their own products and services. Our Open Platform supports rapid business deployment in key industry sectors such as smart devices, healthcare, cultural tourism, finance, and internet services — making it one of the world's largest open platforms for enterprises in terms of average daily token volume, signifying widespread adoption.

Our Open Platform supports business customers in key industry sectors such as smart devices, healthcare, cultural tourism, finance, and Internet services.

Freemium, token-based API billing, enterprise license.

Commencement of revenue generation in May 2023.

Early-Stage Commercialization and Business Sustainability

We are still at a nascent stage in terms of monetization and commercialization as historically we have been largely focused on developing our foundation AI models. During the Track Record Period, we have scaled our product offerings and therefore have experienced rapid revenue growth, reflecting our ability to advance proprietary foundation models while – 308 –

BUSINESS

随着我们在个人用户、开发者及企业客户群体中快速扩大人工智能原生产品的使用规模，我们的收入从2023年的350万美元增长至2024年的3,050万美元，这一增长势头来自面向开发者和企业的开放平台（Open Platform）以及多模态人工智能原生消费者产品的共同驱动。截至2025年9月30日止九个月，我们的收入进一步增长至5,340万美元，而2024年同期收入为1,950万美元。上述增长得益于基础模型智能水平的提升、人工智能原生产品矩阵的扩展、个人用户、开发者及企业客户采用率的提高，以及涵盖订阅、应用内充值、企业API使用及在线营销服务的多元化变现渠道。

随着运营规模的扩大，我们的毛利率得到显著改善，从2023年的负24.7%提升至2024年的12.2%，并进一步提升至截至2025年9月30日止九个月的23.3%。上述改善主要得益于模型智能水平的提升、模型与系统效率的优化、基础设施资源配置的改善，以及收入规模相对于算力强度的持续提升，与我们提升人工智能基础设施效率的战略方向一致。尤其是，在业绩记录期内，人工智能原生产品的毛利率因用户参与度及变现能力的提升以及新变现功能的引入而大幅改善，体现了我们持续强化核心人工智能原生产品变现能力的商业战略重心。

由于对基础模型研发及人工智能基础设施进行了大量前期投入，我们分别于2022年、2023年、2024年以及截至2024年9月30日止九个月和截至2025年9月30日止九个月录得年度/期间亏损7,370万美元、26,920万美元、46,520万美元、30,430万美元及51,200万美元。剔除以股份为基础的薪酬开支、金融工具公允价值变动及上市开支后，我们的经调整净亏损（非国际财务报告准则计量）占收入的百分比大幅收窄，从2023年的逾2,500%下降至2024年的800.2%，并进一步降至截至2025年9月30日止九个月的348.6%。我们的研发开支占收入的百分比从2023年的逾2,000%下降至2024年的619.1%，并进一步降至截至2025年9月30日止九个月的337.4%。由于我们在可预见的未来仍将主要定位为在人工智能研究领域运营的研发导向型公司，预计净亏损状况将持续。

我们的经调整净亏损（非国际财务报告准则计量）主要源于业绩记录期内产生的大额研发开支。随着业务的快速增长，销售及分销开支、行政开支及研发开支的绝对金额在整个业绩记录期内均大幅增加。历史上，在持续开发基础模型和人工智能原生产品并扩大品牌影响力的过程中，我们在研发活动上进行了战略性投入。然而，随着业务规模和范围的扩大，我们预计将持续提升运营效率。

— 309 —

业务

在业绩记录期间，我们主要通过股东的资本注入及融资活动（包括发行可转换可赎回优先股及可转换债券）来满足现金需求。请参阅"历史、重组及公司架构——上市前投资"。截至2025年9月30日，我们的现金及现金等价物录得3.626亿美元。此外，我们持有定期存款及以理财产品形式存在的金融资产，其流动部分分别于2022年12月31日、2023年12月31日、2024年12月31日及2025年9月30日达到6,580万美元、1.075亿美元、4.690亿美元及6.442亿美元。我们的现金总余额足以覆盖经营活动所用净现金流量，并为我们的业务扩张提供充足的流动性。因此，我们认为在考虑可用财务资源后，我们拥有充足的营运资金，包括充足的现金及流动资产。我们预计截至2025年12月31日止年度的净亏损将大幅增加，主要原因在于：随着我们持续提升基础模型的智能水平，预期研发费用将有所增加；同时，由于预期2025年公司估值将有所上升，金融负债公允价值损失亦将随之增加。未来，我们旨在通过以下重点方向维持业务可持续性并实现长期商业化：（i）把握基础模型行业的快速增长机遇；（ii）持续提升基础模型的智能水平；（iii）增强我们人工智能技术的普惠性；（iv）拓宽我们原生AI产品矩阵的变现渠道；以及（v）优化组织效率与可扩展性。

由于在基础模型研发及AI基础设施方面的前期大规模投入，我们于2022年、2023年、2024年及截至2024年9月30日止九个月以及截至2025年9月30日止九个月分别录得亏损7,370万美元、2.692亿美元、4.652亿美元、3.043亿美元及5.120亿美元。我们的累计亏损、经调整净亏损（非国际财务报告准则计量）及经营活动净现金流出，主要源于业绩记录期间产生的大量研发费用，彼时我们的模型商业化尚处于起步阶段。随着业务的快速增长，我们的销售及分销费用、行政费用及研发费用的绝对金额在整个业绩记录期间均大幅增加。我们在业绩记录期间对基础模型研发进行了大量投资，并已在全球范围内推出排名领先的文本、视频及音频模型，这些模型直至近期方才开始产生收入，即：开放平台自2023年5月起、Talkie自2023年6月起、Hailuo AI自2024年10月起。剔除以股份为基础的付款费用、金融工具公允价值变动及上市费用后，我们的经调整净亏损（非国际财务报告准则计量）占收入的百分比显著收窄，由2023年的逾2,500%降至2024年的800.2%，并进一步降至截至2025年9月30日止九个月的348.6%。我们的研发费用占收入的百分比由2023年的逾2,000%降至2024年的619.1%，并进一步降至截至2025年9月30日止九个月的337.4%。我们预计在可预见的未来（包括截至2025年12月31日止十二个月）将持续录得净亏损，同时我们仍将主要作为一家专注于研发、在AI研究领域运营的公司。在业绩记录期之前，我们亦于2021年录得累计亏损，主要原因在于我们的天使轮融资及估值上升，导致在对优先股进行重新计量时产生金融负债公允价值损失。

— 310 —

BUSINESS Path to the Commercialization of our Specialist Technology Products Leveraging the Rapid Growth of the Foundation Model Industry

Since our inception, we have capitalized on the tailwinds of the rapidly expanding foundation model industry, which is undergoing unprecedented growth and reshaping human society at a remarkable pace. According to CIC, as these technologies mature and end-users' willingness to pay continues to rise, the global foundation model market in terms of model-based revenue is expected to grow rapidly — from US$10.7 billion in 2024 to US$206.5 billion by 2029, representing a CAGR of 80.7%.

We are equipped to leverage this market trend with the following core competitive edges: (i) we are built on a technological vision and a commitment to scalability, focusing on expanding multi-modal capabilities and pursuing model algorithm innovation, including our adoption of the MoE architecture and the implementation of Linear Attention mechanisms; (ii) we have developed proprietary AI infrastructure, including AI training and inference framework, unified training and inference computing resources, and multi-cluster load balancing to enable scalable development and inference of foundation models; (iii) we adopt a scalable commercialization approach with global adoption, serving both individual users as well as enterprise and developer customers across the globe, providing a customer foundation for our future suite of AI-native offerings; (iv) we operate under a flat and nimble organizational structure.

In particular, we are well positioned to capture the growth potential of the video generation and editing market as well as the AI agent market.

• Video generation and editing market: As video generation continues to converge with advanced text understanding, we expect to unlock a broader set of application scenarios, including automated content production and professional-grade agentic video editing. By combining our track record in developing leading video generation and text models, we are well positioned to capture this emerging market opportunity and address increasingly sophisticated creator and enterprise use cases. We plan to further commercialize our Hailuo video generation model series by introducing subscription-based and enterprise-grade solutions that integrate video generation into marketing, media and design workflows, thereby expanding monetization channels and improving operating efficiency.

• AI agent market: According to CIC, the application scenarios of AI agents are expected to experience significant growth driven by the continuous advancement of model intelligence. Gartner, a global research and advisory firm, estimates that by 2027, 50% of business decisions will be strengthened or automated by AI agents. We are among the first Asia-based companies to enter the Asia agent markets previously dominated by international players. Our latest MiniMax-M2 model has already demonstrated autonomous decision-making and execution capabilities applicable to workflow automation, coding and other productivity scenarios. Upon its release, — 311 —

BUSINESS

MiniMax-M2在全球所有开源模型中，在Artificial Analysis人工智能指数中排名第一，反映出其认知与推理能力，进一步强化了我们在智能体应用领域的技术领先地位。

展望未来，随着基础模型技术在各行各业和垂直领域得到更广泛的应用，我们相信自身的核心竞争优势将使我们能够识别并把握新兴市场机遇，从而持续推动收入增长。

**持续提升基础模型智能水平**

我们认为，模型的智能水平与能力强弱直接影响市场对我们AI原生产品套件的需求，进而影响我们的收入增长。自成立之初，我们便致力于开发下一代基础模型，以满足文本、视频、语音等多模态格式不断演变的需求。2025年10月，我们发布并开源了MiniMax-M2——这是我们专为智能体及代码相关应用而设计的最新推理模型。MiniMax-M2以大幅降低的成本和更快的推理速度实现了高性能能力。依托经过优化的激活参数设计，MiniMax-M2在智能水平、速度与成本效益之间实现了更佳平衡，为专业用户和消费用户提供响应迅速、经济高效的AI基础模型。具体而言，MiniMax-M2采用先进架构，激活参数约为100亿（总参数量为2,300亿），其API定价约为每百万输入token 0.30美元、每百万输出token 1.20美元，约为领先境外模型成本的8%。将上一代模型发布前最后一个月（6月）与MiniMax-M2发布后第一个完整月（11月）相比，我们在各模态的使用量均实现了显著提升：文本模型的月token消耗量增长了770.7%，视频模型的月视频生成数量增长了652.9%，音频模型的月处理字符数增长了70.4%。上述增长反映了我们现一代模型在文本、视频和音频使用场景中能力的提升以及更广泛的应用普及。

展望未来，我们致力于通过对下一代多模态系统的持续投入，进一步巩固我们在基础模型开发领域的领先地位。我们计划陆续推出能力持续增强的核心模型迭代版本——例如改进的记忆功能与上下文理解能力、更高分辨率的多模态融合，以及更快的实时推理速度——以满足企业和个人用户的需求，推动产品的更广泛应用与用户参与度提升。借助我们优化的训练系统，我们旨在进一步缩短模型开发周期，并显著提升算力资源利用率。具体而言，我们计划在未来五年内，每年持续在全球范围内招募30名顶尖基础模型及AI基础设施研究人员、工程师和科学家，同时投资于内部人才的成长与发展。我们还将投资于下一代、更高投资回报率的算力及相关AI基础设施，包括与云合作伙伴共同推进升级，以支持更大规模、更快速的多模态模型，同时持续降低训练和推理成本。

– 312 –

BUSINESS

提升我们人工智能技术的可负担性

自成立以来，我们认识到，我们人工智能技术的可及性与可负担性（主要由成本驱动）将是我们长期收入增长的关键决定因素。我们长期以来坚信，人工智能模型训练和推理的高昂成本是人工智能技术广泛普及的重大障碍，包括我们的人工智能模型套件及人工智能原生产品在全球范围内的推广应用。

通过在模型架构和基础设施方面的持续创新，我们在2023年至2024年期间降低了与推理相关的成本。我们效率的提升源于通过动态资源分配和统一的训练-推理框架保持较高的算力利用率，并辅以支持实时工作负载弹性扩展的高并发调度系统。我们的销售成本占收入的百分比从2023年的124.7%下降至2024年的87.8%，随后从截至2024年9月30日止九个月的97.4%进一步下降至2025年同期的76.7%。这一显著降幅反映了推理效率的提升以及基础设施利用率提高所带来的规模经济效益。我们预计将继续降低每个token的边际成本，从而进一步提升我们人工智能原生产品的可负担性。因此，人工智能原生产品的毛利率从2023年的负380.2%改善至2024年的负8.1%，并从截至2024年9月30日止九个月的负23.5%改善至2025年同期的4.7%。我们致力于通过提升用户变现能力和推理成本效率，进一步改善人工智能原生产品的毛利率。开放平台及其他基于人工智能的企业服务毛利率从截至2024年9月30日止九个月的62.3%提升至2025年同期的69.4%。我们致力于在未来持续维持并优化开放平台及其他基于人工智能的企业服务的毛利率水平。

高效的模型训练对我们的长期成功同样至关重要。与模型训练、微调及实验相关的成本被计入研究与开发（研发）费用。为提升训练效率，我们组建了内部基础设施团队，并独立开发了一套针对大规模计算集群量身定制的高性能训练框架。我们的人工智能基础设施从算力优化到跨集群资源调度进行整体设计，以支持模型训练的执行。尽管研发仍是我们最大的投资领域，但研发费用占收入的百分比已大幅下降，从2023年的逾2,000%降至2024年的619.1%，并进一步从截至2024年9月30日止九个月的712.9%降至2025年同期的337.4%。我们研发费用的同比增长率在2024年及截至2025年9月30日止九个月分别为170.0%和30.0%，显著低于同期782.2%和174.7%的收入增长率，这充分体现了我们训练效率的提升以及基础设施的可扩展性——随着我们的业务从研究密集型开发阶段向规模化商业部署阶段转型。与此同时，我们基础模型的推理成本持续稳步下降，使其能够在高并发场景中得到更广泛的应用。

— 313 —

BUSINESS

industry scenarios. This decline has been driven by a combination of model architecture innovations, inference efficiency improvements, engineering optimizations, and reductions in the cost of compute. These factors are expected to continually lower our costs at a predictable rate.

Lower costs associated with model training and inference activities directly translates to more affordable AI-native products for future customers, thereby expanding our user base. Our average MAUs increased from approximately 3.1 million in 2023 to approximately 19.1 million in 2024, and further to approximately 27.6 million in the nine months ended September 30, 2025, demonstrating the rapid expansion of our user base. We aim to continue improving training and inference-related cost efficiency, targeting further reductions as a percentage of revenue through optimized infrastructure deployment, algorithmic innovation, and tighter integration of our training and inference workflows. These efforts will not only solidify our position as a global leader in high-performance foundation models but also enable us to deliver cost-effective AI-native offerings.

拓宽我们AI原生产品套件的变现渠道

我们的收入增长主要由AI原生产品及开放平台的快速扩张驱动，并采用多元化变现策略。我们已成功扩展和多样化变现来源，主要通过面向客户的AI原生产品的快速规模化以及面向开发者和企业的开放平台来实现。

我们的AI原生产品——如Hailuo AI及Talkie/星野——凭借领先的模型智能、多模态能力、用户体验以及差异化的模型驱动功能，展现出显著的市场吸引力。我们AI原生产品的吸引力推动了用户参与度和变现能力的增长，月活跃用户数（MAU）及付费用户数均实现快速增长。我们的平均月活跃用户数从2023年约310万增长至2024年约1,910万，并从截至2024年9月30日止九个月的约1,460万增长至2025年同期的约2,760万。AI原生产品的付费用户数从2023年的约119,700人增长至2024年的650,300人，并从截至2024年9月30日止九个月的约489,100人增长至2025年同期的约1,771,600人。我们通过多种方式实现产品变现，包括订阅、应用内充值、企业API使用以及在线营销服务。此外，我们通过引入新的付费功能和变现渠道提升了用户变现水平，例如推出MiniMax Agent及相应的高级订阅套餐。用户需升级至付费套餐或购买积分方可解锁特定MiniMax Agent服务，包括高峰时段优先访问权限及抢先体验测试版功能，这有助于提升每位付费用户的平均消费金额。我们AI原生产品的每位付费用户平均消费金额从2023年约6美元增长至2024年约11美元，并从截至2024年9月30日止九个月的约7美元增长至2025年同期的约15美元。因此，我们来自AI原生产品的收入显著增长，

– 314 –

BUSINESS

increasing from US$0.8 million in 2023 to US$21.8 million in 2024, and from US$13.5 million in the nine months ended September 30, 2024 to US$38.0 million in the nine months ended September 30, 2025, underscoring the commercial traction and user loyalty generated by our market-leading products.

Complementing our monetization strategy, our developer and enterprise facing Open Platform has become a driver of revenue growth, enabling enterprises to integrate our foundation models into their own applications and services. We plan to continue investing in the development and refinement of our AI-native products and Open Platform, with a focus on improving engagement, multi-modal capabilities and enterprise adoption. We will expand R&D to enhance existing products and to launch new AI-native applications that leverage ongoing advances in our models. To support this, we intend to expand our product development and commercialization organization, as well as our international sales and marketing team, by hiring approximately 14 additional specialists per annum over the next five years. These hires will enable us to improve user experience and scale adoption in overseas markets. Driven by our technology leadership and multi-modal capabilities, our paying users, defined as users who have individually consumed no less than US$50 worth of API calls (or its equivalent in other currencies), increased from approximately 400 in the nine months ended September 30, 2024 to approximately 2,500 during the same period in 2025. As a result, revenue from the Open Platform and other AI-based enterprise services grew significantly from US$2.7 million in 2023 to US$8.7 million in 2024 and from US$5.9 million in the nine months ended September 30, 2024 to US$15.4 million in the nine months ended September 30, 2025.

Moving forward, we plan to further deepen monetization through strategic enhancements of our AI-native product offerings — including the introduction of premium subscription tiers, enhanced features, and targeted geographic expansion into international markets. Additionally, by continuously refining our model capabilities and user engagement strategies, we aim to strengthen our competitive positioning, further boosting user acquisition and retention for individual users as well as developer and enterprise customers, and overall revenue growth.

Optimizing Organizational Efficiency and Scalability

We have strategically developed a lean, agile, and cross-functionally integrated organizational structure that significantly enhances our ability to rapidly innovate and scale with discipline. This flat organizational design fosters close collaboration between our research, product, and marketing teams, enabling rapid iterations and allocation of resources to the most impactful opportunities. As a result, despite growth in operational scale, we maintained a disciplined approach to organizational expansion, achieving significant productivity improvements and cost efficiency gains. For example, our administrative expenses as a percentage of revenue decreased sharply from 220.1% in 2023 to 47.1% in 2024, and further decreased from 49.4% in the nine months ended September 30, 2024 to 41.3% in the nine months ended September 30, 2025, and selling and distribution expenses were reduced from 659.7% in 2023 to 285.0% in 2024, and further decreased from 274.4% in the nine months ended September 30, 2024 to 73.6% of revenue in the nine months ended September 30, 2025, demonstrating effective cost management.

– 315 –

BUSINESS

展望未来，我们计划通过持续优化组织架构和工作流程来进一步提升运营效率，并继续采取有机用户获取策略。我们打算利用已建立的内部流程，在控制增量成本的同时快速扩展基础模型及AI原生产品。为进一步提升效率，我们正将自研模型技术整合至内部运营中，包括部署内部开发的AI智能体，用于软件开发支持、工作流自动化及日常任务处理。我们对保持精简而高效的团队的持续承诺，将使我们在规模扩张过程中实现可持续的成本管控，从而确保组织效率始终作为我们商业化路径的基础支柱。我们计划继续采取有机的用户及客户获取策略，在不依赖大规模品牌推广及用户获取支出的前提下扩展全球用户及客户群。上述举措旨在确保组织效率与营收规模同步提升，从而支持利润率随时间持续改善。我们的长期目标是维持一支精干、执行力强的团队，能够在不按比例增加成本的前提下，实现引领细分市场的创新及全球增长。

基于上述情况，我们预计将在截至2025年12月31日止十二个月内，达到《上市规则》第18C章关于商业公司的收入要求，主要驱动力来自面向消费者产品矩阵的商业扩张以及开放平台（Open Platform）采用率的持续提升。为展示我们在追踪记录期间的快速增长及变现潜力，就我们的AI原生用户产品而言，付费用户数量从2023年的约119,700人增长至2024年的约650,300人，并进一步增长至截至2025年9月30日止九个月的约1,771,600人。就我们的开放平台而言，付费用户数量（定义为单独消费不少于50美元API调用费用（或其他货币等值金额）的用户）从截至2024年9月30日止九个月的约400人增长至2025年同期的约2,500人。凭借我们所建立的坚实基础以及我们持续把握的技术与市场机遇，我们相信自身具备维持并发展业务的能力。然而，我们对达到商业公司资格的预期存在各种不确定性，并取决于我们在行业多个关键维度上有效竞争的能力，包括我们基础模型的智能水平、我们AI技术的可负担性、我们AI原生产品的变现能力，以及我们组织的效率与可扩展性。此外，我们实现商业成功的能力还取决于我们的技术创新速度、定价策略、吸引和留住顶尖人才的能力，以及品牌知名度和客户信任度。有关估计达到及实现《上市规则》第18C.03(4)条收入要求的时间框架所涉及的相关风险及潜在障碍，请参阅"风险因素——与我们产品商业化相关的风险"。基于上述情况，我们的董事认为，且联席保荐人亦同意，我们的业务具有可持续性。

－316－

BUSINESS DATA SECURITY AND PRIVACY Data security and privacy protection are among our top priorities. We provide services based on proprietary AI models developed in-house, including applications targeted at individual users, such as MiniMax, Hailuo AI, MiniMax Audio and Talkie/Xingye, and enterprise-oriented services offered through the Open Platform. At the model level, we have implemented organizational safeguards and technical control measures at the input, model, and output levels in accordance with regulatory requirements applicable to AI companies. Model Input For model training, we use lawfully authorized datasets, and other legally available data. Such training data do not involve customer data or personal data that can identify natural persons. We take multiple measures to ensure the quality and reliability of training data, including but not limited to procure data from professional data providers, establishing annotation guidelines, enforcing personnel management, and conducting training and assessments to ensure annotation quality and consistency. Annotation (or data labelling) refers to the process of tagging or adding information to training data so that the model can understand the meaning and intended use of such data. For example, data inputs may be annotated according to its topic, sentiment, format or whether it contains illegal or harmful content. Annotation enables the model to better interpret input data and produce outputs that are aligned with expected, safe and legally compliant behaviour. In addition, by annotating and identifying harmful or inappropriate content, the risk of the model generating illegal, discriminatory or otherwise harmful information is reduced. On the basis of such datasets, we further engage domain experts to generate high-quality training data tailored to our specific needs. For training data, we adopt methods such as screening and test questions to ensure fairness, non-discrimination, and compliance with third-party rights. Specifically, in the data preparation stage, we process and clean training datasets to remove harmful content, including content that is illegal or non-compliant, endangers public security, is pornographic or violent in nature, or contains discriminatory elements. During model training, we analyze preliminary model outputs to identify risk-related samples (such as harmful or discriminatory content). These samples are then subject to human review, including annotation, rewriting or other corrective processing, to generate positive and negative examples. The resulting annotated and refined datasets are subsequently incorporated into further rounds of training to enhance the model's ability to recognize and appropriately manage compliant content and potential risks.

– 317 –

BUSINESS

模型与应用安全

我们已依据中国关于生成式人工智能的相关法规，就我们的自研模型（包括"abab"模型系列、"abab多模态"模型系列及"MiniMax"模型系列）向上海市互联网信息办公室完成了必要的模型备案。对于在提供网络信息服务过程中涉及算法推荐技术的应用服务，我们已通过互联网信息服务算法备案系统完成算法备案程序并获得相应备案结果。在适用情形下，我们亦已在我们的网站及应用程序的显著位置公示备案信息及备案编号。

模型输出

我们对模型生成内容的合规性实施严格管控。所有人工智能生成内容均依据适用的内容标识要求予以明确标注。我们要求用户签订服务协议，明确双方权利与义务，以防止用户利用我们的服务生成可能违反法律法规、危害网络安全或数据安全、或侵犯他人权利的内容。此外，我们高度重视对未成年人的保护，已制定"未成年人保护政策"，以保障未成年用户的内容安全。对于模型输出，我们亦采用审核、抽查及测试题目等方式，以确保公平性、非歧视性及对第三方权利的合规保护。

应用安全

在提供人工智能服务的过程中，我们会处理用户的个人信息。在收集和处理相关信息之前，我们通过隐私政策向数据主体告知处理目的，并依据适当的法律依据开展相关处理活动。我们已采取适当的技术与组织措施，以确保个人信息安全及数据主体权利的保护。

数据安全

为支持我们产品和服务的功能运行，并遵从相关法律法规的要求，我们在日常业务过程中会根据需要收集、处理和存储用户的若干类型个人数据，包括基本信息（如账户信息）、设备与网络数据（如IP地址）以及交易相关信息（如适用）。我们亦会处理通过自研模型生成的自创数据。我们使用上述数据用于运营和维护我们的服务、提升用户体验、开展数据分析以改善性能、检测有害活动和滥用行为，以及履行法律法规义务。

－318－

BUSINESS

We outline below the categories of data involved during our operations. As the specific data types vary depending on product functionality, the principal categories are as follows:

| 数据类型（类别）Data Type (Category) | 数据字段示例 Examples of Data Fields | | |---|---|---| | 用户提供的数据 User-provided data 对于用户数据的收集和使用（如有），我们通过用户协议和隐私政策事先获得用户同意。For the collection and use of user data (if any), we obtain users' prior consent through our user agreements and privacy policies. | 用户基本信息 Basic user information | 例如，用于账户注册的手机号码或电子邮件地址 E.g. mobile number or email address used for account registration | | | 设备信息 Device information | 例如，设备类型和操作系统版本 E.g. device type and operating system version | | | 支付信息 Payment information | 例如，购买付费服务时产生的支付数据 E.g. payment data generated when purchasing paid services | | 用户输入数据 User input data | 用户输入的内容 Content input by users | 例如，使用生成式人工智能产品时输入的提示词 E.g. prompts entered when using generative AI products | | 系统生成／输出数据 System-generated/output data | 产品输出的内容 Content output by products | 例如，基于用户提示词生成的响应内容 E.g. responses generated based on user prompts |

In limited cases where enterprise customers deploy our solutions on their own systems or servers, the data generated are generally managed and processed by such customers. We set out the respective rights and obligations of the parties in the relevant contracts following mutual negotiation with our customers.

Our data processing activities are subject to applicable laws and regulations on cybersecurity, data privacy, and personal data protection in the jurisdictions where we operate. To maintain compliance with applicable laws, regulations, and industry best practices, we have adopted data protection and information security measures, such as privacy policies, technical safeguards and governance measures, which cover the entire lifecycle of data we collect, process, and store.

Privacy Policies: We have established privacy policies for relevant products and have informed users of such policies about the types of and the usage of personal data to be collected. To minimize the volume of personal data collected, we collect, use, and store users' personal data solely within the scope necessary to achieve the stated processing purposes.

Technical Safeguards: We have implemented a series of technical safeguards to enhance data security, such as data encryption and backup strategies, role-based access control, and audit of relevant logs. Queries, downloads, and printouts involving sensitive personal information require approval and a legitimate purpose. Personal data is stored securely with appropriate technical and organizational safeguard. We store the personal data of PRC and overseas users on separate locations. Based on due inquiry of our PRC legal advisor, at the personal data level, our services and products do not involve any cross-border data transfer between our PRC and overseas storage locations.

– 319 –

BUSINESS

治理措施：我们亦已采取组织及治理措施，以确保数据安全。鉴于员工了解数据保护原则的重要性，我们制定了信息安全培训政策，并对系统管理员及管理人员进行技术培训。我们已委任专职个人信息保护负责人，负责监督相关合规事宜。我们亦已建立事件响应程序，以迅速处理并减轻任何潜在数据泄露事件的影响。

使用第三方云服务及公共云基础设施：我们意识到，使用第三方云服务及基础设施提供商以及采用"公共"云服务可能使我们面临服务中断、终止或延迟等风险，以及由此类第三方服务导致的潜在数据泄露或损毁风险。为降低上述风险，我们已采取以下措施：

• 第三方供应商管理：我们选择可靠的服务提供商，并在合同中明确规定其数据安全义务及责任。我们持续进行监督与评估，以确保服务提供商具备提供安全稳定服务的能力。

• 内部数据保护措施：我们已建立内部管理制度，包括《数据安全管理政策》，并设立应急响应小组，负责处理因第三方问题引发的潜在数据安全事件。一旦发生安全事件，我们将及时展开调查并采取补救措施。为防止数据丢失或损毁，我们已实施有效的数据备份与恢复策略。

于追踪记录期间及截至最后实际可行日期，在我们经营所在的中国及其他司法管辖区，(i) 我们未收到相关主管机关发出的任何通知，表明我们处理的数据已被列为重要数据或核心数据；(ii) 就我们所知，我们未曾发生任何与数据安全相关的重大数据泄露、违规或其他损失；(iii) 我们未收到任何第三方就重大违反数据隐私或保护法律提起的法律诉讼。根据我们中国法律顾问的意见，本集团的网络安全及数据处理活动在所有重大方面均符合中国适用的数据隐私及安全相关法律法规的规定。鉴于我们人工智能原生产品的用户数量及美国市场所产生的相关收入，我们已聘请美国数据法律顾问就数据保护合规问题进行尽职调查，根据我们美国数据法律顾问的意见，从美国角度而言，(i) 于追踪记录期间及截至最后实际可行日期，本集团在美国的主要业务运营在所有重大方面均符合适用的数据保护法律法规；(ii) 根据其对本集团业务运营的审查及美国目前法规的不确定性，主管机关目前对本集团采取监管执法行动的可能性较低。

－320－

BUSINESS

Based on the relevant diligence findings, as advised by our U.S. data legal advisor, from a U.S. legal perspective, that our business operations in the U.S. have been carried out in material compliance with applicable data protection laws and regulations during the Track Record Period and up to the Latest Practicable Date. Based on our U.S. data legal advisor's review of our business operations and considering the current regulatory uncertainties in the U.S., such legal counsel further concludes that the probability of a regulatory enforcement action against us is currently low. Based on the view of our U.S. data legal advisor and the Joint Sponsors' review of the legal due diligence report prepared by such counsel, the Joint Sponsors have reasonable grounds to believe that the view expressed fairly represents the views of the U.S. data legal advisor. However, we cannot rule out the possibility of future enforcement actions. This is primarily due to (i) the evolving nature of the U.S. regulatory environment for AI technologies and (ii) the increasing legislative and enforcement initiatives at both federal and state levels concerning AI transparency and data usage.

Our Directors confirm that, as advised by our PRC legal advisor, U.S. data legal advisor and Singapore legal advisor, (1) we had complied with all applicable laws and regulations relating to data protection and privacy in all material aspects; (2) did not experience any material data leakage, breach or other losses in relation to data security; and (3) had not received any investigation, penalty or material third-party claim with respect to third party's rights to data protection; and (4) as advised by our PRC legal advisor, our business did not involve any cross-border transfer of personal data between our PRC and overseas data storage locations during the Track Record Period and up to the Latest Practicable Date, and as advised by our U.S. data legal advisor and Singapore legal advisor, our operations in the U.S. and Singapore, respectively, are compliant in all material aspects with U.S. and Singapore cross-border transfer obligations.

COMPETITION

We operate in a fast-evolving and increasingly competitive global market for foundation models. Our market is characterized by rapid innovation cycles, growing demand for AI-native applications, and increasing interest from both customer and enterprise users.

Our principal competitors include multinational technology companies with extensive R&D capabilities and computing resources, regional technology companies with established distribution channels and vertical domain expertise, and specialized start-ups focused on vertical model optimization and application-layer deployment. In China, we compete with both domestic AI labs and leading internet companies as well as international foundation model developers offering comparable foundation model capabilities.

The principal competitive factors in our industry include model intelligence and versatility, scalability of computing infrastructure, efficiency in long-context modeling, capabilities in multi-modal alignment and generation, user experience, breadth and depth of commercial applications, ecosystem development, technological innovation, pricing strategy, talent acquisition, brand recognition and customer trust.

– 321 –

业务

我们通过专注于长上下文建模和可扩展的多模态架构设计来实现差异化竞争，这使我们能够构建具备处理文本、视觉和音频之间复杂、多维度交叉能力的模型。

我们相信，凭借我们专有的模型开发管道、对基础研究的持续投入、在消费者和企业市场中快速扩展的产品应用场景，以及支持创新与采用的不断壮大的开发者生态系统，我们具备在上述各维度有效参与竞争的能力。

然而，我们的许多现有及潜在竞争对手拥有更长的运营历史、更广泛的全球布局、更成熟的用户社区，以及更便捷获取数据、人才和计算基础设施的渠道。基础模型市场格局仍高度动态且资本密集，底层算法的持续进步、商业化努力的不断加强以及监管标准的持续演变，可能进一步加剧未来的竞争。

请参阅"行业概览"及"风险因素——与我们产品商业化相关的风险——我们所处的全球基础模型行业快速演变且竞争日益激烈。我们的业务面临持续的技术进步与行业变革。若我们未能持续创新并适应客户需求的变化，我们的竞争地位将受到影响，我们的业务、财务状况及经营业绩可能受到重大不利影响。"

保险

我们认为我们的保险覆盖范围充分，符合一般市场惯例及中国适用的法律法规要求。我们依据相关中国法律法规为员工提供法定社会保险，包括养老保险、医疗保险、生育保险、失业保险、工伤保险及住房公积金缴存。除法定保险外，我们还为员工提供商业医疗保险，以增强其健康保障并促进整体健康水平。

与中国现行市场惯例一致，由于以下险种在中国法律下并非强制要求，我们不维持营业中断险、产品责任险、关键人物寿险，以及涵盖我们网络基础设施、数据中心或其他信息技术系统损害的保险。我们会不时评估我们的保险需求，并可能根据业务运营、风险状况及市场条件的变化对保险覆盖范围作出调整。

— 322 —

BUSINESS

During the Track Record Period and up to the Latest Practicable Date, we did not experience any material incidents that would have resulted in a significant insurance claim, nor did we make any material insurance claims in relation to our business operations. For more details, see the section headed "Risk Factors — Risks Relating to Our Business and Industry — We may not have sufficient insurance coverage to cover our business risks" in this prospectus.

ENVIRONMENTAL, SOCIAL AND GOVERNANCE

Overview

We consider environmental, social and governance ("ESG") matters to be an integral part of our operations. With a view to becoming a socially responsible enterprise, we are firmly committed to enhancing our ESG practices and promoting long-term sustainable development. During the Track Record Period and up to the Latest Practicable Date, as advised by our PRC Legal Advisor, we had not been subject to any material claim or penalty or accident in relation to health, work safety, social and environmental protection, and we had been in compliance with the relevant PRC laws and regulations in all material aspects.

Our ESG Governance Structure

To further strengthen the management of ESG-related matters, we have established a comprehensive governance structure led by our Board, which holds ultimate responsibility for the oversight, review and decision-making of ESG affairs. The Board plays a central supervisory role in ensuring that our ESG strategies are effectively aligned with our overall business objectives. It also provides leadership in implementing ESG initiatives, while ensuring compliance with applicable laws, regulations, standards and the requirements of relevant regulatory authorities. In addition, the Board is responsible for identifying and managing ESG-related risks to ensure the effective execution of our ESG objectives throughout the organization. We have conducted ESG-related knowledge training sessions for the Board, covering dimensions including concepts and regulations, governance, management and disclosure. We are committed to continuously enhancing the Board's understanding of ESG-related matters and to helping the Board make decisions that align with the Company's sustainable development plans.

In order to institutionalize and standardize our ESG practices, we have formulated an ESG Management Policy, which clearly defines the responsibilities of each relevant party in the day-to-day management of key ESG matters. This policy serves as a foundation for ensuring the effective execution and continuous improvement of our ESG management framework.

– 323 –

BUSINESS

ESG相关风险及管理

我们高度重视ESG相关风险，并已将其纳入整体战略规划之中。为此，我们已制定明确的ESG管理原则及策略。我们的董事会承担监督职责，并定期审查ESG相关事宜，以确保与业务发展目标保持一致，同时管控ESG相关风险，确保ESG目标及实施流程的有效性。

我们认为，回应利益相关方的期望是推动企业长期增长的关键动力。我们的利益相关方涵盖内部及外部各方，包括员工、客户、供应商、业务合作伙伴、投资者、监管机构及各类社会群体。通过建立有效的沟通渠道和参与机制，我们致力于倾听并了解利益相关方的关切，从而使我们能够作出更为明智、有效的ESG相关决策。

结合利益相关方的关切及我们业务的具体特点，我们已识别并评估了重要ESG相关议题及其相关风险与机遇。为有效管理上述议题，我们已建立完善的ESG风险评估机制，并定期组织管理层开展专项评估工作。我们采用多维度评分体系，从风险影响程度、发生可能性及时间紧迫性等核心指标出发，对各类ESG风险进行全面量化评分，以确保风险管理的科学性。通过这一机制，我们能够及时识别优先风险并制定相应的应对策略。被认为对我们业务具有重大影响的ESG相关议题包括：气候变化应对、人力资本发展、商业道德、数据与隐私安全、供应链管理、知识产权保护及普惠科技。

环境

凭借我们在人工智能技术开发与应用方面的专业积累，我们深刻认识到气候变化对全球生态系统及商业环境所带来的影响。我们已将气候行动纳入战略部署，并致力于通过技术创新与负责任的运营方式减少碳足迹，以推动构建更加绿色、可持续的数字未来为愿景。

– 324 –

BUSINESS Climate Risk Management We recognize the interconnection between our business and climate change, and have identified and assessed the climate-related risks that may affect our operations, supply chain and stakeholders. These risks primarily include:

| Risk Type | Potential Impact | Risk Management Actions | |---|---|---| | Physical Risks • Acute risks such as typhoons and floods, and chronic risks such as rising sea levels and long-term shifts in climate patterns. | Disruption to employee commuting, damage to office infrastructure, service interruptions at computing facilities, impacting business continuity; adverse effects on operations, leading to increased operating costs and potential decline in revenue. | Establish and continuously refine contingency plans and recovery measures to respond to extreme weather events; safeguard employee health and safety; ensure product and service stability during customer use. | | Transition Risks • Policy risks arising from tightening carbon emission regulations; reputational risks due to increasing stakeholder expectations for green development. | Potential operational challenges and increased compliance costs; pressure to adapt products and services to meet low-carbon standards. | Continuously monitor the evolving regulatory environment; actively engage stakeholders; develop and implement more sustainable production and operational models to meet regulatory and market demands. |

We actively implement the concept of green development and provide clear direction for environmental management efforts by setting environmental objectives. We have developed policies regarding ESG targets management, as well as the ESG target review mechanism, requiring the management to review the target regularly. Based on the international ESG-related standards, industry peers and development situation, we have established the following environmental objectives:

• Greenhouse Gas Emission Reduction Objective: By 2030, reduce scope 1+2 greenhouse gas emissions per unit of operating revenue by 55% compared to 2024.

• Energy Efficiency Objective: By 2030, reduce total energy consumption per unit of operating revenue by 55% compared to 2023.

• Water Efficiency Objective: By 2030, reduce total municipal water consumption per unit of operating revenue by 20% compared to 2023.

• Waste Reduction Objective: Continuously promote the concept of green office practices.

– 325 –

BUSINESS

环境与气候相关目标及指标

我们已将办公场所纳入环境和气候相关指标的计算范围。由于我们业务的性质，我们不从事实物产品制造，也不拥有自有车队，因此不产生大气污染物，亦不使用实物包装材料。主要绩效指标如下：

温室气体排放

根据世界资源研究所与世界可持续发展工商理事会发布的《温室气体核算体系》、政府间气候变化专门委员会发布的《2006年IPCC国家温室气体清单指南》，以及中华人民共和国国家发展和改革委员会发布的《其他行业工业企业温室气体排放核算方法与报告指南（试行）》，我们已将主要温室气体排放源识别为范围二排放。此外，我们已建立完善的ESG管理政策以应对温室气体排放，从而实现有效且负责任的排放管理。

| 指标/单位 | 2022年 | 2023年 | 2024年 | 截至2025年9月30日止九个月 | |---|---|---|---|---| | 温室气体排放总量（1）（吨二氧化碳当量） | 11.94 | 124.02 | 320.53 | 425.76 | | – 范围二温室气体排放 | 11.94 | 123.97 | 320.11 | 425.09 | | – 范围三温室气体排放（2） | – | 0.05 | 0.42 | 0.67 | | 单位收入温室气体排放量（吨二氧化碳当量/百万美元） | N/A | 35.84 | 10.50 | 7.96 |

截至12月31日止年度

注：

（1） 由于我们目前不拥有任何车辆，范围一温室气体排放不适用。

（2） 温室气体排放（范围三）的计算范围包括废水排放处理过程中产生的排放量。

－ 326 －

BUSINESS

| Indicators/Unit | 2022 | 2023 | 2024 | For the nine months ended September 30, 2025 | |---|---|---|---|---| | Energy consumption(3) (kwh) | 20,983.24 | 214,691.12 | 566,043.73 | 735,824.16 | | Energy consumption per unit of revenue (kwh/US$ ten thousand) | N/A | 62,049.46 | 18,544.83 | 13,760.67 | | Municipal water consumption(4) (ton) | 0 | 68.37 | 577.08 | 935.64 | | Municipal water consumption per unit of building area (ton/m²) | – | 0.01 | 0.07 | 0.11 | | Wastewater discharge (ton) | 0 | 54.70 | 461.66 | 748.51 | | Wastewater discharge per unit of building area (ton/m²) | – | 0.06 | 0.02 | 0.09 |

For the year ended December 31,

Notes:

(3) As we currently do not own any vehicles and are not involved in any production processes, our total energy consumption solely comprises electricity usage within our existing office premises in Beijing, Shanghai, Chongqing and Chengdu during the relevant period, including lighting, air conditioning, and office equipment.

(4) The scope of municipal water consumption statistics includes our offices in Beijing, Shanghai, Chongqing and Chengdu.

Due to the nature of our business, our environmental impact is relatively limited. Nevertheless, we are committed to the principles of green development. We consistently comply with environmental laws and regulations, including the Environmental Protection Law of the People's Republic of China, and continuously monitor and respond promptly to changes in external environmental regulations to ensure compliance. We have formulated environmental protection policies to manage environmental matters within our office premises.

Our operations rely on high-efficiency, energy-saving AI infrastructure, prioritizing the use of low-power consumption equipment. We employ next-generation energy-efficient servers and intelligent cooling systems, utilizing dynamic management technologies and energy-saving solutions to maintain high power usage effectiveness (PUE) values. We actively encourage and cooperate with our data center suppliers to achieve carbon-neutral operations, working on reducing our overall environmental footprint.

– 327 –

BUSINESS

在节能减排方面，我们制定了能源管理政策，对办公室内的能源消耗进行规范管理。我们致力于降低单位面积能耗，提高产品和服务的能源效率，并积极探索数据中心的可持续能源选项。

在废弃物管理方面，我们严格遵守《中华人民共和国固体废物污染环境防治法》及其他相关废弃物管理法规。鉴于我们的业务性质，所产生的废弃物主要为办公废弃物。

在水资源管理方面，用水主要发生在办公区域。我们鼓励员工养成节约用水的行为习惯。

社会责任

劳工标准、健康安全与发展

我们遵守《中华人民共和国劳动法》及其他适用法律法规，并制定了人力资源管理制度及员工手册等人力资源政策。上述政策明确规定了我们在培训、激励、福利及其他雇佣相关事项方面的标准与要求。我们遵循国际劳工组织（ILO）的相关公约，承诺在雇佣的各个方面平等对待所有员工。我们承诺禁止任何形式的童工和强迫劳动，积极维护所有员工结社自由和集体谈判的权利。我们致力于禁止任何基于种族、性别或其他因素的就业歧视，且不会以此类原因解雇员工。我们为所有员工提供同工同酬及其他员工福利，保障其合法权益。我们为所有员工提供平等的发展机会，并持续提升劳动力多元化水平。

我们致力于平等对待每一位员工，无论其性别、年龄、国籍或文化背景，并努力维护透明、公平、公正的工作环境。此外，我们遵守《未成年人保护法》及《禁止使用童工规定》等法律法规。我们坚决反对并禁止使用强迫劳动或童工，任何经查实的违规行为将受到严肃处分，以保护员工的合法权益。我们依法参加各项强制性社会保险，包括养老保险、医疗保险、失业保险、工伤保险和生育保险，以及住房公积金。详见"业务——我们的组织与人员——基本信息"。

我们致力于保障员工的健康与安全，遵守中国所有适用的职业健康与安全法律、法规及标准。我们已制定职业健康与安全管理内部政策，并定期开展工作场所安全检查，以识别和消除潜在危害。上述措施

— 328 —

业务

包括对消防安全设备进行例行维护和升级，以确保安全的工作环境。在业绩记录期间及直至最后实际可行日期，我们未发生任何工伤或死亡事故，亦未经历任何重大运营或行政事故。

我们制定了《员工培训管理政策》并建立了完善的培训体系。我们为员工提供培训机会，例如每周五举行的午间研讨会，由经验丰富的员工或外部特邀嘉宾分享其在技术和行业方面的见解。

供应链管理

为规范供应商在环境、社会及企业管治方面的表现，并有效管理整个供应链中与环境、社会及管治相关的风险，我们制定了《供应商行为准则》。该准则规定了供应商在与我们开展业务时须遵守的标准。在环境方面，我们鼓励供应商采取适当的环境管理措施，以尽量减少其运营过程中的碳排放及对环境的不利影响。在社会方面，我们要求所有供应商遵守适用的法律法规。供应商亦须妥善保护其处理或加工的任何机密、专有或个人信息，并避免从事任何可能侵犯我们知识产权或损害我们声誉的活动。在管治方面，供应商须避免任何旨在谋取商业机会或获取不当利益的不正当交易。

道德与合规

我们致力于坚守最高标准的商业道德，并严格遵守所有适用的法律法规，包括但不限于《中华人民共和国刑法》中有关贿赂、勒索、欺诈及洗钱的相关规定。我们制定了反腐败政策，明确禁止一切形式的贿赂、勒索、欺诈及洗钱行为，以确保我们业务运营的合法性与诚信性。为有效预防和打击腐败，我们为董事及员工提供培训，以强化道德行为规范。在业绩记录期间，我们未遭遇任何对我们的业务或运营已产生或可合理预期将产生重大不利影响的腐败、欺诈或其他不当行为事件。

包容性技术

我们致力于通过技术创新推动包容性发展。我们通过技术赋能、教育支持及文化交流促进社会的可持续进步。通过提供多语言产品，我们致力于确保每种语言都能被倾听，每种文化都能被理解。

－329－

BUSINESS

作为我们社区参与工作的一部分，我们积极开源我们的模型，并构建开放平台，为全球开发者提供便捷的人工智能基础设施，从而降低技术壁垒，培育充满活力的创新生态系统。

PROPERTIES

为支持我们的业务运营，包括研究与开发、工程及行政管理，我们在中国大陆多个重要城市租赁了物业，包括北京、上海、成都、深圳及重庆。我们的主要行政办公室位于中国上海。截至最后实际可行日期，我们共租赁了七处物业，合计建筑面积约为5,983.3平方米，主要用于办公及研究与开发用途。

我们不拥有任何物业。截至最后实际可行日期，我们租赁的物业中，没有任何一处的账面价值等于或超过我们综合总资产的15%。因此，根据《上市规则》第5章及《公司（豁免公司及招股说明书遵从条文）公告》第6(2)条，我们获豁免遵守《公司（清盘及杂项条文）条例》第342(1)(b)条有关第三附表第34(2)段的规定，否则须就我们于土地或建筑物的权益提交估值报告。

截至最后实际可行日期，我们七处租赁物业中有五处尚未在相关中国政府主管部门办理登记。我们的中国法律顾问建议，未经登记不影响租赁协议在中国法律下的有效性或可执行性；然而，相关政府主管部门可能要求我们在规定期限内办理租赁协议登记，且每份未登记租赁可能被处以人民币1,000元至人民币10,000元的罚款。我们将继续尽最大努力与出租方协调，推动所有未登记租赁协议向主管部门办理登记，并将持续监控对适用中国物业租赁法律法规的合规情况。请参阅"风险因素——与我们业务及行业相关的风险——我们可能因未能按照适用法律法规登记和备案租赁协议而承担责任，这可能使我们受到行政处罚。"

就上述租赁物业而言，我们的董事确认，鉴于市场上存在可比替代物业，预计无需花费大量时间或成本来寻找可比替代物业或将我们的业务迁移至可比替代物业。我们的董事进一步确认，如需进行迁移，该迁移不会对我们的运营及财务状况产生重大影响。

– 330 –

BUSINESS 法律程序及合规情况 法律程序 于往绩记录期间及直至最后实际可行日期，本集团并未涉及任何我们认为会对我们的业务、经营业绩、财务状况或声誉及合规情况产生重大不利影响的实际或待决法律、仲裁或行政程序（包括任何破产或接管程序）。 版权侵权诉讼 2025年9月16日，包括迪士尼（Disney）、环球影业（Universal）及华纳兄弟探索（Warner Bros. Discovery）在内的一批主要美国电影制片公司（"原告方"）就本集团旗下视觉生成平台Hailuo AI，向美国加利福尼亚州中区联邦地区法院提起民事诉讼（"诉状"）。

在诉状中，原告方依据《美国版权法》提出直接版权侵权及间接版权侵权（包括辅助侵权及替代侵权）的指控。上述指控属商业性质索赔。

• 指控一——直接版权侵权：原告方指称，本公司通过Hailuo AI自行生成并展示了描绘原告方所拥有的多个知名电影及动画角色的视频和图像，并据此主张本公司应被视为直接复制、展示及传播了其作品。诉状中附有原告方自行生成的截图以佐证上述主张。

• 指控二——间接侵权，包括辅助侵权及替代侵权：原告方指称，即便相关内容由个别用户生成，本公司仍应依据辅助侵权及替代侵权原则承担法律责任。原告方指称，本公司知晓或理应知晓用户可能生成描绘原告方角色的内容，且本公司据称从该等使用中获益。

在救济请求方面，原告方主要寻求（除其他事项外）以下救济：

• 以实际损害赔偿及可归因于涉嫌侵权行为的利润追缴为形式的金钱救济；或由其选择，适用法定损害赔偿；

• 禁令及其他衡平法救济，以禁止本集团侵犯原告方享有版权的作品，并要求就Hailuo AI实施有效的版权保护措施；

• 判令原告方获得合理律师费及诉讼费用；及

• 法院认为公正合理的其他法律或衡平法救济。

– 331 –

BUSINESS

Legal Analysis

These claims are commercial disputes in nature, and having considered advice from our U.S. litigation advisor, our Directors believe that they are without merit in all material respects and that there is insufficient evidence to support them, based on the below analysis for the two major allegations. Based on the Joint Sponsors' due diligence conducted, there was no reasonable basis for the Joint Sponsors to disagree with the Directors' view that the claims are without merits in all material respects:

Allegation 1: Direct copyright infringement

Having considered advice from our U.S. litigation advisor, the Company categorically denies this allegation for the reasons set out below:

• Hailuo AI users trigger content generation, not the Company. Direct infringement requires volitional conduct under the standard set in VHT, Inc. v. Zillow Group, Inc., 918 F.3d 723 (9th Cir. 2019). Hailuo AI only generates an output when a user types in a prompt or uploads a starting image. In that situation, it is the user who is deciding the output that is generated, not the Company. The Company provides the tool; it does not select any specific characters or scenes. Further, Hailuo AI's outputs are generated through algorithmic transformation rather than passive copying and based on our internal records, the overwhelming majority of Hailuo AI outputs user-created content have nothing to do with the Plaintiffs' characters, further undermining the presence of any volitional conduct on the part of the Company. That means a key element of direct infringement is missing, as it is not the service provider itself carrying out the copying, and volitional conduct cannot be demonstrated.

• The Plaintiffs have stretched what is actually protected. The Plaintiffs assume that every image that looks like one of their characters is automatically covered by their U.S. registrations, including those depicting realistic or live-action characters that may not be protected, as human beings are not usually subject to copyright protection. In order for a copyright infringement claim to succeed, the Plaintiffs must in fact have copyright registrations that cover the expressive elements of characters that are eligible for copyright protection. The Plaintiffs are overstretching the bounds of their copyright protections to suggest that a particular depiction of a human-like character is itself protected, and that the registration they rely on actually covers that depiction. As such, the Company may raise defenses based on the scope of the copyright protections claimed by the Plaintiffs and the underlying copyrightability of some of the characters identified by the Plaintiffs.

– 332 –

BUSINESS

•

在模型开发中使用相关材料可能构成"合理使用"。本公司预计将主张，即便相关使用行为确实存在，该等使用类型亦应受"合理使用"抗辩的保护。该抗辩已被编纂于《版权法》第107条。合理使用是一项促进言论自由的法律原则，允许在特定情形下未经许可使用受版权保护的作品。

本公司预计将主张，在用于训练海螺AI模型之前对视觉内容进行解构的过程具有高度转化性。原告的理论所描述的是一种中间性、非表达性的过程，其中相关作品据称作为输入数据，使模型能够学习统计规律和一般特征，而非复制或替代任何表达性内容。多家美国地区法院已将此类所谓的中间性、分析性使用——例如对图像进行索引、搜索及其他非表达性处理以用于训练AI模型——认定为具有合理使用意义上的转化性。

**指控二：间接（辅助/替代）侵权**

在听取美国诉讼顾问的意见后，本公司以下列理由否认该指控，认为其缺乏依据：

•

海螺AI的创建目的及主要用途均为正常、合法的内容。我们认为，海螺AI不应被认定为协助和促进用户的违法使用行为。根据美国版权法，技术提供商不因提供一种具有实质性非侵权用途的工具而自动对用户的潜在滥用行为承担责任。提供通用AI工具不构成协助、教唆或促进用户未经授权行为的行为。海螺AI是一款专为合法用途而创建的通用创意工具，能够生成大量与原告角色完全无关的原创内容。根据美国最高法院在Sony Betamax案中确立的"商业主要商品原则"，能够用于实质性非侵权用途的主要商品的销售者，仅凭此点不足以承担间接侵权责任。一项主要用于普通、非侵权目的的通用服务，不应被定性为主要以侵权为目的而设计。因此，本公司可就辅助侵权主张援引商业主要商品抗辩。根据第九巡回法院关于辅助侵权的认定标准，一项被广泛使用且以合法用途为主的服务的提供者，不能仅因该服务可能被滥用而承担责任；还须证明提供者对特定侵权行为具有充分知情并予以鼓励，而本公司对此予以否认。

– 333 –

BUSINESS

•

The Company has no direct financial benefit tied to alleged infringements. According to the standard set in Perfect 10, Inc. v. Giganews, Inc., 847 F.3d 657, 673 (9th Cir. 2017), "to prevail on a claim for vicarious infringement, a plaintiff must prove 'the defendant has (1) the right and ability to supervise the infringing conduct and (2) a direct financial interest in the infringing activity'". Under Giganews, the "direct financial benefit" prong requires a causal nexus between infringement of the plaintiff's specific works and revenues received by the defendant. Id. The Company charges for Hailuo by access level (for example, higher resolution or priority features), not by whether the user tries to generate a character that looks similar to the Plaintiffs' copyrighted works. In other words, the Company does not earn extra because a user tries to create an image of a copyrighted character instead of an original cartoon. Therefore, the Company does not get any direct financial benefit from the specific conduct they complain about as the same fee is charged for non-infringing use. Thus, the Company may raise a defense that it does not have any direct financial interest in the infringing activity and is not liable for contributory infringement.

•

Knowledge/willfulness on the part of the Company is not established. The Plaintiffs have not shown that the Company had advance, specific knowledge of each alleged infringing work or that the Company refused to act after receiving such notice. There is no allegation that the Plaintiffs ever notified the Company of potential infringement beforehand. Without such proof, the heightened "knew and continued" characterization in the complaint — used to suggest willful conduct and increase damages — is unsupported. The Company had no knowledge of the Plaintiffs' concerns until the Company received a letter dated August 27, 2025 asserting in general terms that the Plaintiffs believed that the Hailuo AI was being used to infringe their copyrights. The Company responded promptly and began an investigation. Shortly thereafter, on September 16, the Plaintiffs for the first time identified specific characters whose copyrights they alleged were being infringed. Again, the Company promptly responded by adopting measures to prevent any possible infringements of those characters.

Based on the independent due diligence steps performed by the Joint Sponsors, including, among other things, (a) reviewing the plaintiffs' complaint and a legal memorandum prepared by the Company's U.S. legal advisors on the general principles of U.S. copyright law and recent case-law developments in the AI industry, (b) interviewing and discussing with the Company's U.S. litigation advisor regarding analysis of the merits of the claims, (c) examining the operation of Hailuo AI and the Company's risk-mitigation measures, including but not limited to its complaint and reporting mechanisms and its review and filtering processes, and (d) discussing with CIC, who is of the view that the Group's business practices are comparable to those of its industry peers, there was no reasonable basis for the Joint Sponsors to disagree with the Directors' view that the claims are without merits in all material respects.

– 334 –

BUSINESS

Potential Liability Exposure and Impact on the Company

Having considered advice from our U.S. litigation advisor, our Directors are of the view that the quantum of statutory damages in U.S. copyright claims is generally determined by two independent factors: (i) the statutory damages awarded per work, and (ii) the number of works deemed infringed.

The Plaintiffs have alleged in their Complaint that they are entitled to statutory damages of up to US$150,000 per infringed work, which is the maximum amount awardable per work under U.S. copyright law and is only awarded if infringement is found to be willful. The attachments to the Complaint identify approximately 500 registrations for motion pictures and television programs that are at issue in this case. Under 17 U.S.C. § 504(c)(1), courts calculate and award statutory damages on a per-work basis, regardless of the number of copies made, outputs generated, or instances of alleged infringement involving that work. In other words, the 500 copyright registrations listed in the attachments to the complaint is not counted based on number of times a video is generated using the work. Therefore, assuming the plaintiffs prevail and fully succeed in their claims, the worst-case scenario, as alleged by the plaintiffs, would be a monetary claim of US$75 million in statutory damages, calculated by multiplying the number of alleged works by the alleged maximum statutory damages amount, and injunctive relief.

The overwhelming majority of Hailuo AI outputs user-created content have nothing to do with the Plaintiffs' characters. Having considered advice from our U.S. litigation advisor, our Directors believe that the probability of the Plaintiff prevail and fully succeed in their claims and we are found to have willfully infringed 500 registrations resulting in a maximum statutory damages award of US$75 million against us is extremely remote, for the following reasons:

• It is highly unlikely for the court to deem the number of infringement based on the approximately 500 registrations for motion pictures and television programs identified in the Complaint: U.S. law permits statutory damages on a per-"work" basis, not a per-registration basis, regardless of how many copies of that work are made. 17 U.S.C. § 504(c)(1). The Complaint alleges that plaintiffs used the Hailuo AI tool to generate videos featuring 68 characters for which they claim copyright ownership, suggesting that each character is treated as the relevant "work." While the Plaintiffs claim each of the 500 registrations they have identified is eligible to be counted as a "work", many of these 68 characters appear across multiple registrations within the more than 500 registrations attached to the Complaint. Only one registration per infringed "work" should give rise to statutory damages. Additional registrations for the same character — which merely reflect later appearances without new protectable expression — should not expand the statutory damage count. Accordingly, a court could reasonably determine that Plaintiffs may recover statutory damages only for the original registered depictions of each character, and that subsequent registrations do not independently support additional awards.

– 335 –

BUSINESS

Moreover, it would be possible that certain of the allegedly infringing videos contained in the Complaint were generated by the Plaintiffs themselves. For such video outputs, if the Plaintiff had not intentionally forced the tool to create those examples, those particular outputs would not exist. As a result, the number of works that can be attributed to the conduct of the Company or its users, as opposed to the Plaintiff's own test prompts, could be even lower.

•

It is highly unlikely that the Court would award the Plaintiffs the maximum statutory damages of US$150,000 per infringed work: as claimed: Statutory damages are typically capped at around US$30,000 per work for non-willful infringement, while the higher ceiling of roughly US$150,000 requires a finding of willfulness — a standard seldom met in comparable cases. A 2019 UCLA Law Review study examining approximately 1,000 federal copyright cases from 2005–2008 found that although about 80% of plaintiffs alleged willful infringement, courts found willfulness in only about 2% of plaintiff-favored cases, and maximum statutory damages of US$150,000 were awarded in just 0.2% of them. Those cases where the maximum statutory damage of US$150,000 were awarded usually involved extreme fact patterns, such as counterfeit software operations, repeated infringement after clear notice, deliberate evasion, or defaulting defendants, making them outliers even within the already small universe where a court finds willfulness.

We believe we have solid basis to argue that our conduct as alleged in the Complaint was not willful, considering that (1) it is the users, rather than the Company, who prompt the Hailuo AI model to produce outputs that allegedly infringed the Plaintiff's copyrights without any intentional assistance or facilitation by the Company; and (2) we did not have knowledge of the alleged infringement without receiving any advance notice or request from the Plaintiffs until we became aware of the Plaintiffs' ungrounded allegations that the model could be used to infringe the copyright of their alleged works. See "— Legal Analysis." Moreover, our good faith efforts in communication with the Plaintiffs, and the good faith mitigating measure we have taken since the onset of the litigation clearly distinguish us from the extreme cases where the maximum statutory damage of US$150,000 were awarded.

For the Plaintiffs' alleged actual damages and disgorgement of profits theory, having considered advice from our U.S. litigation advisor, our Directors are of the view that determining any potential recovery amount requires extensive factual development and likely expert analysis. Therefore, at this stage, it is not possible to provide an estimate of the alleged amount, and the Plaintiffs did not include a concrete figure in their Complaint. Under 17 U.S.C. § 504(b), a copyright owner may recover any profits of the infringer that are attributable to the infringement. Having considered advice from our U.S. litigation advisor, our Directors are of the view that applicable U.S. copyright laws indicate that subscription revenue is not profit — it must be reduced by infrastructure, personnel, development, marketing, and operational costs.

– 336 –

BUSINESS

根据联席保荐人所执行的独立尽职调查步骤，包括与本公司美国诉讼顾问进行访谈及讨论，联席保荐人有合理理由相信，董事会上述所表达的观点公正地代表了本公司美国诉讼顾问的意见。

董事及保荐人的观点

基于上述内容，董事认为上述索赔不会对本公司的业务、经营业绩或财务状况产生重大不利影响，原因如下：

• 最坏情形下法定赔偿金相较于流动资源的影响有限：在极为罕见的情况下，即原告胜诉并完全获得其索赔主张，且本公司被认定故意侵犯500项注册权利，从而被判处最高法定赔偿金7,500万美元，该7,500万美元的法定赔偿金总额仅相当于以下两项之和的约4.9%：(i) 本公司于2025年9月30日的可用财务资源，包括截至该日的现金及现金等价物以及金融资产流动部分；及(ii) 预期首次公开发售所得款项，考虑到本案预计将延续至上市日期之后。由于美国民事诉讼通常旷日持久，且本公司预计财务资源将持续增长，我们预期该百分比将随时间推移进一步下降。

• 法院颁布禁令救济的可能性较低，风险属于远期风险：在考虑了本公司美国诉讼顾问的意见后，本公司认为法院颁布导致本公司业务运营受到重大干扰的禁令的风险较低。根据美国法律，申请临时禁令要求原告证明若干因素，包括实质胜诉的高度可能性、无禁令情形下的不可弥补损害以及紧迫性，而原告在诉状中均未对上述任何因素加以证实。此外，可通过损害赔偿来弥补相关损失，这进一步降低了获得该等救济的可能性。永久禁令（如经考虑）仅在作出终局判决后方可颁布，且需要原告在责任认定方面胜诉，并克服包括合理使用在内的若干抗辩事由；本公司认为永久禁令的风险较低，原因在于本公司认为原告的索赔在所有重大方面均不具有实质依据，且现有证据不足以支持其主张。此外，即使在原告最终获得终局判决胜诉的不太可能情形下，法院在技术案件中通常拒绝颁布宽泛的禁令命令，倾向于采用针对特定作品的窄范围救济措施，而非导致全面停业的永久禁令。在考虑了本公司美国诉讼顾问的意见后，董事认为本公司将能够遵守任何此类禁令命令（如有），且禁令干扰本公司运营的整体概率被认为较低。即使法院命令颁布经专门调整的禁令救济，要求本公司对原告所主张的受版权保护角色实施增强的内容过滤，对运营的影响应属轻微。核心海螺AI服务

－337－

BUSINESS

在除被指控版权角色以外的所有使用场景中，该平台仍将保持完全正常运营，包括面向营销、教育、娱乐及使用非侵权内容进行创意表达的视频生成。

•

**原告知识产权对海螺AI运营及商业可行性影响有限：** 董事认为，如"——本集团采取的额外措施"所述，禁止在海螺AI运营中使用原告知识产权，不会对本集团的业务、经营业绩或增长前景产生重大影响。从产品设计角度而言，海螺AI是一个通用型多模态视觉生成平台，能够生成大量与原告知识产权无关的原创内容，其设计目的并非用于或依赖于侵权活动。该产品被创作者、广告商及普通用户广泛用于电影级视频生成、富有表现力的多媒体叙事及专业级项目，而非专门用于生成投诉书中所提及的角色或作品。从运营及财务指标角度而言，在业绩记录期间，海螺AI生成的绝大多数输出内容均为与原告知识产权无关的原创用户生成内容；就未来而言，董事不认为原告知识产权构成海螺AI用户参与度或收入产生的重要驱动因素。因此，董事认为，禁止在输出内容中使用原告知识产权，不会对海螺AI的平台使用情况、内容多样性、商业价值或商业可行性产生实质性改变，亦不会对本集团整体的业务、经营业绩及财务状况造成任何重大不利影响。

•

**原告寻求的其他救济可能微不足道。** 就原告在其投诉书中寻求的其他救济而言，在考虑了美国诉讼顾问的意见后，董事认为此类救济不太可能具有重大影响，具体包括：(a) 就以实际损害赔偿及归还归因于涉嫌侵权行为所得利润形式的金钱救济而言，版权所有人须在实际损害赔偿（包括侵权方的任何额外利润）与法定损害赔偿之间作出选择，两者为相互排斥的救济途径，不得就同一作品同时获得；因此，若原告就任何特定作品选择法定损害赔偿，则其将丧失就该作品主张实际损害赔偿及归还利润的权利；我们预期原告将选择法定损害赔偿，原因在于通过实际损害赔偿及归还利润所能追回的金额可能较低，致使寻求此类救济在经济上缺乏合理性；以及 (b) 就裁定原告合理律师费及诉讼费用而言，根据美国版权法，律师费的裁定属于法院自由裁量权范畴，而非自动授予，取决于法院对案件整体情况的评估，且任何费用裁定均须经合理性审查（包括在费率过高时予以潜在削减），同时费用承担具有对等性，即若我们成功抗辩相关索赔，我们亦有权寻求

— 338 —

BUSINESS

recovery of our own attorneys' fees under the same discretionary framework. Accordingly, while attorneys' fees are a component of risk, they do not fundamentally change the overall exposure profile described above.

As the case is still at an early stage and a reliable estimate of the amount of the obligation cannot be made with certainty, the Directors, having given due consideration to the legal advice and the relevant facts and circumstances, are of the opinion that the above matters give rise to contingencies for the Group and hence no provision should be recognized as at September 30, 2025. See Note 28 to the Accountant's report in Appendix I to the prospectus. The Reporting Accountants conducted their work in accordance with Hong Kong Standard on Investment Circular Reporting Engagements 200 Accountants' Reports on Historical Financial Information in Investment Circulars issued by the Hong Kong Institute of Certified Public Accountants. This standard requires that the Reporting Accountants plan and perform their work to obtain reasonable assurance about whether the Historical Financial Information as a whole is free from any material misstatement. The Reporting Accountant's opinion on the Historical Financial Information of the Group for the Track Record Period as a whole is set out on page I-1 to I-3 of Appendix I to the prospectus.

Based on the independent due diligence steps performed by the Joint Sponsors, including, among other things, (a) discussing with the Company's U.S. litigation advisor regarding the potential worst-case scenario, the possibility of granting the injunctive relief and their legal analysis in this regard, (b) reviewing of the Group's financial information, and (c) examination of the MAU data of Hailuo AI from June to November 2025 and the system-check results of Hailuo AI's outputs, the Joint Sponsors concur with the Directors' view above.

We intend to defend ourselves vigorously against the allegations and will respond to the complaint in accordance with U.S. civil procedure. However, as this case is still at an early stage, we cannot predict with certainty its timing, outcome, potential damages, or expenses that may be incurred, and there can be no assurance that we will prevail. Additionally, the potential damages scenario mentioned above is inherently speculative given the early stage of the case, the absence of discovery, and the unresolved questions regarding the Plaintiffs' claims, including the number of works that may ultimately be found to have been infringed, if any, and the appropriate per-work amount of statutory damages. Any adverse outcome of this case could result in payments of monetary damages and divert our management's attention from day-to-day operations, and thus have an adverse effect on our business, results of operations, financial condition and reputation. For the potential impact of legal proceedings on us, see "Risk Factors — Risks Related to Our Business and Industry — We, our directors, management, employees and shareholders and their affiliates may be subject to lawsuits, contract disputes, employment-related controversies, and other legal and administrative proceedings or fines, which could have a material adverse effect on our business, results of operations, financial condition and reputation."

— 339 —

BUSINESS

本集团采取的额外措施

尽管我们认为原告在诉讼中的索赔在所有重大方面均不具有实质依据，我们已积极主动地在持续合规及风险管理框架内实施相关措施。为防止用户输入任何不当或非法内容，并最大程度降低我们的风险敞口、避免卷入类似索赔及纠纷，我们在服务条款中明确告知用户，不得输入违法、不合规或不当内容，不得使用我们的产品从事违法、不合规活动或用于非法目的。我们的条款进一步规定了违规行为的后果，包括我们有权删除或屏蔽违禁内容、暂停用户账户或采取其他执法措施。

我们亦建立了投诉及举报机制。如服务条款所述，若用户发现违法、侵权或其他不合规的内容或行为，可提交投诉。收到此类举报后，我们将根据服务条款及时核实并处理相关问题，包括删除或屏蔽违法、侵犯他人知识产权或其他不合规内容，或在适当情况下采用关键词屏蔽措施。

我们在输入和输出两个阶段对违法或不合规内容进行监控和管理。我们依据适用法律、法规及运营经验制定了内容审核标准和管理规则（例如，《互联网信息服务管理办法》及《生成式人工智能服务管理暂行办法》中规定的禁止类别）。我们的自动化及人工审核机制可对有害内容进行过滤、屏蔽或以其他方式处理，包括违法内容、危害公共安全的内容，以及色情、暴力或其他违禁内容。

为支持内容治理框架的规模化实施，我们在所有产品的输入和输出两个阶段均采用自动化及人工审核与管理控制措施，并通过内部指标"过滤有效率"追踪上述控制措施的有效性。"过滤有效率"的定义为：（1）审核项目中成功过滤的总数量，除以（2）审核项目中应当过滤的总数量所得之比率。该指标主要通过基于抽样的方法进行衡量——我们部署包含应被过滤内容的测试数据集，并评估我们系统成功过滤的比例。我们认为，与仅依赖在真实用户流量中检测到的遗漏相比，此方法更为可靠，因为可能存在未被检测到的遗漏。作为参考，在截至2025年11月30日的当月，我们旗下AI原生产品（包括MiniMax、Hailuo AI、MiniMax Audio及Talkie/星野）的过滤有效率约为96.8%。

除上述关键绩效指标测试外，实际运营中潜在的"遗漏"可通过多种渠道识别，包括：（i）用户投诉；（ii）我们安全团队的例行检查及测试；以及（iii）在适用情况下来自监管机构的反馈。对于任何已识别的遗漏，我们将（a）迅速实施屏蔽措施以防止再次发生（包括

— 340 —

BUSINESS

taking down, blocking or restricting the relevant content/output, as appropriate), and (b) continuously improve our moderation controls based on root-cause analysis, including refining our review models and expanding our keyword libraries for filtering, with a view to further enhancing the effectiveness of filtering on an ongoing basis.

Our legal department manages infringement. In daily operations, when intellectual property disputes are identified, the responsible business department promptly reports them to the legal department. The responsible department and the legal department jointly investigate the matter, determine a response strategy, and take appropriate actions. In cases involving litigation or arbitration, the legal department also adheres to our internal regulations regarding litigation and arbitration cases. Additionally, the legal department conducts searches through public channels to identify potential infringement issues. For any confirmed infringement incidents, the legal department follows up and manages their resolution and keep the follow up records.

The Directors view these measures as adequate and effective and consistent with industry practice. Based on the Joint Sponsors' independent due diligence steps, including the discussions with CIC on whether the Group's business practices are comparable to those of its industry peers, the discussion with the internal control consultant and the review of the internal control report issued, nothing has come to the attention which would cause them to disagree with the Directors' view above.

General Legal Compliance

With respect to Singapore where we maintain a material subsidiary, as advised by our Singapore legal advisor, the Group's business operations have been conducted in compliance with all material aspects of applicable laws and regulations throughout the Track Record Period and up to the Latest Practicable Date.

While we do not possess any subsidiaries in the United States, given the volume of users of our AI-native products and the associated revenue generated from the U.S. market, we have engaged U.S. legal counsels to conduct legal due diligence, with a particular focus on U.S. sanctions and export control issues and data protection compliance issues.

Sanctions and Export Controls

Based on the relevant diligence findings, as advised by our international sanctions legal advisor, from a U.S. legal perspective, (a) our Group is not engaged in activities in violation of U.S. export controls and (b) our Group is not currently engaged in primary sanctioned activity or secondary sanctionable activity; and (c) our Group is not subject to U.S. tariff rules and regulations in any material aspects, since U.S. import tariffs only apply of export of physical goods to the United States and we do not export physical goods to the United States. As advised by our international sanctions legal advisor, during the Track Record Period and up to the Latest Practicable Date, our Group has not been subject to sanctions, and we have not engaged in any material activities in comprehensively sanction countries, or entered into

– 341 –

BUSINESS

material service contract or engaged in any material activities with any customers that are targets of U.S. sanctions. Therefore, as advised by our international sanctions legal advisor, we have been in compliance with rule and laws in US export control and sanctions in all material aspects, and U.S. sanctions are not likely to have any material adverse impact on us. Based on, among other things, the review of (a) the legal memorandum issued by the international sanctions legal advisor, (b) the Group's financial information during the Track Record Period and (c) the results of background check conducted against the Group, and third party due diligence sessions conducted by the Joint Sponsors, the Joint Sponsors have reasonable grounds to believe that the view expressed fairly represents the views of the sanction expert. Based on our Directors' knowledge, we have conducted our business operations in compliance with applicable laws and regulations in all material respects in all jurisdictions in which we operate.

Outbound Investment Rules

Effective on January 2, 2025, the final rule issued Treasury to implement the executive order of August 9, 2023 (the "Final Rule") imposes investment prohibition and notification requirements on U.S. Persons for a wide range of investments in entities associated with China (including Hong Kong and Macau) that are engaged in activities relating to three sectors: (i) semiconductors and microelectronics, (ii) quantum information technologies, and (iii) AI systems. U.S. persons subject to the Final Rule are prohibited from making, or required to report, certain investments in covered foreign persons, which are defined as "covered transactions," and include acquisitions of equity interests (including contingent equity interests), certain debt financing, joint ventures, and certain investments as a limited partner in a non-U.S. person pooled investment fund. The Final Rule excludes some investments from the scope of covered transactions, including certain ones in publicly traded securities. The Final Rule is aimed at exerting greater U.S. government oversight over U.S. direct and indirect investments involving China, and may introduce new hurdles and uncertainties for cross-border collaborations, investments, and funding opportunities of China-based issuers including us. Since our principal place of business is in China and we engage in the development of certain AI models, we are likely to be deemed as a "covered foreign person" as described in the Final Rule.

Pursuant to the Final Rule, U.S. persons' purchases of certain publicly traded securities are neither prohibited nor subject to notification to Treasury under an exception in the Final Rule that applies to U.S. persons' purchase of "any publicly traded security, with 'security' as defined in the U.S. Exchange Act, denominated in any currency, and that trades on a securities exchange in any jurisdiction" (the "Publicly Traded Securities Exception"), provided that such U.S. persons or their non-U.S. person subsidiaries are not afforded rights beyond standard minority shareholder protections with respect to the Company. But it appears likely, based on information we provided to our international sanctions advisor, that certain purchases of our Shares by U.S. persons or their non-U.S. person subsidiaries in the Global Offering would be ineligible for the Publicly Traded Securities Exception. Accordingly, it appears likely that some U.S. persons that purchase our Shares in the Global Offering or are the parents of

– 342 –

BUSINESS

non-U.S. person subsidiaries that purchase our Shares in the Global Offering would be required to file notifications regarding their or their subsidiaries' purchases with Treasury no later than 30 days after such purchases of the Shares.

As advised by our international sanctions legal advisor, our Directors are of the view that the Final Rule will not have a material effect on our business, results of operations or financial condition, in part because: (i) in light of the totality of the circumstances of the Global Offering, including that it is expected to be marketed to, and capable of being supported by, a broad investor base (including non-U.S. investors), the Final Rule is not expected to materially constrain investor participation in the Global Offering; (ii) based on information we provided to our international sanctions advisor, following completion of the Global Offering, it appears likely that the Publicly Traded Securities Exception would generally be available to U.S. persons (or their non-U.S. person subsidiaries) seeking to purchase our Class A Shares after they become listed and traded on the Stock Exchange, provided that such U.S. persons or non-U.S. person subsidiaries are not afforded rights beyond standard minority shareholder protections with respect to the Company; and (iii) although some U.S. person investors (or their non-U.S. person subsidiaries) may be unable to rely on the Publicly Traded Securities Exception in connection with the purchase of our Shares in the Global Offering, any resulting impact would be expected to relate primarily to the composition of our shareholder base, rather than our ability to continue operating our business in the ordinary course.

Compliance with Applicable PRC AI Laws and Regulations

The regulations concerning generative artificial intelligence (AI) services in the PRC mainly include the Interim Measures for the Administration of Generative Artificial Intelligent Services (《生成式人工智能服務管理暫行辦法》) (the "AIGC Administration Measures"), The Administrative Provisions on Algorithm Recommendation of Network Information Services (《互聯網信息服務算法推薦管理規定》), and the Administrative Provisions for Deep Synthesis as an Internet Information Service (《互聯網信息服務深度合成管理規定》). These regulations set forth specific compliance requirements regarding the record-filing of generative AI services, algorithm filing, security assessments, training data processing activities, data labeling, service transparency, generation content identification, and content compliance. See "Regulatory Overview — Laws and Regulations in the PRC — Government Policies on Artificial Intelligence" for details.

As of the Latest Practicable Date, we have completed the record-filing procedures for the large models and algorithms related to generative AI services in accordance with the aforementioned regulations. We have filed the requisite model registration for our proprietary models, including the "Abab" model series, the "Abab multi-modal" model series, and the "MiniMax" model series, with the Shanghai Municipal Cyberspace Administration in accordance with PRC regulations on generative artificial intelligence. As advised by our PRC data legal advisor, our Directors are of the view that the Group had complied in all material respects with the Measures and all applicable AI-related laws and regulations in the PRC during the Track Record Period and up to the Latest Practicable Date. Based on the view of our PRC legal advisor and the Joint Sponsors' discussion with their PRC legal advisor, the Joint

– 343 –

BUSINESS

Sponsors concur with the Directors' view above. We attach great importance to the protection of minors and have adopted a "Minor Protection Policy" to safeguard content safety for underage users. As providers of generative artificial intelligence services, We respect intellectual property rights and business ethics, keep trade secrets, and respect the legitimate rights and interests of others. We have service agreements with users of generative artificial intelligence services who have registered for our services to clarify the rights and obligations of both parties. We have also conducted the required security assessments for the relevant services as per the regulations. In addition, we have implemented compliance measures in areas such as training data processing, data labeling, service transparency, generation content identification, and content compliance, in line with the regulatory requirements set forth in the relevant Chinese regulations on generative AI services. On March 7, 2025, the CAC and three other departments jointly issued the Measures for the Identification of AI-Generated and Synthesized Content (《人工智能生成合成內容標識辦法》) (the "Identification Measures"), which came into effect on September 1, 2025. In accordance with the Identification Measures, we add explicit identification to AI-generated and synthesized content such as text, audio, images, video. We also add implicit identification to the file metadata containing AI-Generated and Synthesized content. As advised by our PRC legal advisor, we had fully complied with the Chinese government's policies, laws and regulations on artificial intelligence in all major aspects during the Track Record Period and up to the Latest Practicable Date. For more details, see the section headed "Data Security and Privacy".

Compliance with Applicable PRC Cybersecurity Laws and Regulations

The Measures for Cybersecurity Review (《網絡安全審查辦法》) prescribes the following conditions under which a cybersecurity review must be conducted. A company is required to undergo such a review if any of the following circumstances apply:

• Critical information infrastructures operators that purchase network products and services shall anticipate the potential national security risk of products and services after they enter operation, and they influence or could influence national security;

• Online platform operators engage in data processing activities that may have an impact on, or potentially affect, national security;

• Online platform operators holding the personal information of more than one million users and listing abroad; or

• Regulatory authorities have initiated a review based on their official prerogative.

See "Regulatory Overview — Laws and Regulations in the PRC — Regulations Relating to Cybersecurity and Data Protection" for details.

– 344 –

BUSINESS

During the Track Record Period and as the Latest Practicable Date, (i) we have not been notified by the relevant industry regulatory or supervisory authorities that we are classified as operators of critical information infrastructure operator; (ii) the data processed under our current business model is unlikely to be classified as important data or core data, and the risks of affecting or potentially affecting national security are considered minimal; (iii) although we hold personal information of more than 1 million users, our listing location is Hong Kong, which does not fall under the scope of "listing abroad"; and (iv) we have not received any notification from the relevant authorities regarding the initiation of a cybersecurity review based on their official authority.

Therefore, our PRC legal advisor specializing in cybersecurity and data privacy protection are of the opinion that the relevant laws and regulations under the Measures for Cybersecurity Review do not apply to us.

社会保险及住房公积金 (Social Insurance and Housing Provident Fund)

In accordance with the requirements of PRC laws and regulations, we participate in various employee social security schemes administered by local authorities for our employees in China, including pension insurance, medical insurance, maternity insurance, work-related injury insurance, unemployment insurance, and the housing provident fund. The requirements for and implementation of employee benefit schemes by local authorities in China are not uniform. Relevant government authorities may inspect employers to ensure full payment of the required employee benefit contributions. Employers failing to make such contributions in full as required may be subject to late payment surcharges, fines and/or other penalties.

As of the Latest Practicable Date, we are not aware of any plans or arrangements by the competent authorities to retroactively collect social insurance and housing provident fund contributions for the Track Record Period. Furthermore, our PRC subsidiaries have obtained certificates of compliance in respect of social insurance and housing provident fund contributions covering the Track Record Period and up to the Latest Practicable Date. As advised by our PRC legal advisor, based on the above circumstances and taking into account the current regulatory policies and the prevailing regulatory stance of the competent authorities in China, the risk of us being subject to material administrative penalties by the relevant competent human resources and social security authorities and housing provident fund authorities in China in relation to social insurance and housing provident fund contribution matters during the Track Record Period is remote.

– 345 –

BUSINESS LICENSES AND PERMITS The following table sets forth the details of the material licenses and permits necessary for the business operations in which we engaged in China.

| License/Permit | Entity Holding the License/Permit | Grant Date | Expiration Date | |---|---|---|---| | Generative Artificial Intelligence Service Filing (MiniMax) (生成式人工智能服務備案 (MiniMax)) | Shanghai Jizhi | June 9, 2025 | N/A | | Generative Artificial Intelligence Service Filing (abab Multi-modal) (生成式人工智能服務備案 (Abab Multi-modal)) | Shanghai MiniMax | June 12, 2024 | N/A | | Generative Artificial Intelligence Service Filing (Abab) (生成式人工智能服務備案) | Shanghai MiniMax | August 31, 2023 | N/A |

Note: (1) During the Track Record Period and up to the Latest Practicable Date, we had obtained all material licenses, permits, approvals and certificates necessary to conduct our actual business operations from the relevant government authorities in the PRC, and such licenses, permits, approvals and certificates remained in full effect.

RISK MANAGEMENT AND INTERNAL CONTROL We have established and currently maintain risk management and internal control systems consisting of policies and procedures that we consider appropriate for our business operations. These include multi-level budget planning and approval processes, internal audits, contract authorization protocols, and a structured system for financial and operational oversight. Our Board of Directors is responsible for the establishment and updating of our internal control systems, while our senior management monitors the daily implementation of internal control procedures and measures across our subsidiaries and functional departments. We are committed to continuously refining our internal control mechanisms to support business growth, financial transparency, and regulatory compliance.

Content Compliance Risk Management As a provider of AI-native products and foundation models, we place particular emphasis on content compliance, safety, and ethical responsibility. We uphold core values in our content management practices, including the protection of minors, the safeguarding of intellectual – 346 –

BUSINESS

property rights, and the respect for lawful rights and interests of third parties. To improve the legality, safety, accuracy, and reliability of content generated by our AI services, we have developed and implemented multi-level content governance policies.

Our review mechanisms include the operation of a 24/7 content monitoring system, coupled with continuous online inspection and refinement of both automated and human-in-the-loop review capabilities. These mechanisms are designed to identify and promptly respond to any content that breaches red lines, infringes legal rights, or contradicts our internal ethical standards. We have established a responsive complaint-handling mechanism to deal with user feedback and potential violations and actively implement mitigation measures where necessary. Our content compliance and review policies are updated continuously to reflect evolving regulatory requirements, user expectations, and technical advancements.

Information System Risk Management

We recognize data privacy and security as essential to our operations. We have adopted a comprehensive set of internal policies and technical safeguards to ensure the secure collection, processing, storage, and usage of data across our systems. These include internal controls over data access, end-to-end encryption protocols, and multi-layered authentication procedures.

We have established internal governance frameworks to address data security and protection of user privacy, which include teams responsible for ongoing monitoring, vulnerability detection, and incident response. Our internal security controls reflect national regulatory requirements, industry standards, and global best practices. In addition to internal efforts, we regularly engage external legal counsel and cybersecurity professionals to ensure our systems and policies remain compliant with applicable laws and internationally recognized data security frameworks.

To support data integrity and security, we continue to expand the scope and quality of our training data through structured procurement and controlled collection channels.

Legal and Regulatory Compliance Risk Management

We have adopted internal procedures to ensure our business operations are conducted in full compliance with applicable laws and regulations, including those governing AI, internet services, content platforms, and data security. Our legal department is responsible for reviewing commercial agreements, product deployment plans, and partnership arrangements to ensure consistency with relevant regulatory requirements. Before releasing new solutions or updates to the public, we conduct regulatory reviews to assess potential legal risks and obtain necessary licenses or approvals from competent authorities.

– 347 –

BUSINESS

We have also implemented a structured framework for intellectual property protection. This includes centralized management of our trademarks, patents, and copyrights, as well as regular audits to track renewal timelines and prevent unauthorized use. Our legal and product teams collaborate to conduct intellectual property clearance checks, and we engage third-party agencies where necessary to support registration and enforcement.

财务报告风险管理

我们已建立内部控制机制，以确保财务报告的准确性、及时性和可靠性。这些控制机制包括有关财务规划、预算审批、现金流管理以及法定账目和管理账目编制的正式政策。我们的财务部门负责监督会计政策的合规执行，并定期开展内部审查，以维护各报告系统的完整性。为提升一致性和质量，我们还定期为财务人员提供有关适用会计准则和报告规程的培训。

内部控制风险管理

我们建立了内部控制框架，支持业务、法律、合规、财务及运营团队之间的跨职能协作。各部门共同评估、实施并持续改进整个组织的控制措施。我们定期开展审查，以评估内部控制和风险管理实践的有效性。

我们的内部控制亦涵盖许可证及运营审批事项。我们制定了相应政策和流程，以确保在各司法管辖区开展业务活动所需的全部许可证和审批均已取得且持续有效。这些控制措施有助于确保符合当地监管要求，并降低运营及声誉风险的敞口。

人力资源风险管理

我们已制定并实施涵盖招聘、入职、培训、绩效评估以及遵守劳动法律法规和职业道德标准的内部政策。我们的行为准则就职业道德、诚信、保密及商业责任等方面设定了明确要求。我们在组织各层级倡导高度诚信的文化，并通过清晰的纪律处分程序强化问责机制。

为降低道德及法律风险，我们制定了反贿赂和反腐败政策，并辅以举报机制，允许员工和利益相关方以保密方式举报涉嫌违规行为。举报案件由我们的合规和法律部门及时、独立、公正地处理。我们还定期为员工及外部代理人提供合规和道德培训，以强化其对适用规则及内部标准的认识。

— 348 —

DIRECTORS AND SENIOR MANAGEMENT

been listed on MIT Technology Review's "35 Innovators Under 35 (China)" list in 2022 and Forbes' "30 Under 30 Asia" list in 2018.

Ms. Yun Yeyi (貟燁禕), aged 31, is our executive Director and chief operating officer. She is primarily responsible for overseeing the overall management and business operation, board affairs, formulating strategies and operation plans particularly on product and commercialization, making major business decisions of our Group.

Prior to joining the Company, Ms. Yun worked at SenseTime Group Inc. for approximately two years in roles related to product management and commercialization.

Ms. Yun obtained a bachelor's degree in computer science from Peking University (北京大學) in the PRC in June 2015 and a master's degree in computer science from Carnegie Mellon University in the United States in May 2017.

Mr. Zhao Pengyu (趙鵬宇), aged 29, is our executive Director and large language model research and engineering leader. He is primarily responsible for the research and development of large language models.

Prior to joining the Company, Mr. Zhao worked at various technology companies in roles related to large language model research and development.

Mr. Zhao obtained a bachelor's degree in computer science from Peking University (北京大學) in the PRC in June 2017 and a master's degree in computer science from Carnegie Mellon University in the United States in May 2019.

---

BUSINESS AWARDS AND RECOGNITIONS

| Award/Recognition | Award Authority | Award Year | |---|---|---| | "Specialized and New" SME (Shanghai Jizhi) ("專精特新"中小企業(上海稀宇極智科技有限公司)) | Municipal Commission of Economy and Informatization (上海市經濟和信息化委員會) | 2024 | | High and New Technology Enterprise (Shanghai Jizhi) (高新技術企業(上海稀宇極智科技有限公司)) | Shanghai Municipal Science and Technology Commission (上海市科學技術委員會) | 2024 | | High and New Technology Enterprise (Beijing Jizhi) (高新技術企業(北京稀宇極智科技有限公司)) | Beijing Municipal Science and Technology Commission (北京市科學技術委員會) | 2023 |

– 349 –

DIRECTORS AND SENIOR MANAGEMENT

BOARD OF DIRECTORS

Upon Listing, the Board will consist of nine Directors, including four executive Directors, two non-executive Directors and three independent non-executive Directors. The following table provides certain information about our Directors:

The following table sets forth the key information about our Directors:

| Name | Age | Position/Title | Time of Joining our Group | Time of Appointment as a Director | Responsibilities | |---|---|---|---|---|---| | **Executive Directors** | | | | | | | Dr. Yan Junjie (閆俊傑) | 36 | Founder, the chairman of the Board, executive Director, chief executive officer and chief technology officer | January 1, 2022 | October 26, 2023 | Oversee the overall management and business operation, board affairs, financial affairs, formulate strategies and operation plans particularly on AI research and development, make major business decisions of our Group | | Ms. Yun Yeyi (貟燁禕) | 31 | Executive Director and chief operating officer | March 31, 2022 | December 16, 2022 | Oversee the overall management and business operation, board affairs, formulate strategies and operation plans particularly on product and commercialization, make major business decisions of our Group | | Mr. Zhao Pengyu (趙鵬宇) | 29 | Executive Director and large language model research and engineering leader | August 1, 2023 | June 23, 2025 | Research and development of large language models | | Mr. Zhou Yucong (周彧聰) | 32 | Executive Director and visual models research and engineering leader | March 28, 2022 | June 23, 2025 | Research and development of visual models | | **Non-executive Directors** | | | | | | | Mr. Chen Yingjie (陳英傑) | 48 | Non-executive Director | March 21, 2024 | March 21, 2024 | Provide strategic advice on the development of the Company | | Mr. Liu Wei (劉偉) | 38 | Non-executive Director | April 28, 2023 | April 28, 2023 | Provide strategic advice on the development of the Company | | **Independent Non-executive Directors** | | | | | | | Mr. Huang Guobin (黃國濱) | 56 | Independent non-executive Director | Listing Date | December 23, 2025 (effective upon Listing) | Provide independent opinion and judgment to the Board | | Dr. Wang Pengcheng (王鵬程) | 54 | Independent non-executive Director | Listing Date | December 23, 2025 (effective upon Listing) | Provide independent opinion and judgment to the Board | | Dr. Zhu Huaxing (朱華星) | 41 | Independent non-executive Director | Listing Date | December 23, 2025 (effective upon Listing) | Provide independent opinion and judgment to the Board |

– 350 –

Executive Directors

Dr. Yan Junjie (閆俊傑), aged 36, is our founder, the chairman of our Board, executive Director, chief executive officer and chief technology officer. He is primarily responsible for overseeing the overall management and business operation, board affairs, financial affairs, formulating strategies and operation plans particularly on AI research and development, making major business decisions of our Group.

Prior to founding the Company, Dr. Yan served at SenseTime Group Inc., a company listed on the Stock Exchange (stock code: 0020), for more than six years with positions such as its vice president, and vice-head of its research institute.

Dr. Yan obtained a bachelor of mathematics in Southeast University (東南大學) in the PRC in June 2010. He then obtained a doctorate degree in the area of artificial intelligence in the Institute of Automation, Chinese Academy of Sciences (中國科學院自動化研究所) in July 2015 and conducted post-doctorate research at Tsinghua University (清華大學). Dr. Yan has

– 351 –

DIRECTORS AND SENIOR MANAGEMENT

published approximately 200 academic articles on top conferences and journals with over 30,000 citations and won several awards and honors. Dr. Yan was awarded (i) the First Prize in the Guangdong Province Technology Invention Award (廣東省技術發明獎) in February 2020; (ii) the Wu Wenjun Artificial Intelligence Natural Science Award (吳文俊人工智能自然科學獎) in October 2019; (iii) Wu Wenjun Artificial Intelligence Technology Progress Award (吳文俊人工智能科技進步獎) in October 2019; (iv) 2024 Shanghai Oriental Talents Program (上海市東方英才計劃) in December 2024; and (v) Senior Professional Title (正高級職稱), a professional title generally denoting individuals who have outstanding achievements in their respective technical or professional fields, by the Shanghai Municipal Professional Title Evaluation Committee in February 2025.

Ms. Yun Yeyi (貟燁禕), aged 31, is our executive Director and chief operating officer. Ms. Yun is primarily responsible for overseeing the overall management and business operation, board affairs, formulating strategies and operation plans particularly on product and commercialization, and making major business decisions of our Group.

Prior to joining our Company, she served in various positions at SenseTime Group Inc. She was the manager of fundraising and strategic investment of SenseTime Group Inc. from September 2017 to August 2018. She was then promoted to CEO executive assistant and head of strategy from August 2018 to January 2021. Ms. Yun then served as its director of innovative business from January 2021 to January 2022.

Ms. Yun obtained a bachelor of science degree in electrical engineering with additional majors in economics and mathematics from The Johns Hopkins University in the United States in 2017.

Mr. Zhao Pengyu (趙鵬宇), aged 29, is our executive Director appointed in June 2025 and the large language model research and engineering leader. Mr. Zhao joined our Company as a natural language processing researcher and engineer since August 2023. He is primarily responsible for the research and development of large language models.

Prior to joining our Company, Mr. Zhao served as a research software development engineer at Beijing Hulu Technology Co., Ltd. (北京葫蘆科技有限公司), a company mainly engaged in the research and development of streaming media technology, between August 2020 and July 2023, where he was primarily responsible for recommendation algorithms.

Mr. Zhao obtained his bachelor's degree and master's degree in computer science and technology from Peking University (北京大學) in the PRC in July 2017 and July 2020, respectively.

Mr. Zhou Yucong (周彧聰), aged 32, is our executive Director appointed in June 2025 and the visual model research and engineering leader. Mr. Zhou joined our Company as a visual model researcher and engineer since March 2022. He is primarily responsible for the research and development of visual models.

– 352 –

DIRECTORS AND SENIOR MANAGEMENT

Prior to joining our Company, Mr. Zhou worked at SenseTime Group Inc. from April 2018 to July 2019 and Huawei Technology Company Limited from August 2019 to March 2022, where he focused on algorithms. Mr. Zhou has also been the legal representative and director of Shanghai MiniMax since January 2023.

Mr. Zhou obtained his bachelor's degree in mathematics and applied mathematics and his master's degree in computer science from Beihang University (北京航空航天大學) in July 2015 and March 2018, respectively.

Non-executive Directors

Mr. Chen Yingjie (陳英傑), aged 48, is our non-executive Director and is responsible for providing strategic advice on the development of the Company. Mr. Chen has been appointed as our Director since March 2024.

Mr. Chen joined Alibaba Group Holding Limited (a company listed on the Hong Kong Stock Exchange (stock code: 9988) and the New York Stock Exchange (symbol: BABA)) ("Alibaba Group") in December 2012. Mr. Chen currently serves as the managing director of the strategic investment department of Alibaba Group. Previously, Mr. Chen served as a non-independent director of DBAPP Security Co., Ltd. (a company listed on the Shanghai Stock Exchange (stock code: 688023)), from May 2020 to January 2024, and as a non-executive director of XPeng Inc., a company listed on the Stock Exchange (stock code: 9868) and the New York Stock Exchange (symbol: XPEV), from February 2022 to November 2023. Prior to joining Alibaba Group, Mr. Chen was a senior manager in the corporate finance department of PricewaterhouseCoopers from September 2007 to December 2012, and a vice president of investment of Shandong Datong Hongye Group from June 2004 to August 2007. Mr. Chen was an auditor at Arthur Andersen from September 1999 to May 2004. He has been qualified as a certified public accountant in Canada since July 1999.

Mr. Chen obtained a bachelor's degree in accounting from Shanghai University of Finance and Economics (上海財經大學) in the PRC in July 1999.

Mr. Liu Wei (劉偉), aged 38, is our non-executive Director and is responsible for providing strategic advice on the development of the Company. Mr. Liu has been appointed as our Director since April 2023.

Mr. Liu co-founded Shanghai Mihoyo Network Technology Co., Ltd. (上海米哈游網絡科技股份有限公司) in February 2012 and currently serves as its president. Mr. Liu has made significant contributions to miHoYo's management, growth, and global expansion. miHoYo is a prominent game developer that has established itself as one of the global leaders in international expansion through the worldwide success of games such as Genshin Impact (原神), the Honkai series (崩壞), Tears of Themis (未定事件簿) and Zenless Zone Zero (絕區零).

– 353 –

DIRECTORS AND SENIOR MANAGEMENT

Mr. Liu obtained a bachelor's degree in information engineering and a master's degree in communications and information systems from Shanghai Jiao Tong University (上海交通大學) in the PRC in July 2009 and March 2012, respectively.

Independent Non-executive Directors

Mr. Huang Guobin (黃國濱), aged 56, was appointed as an independent non-executive Director with effect from Listing Date. He is responsible for providing independent opinion and judgment to the Board.

Mr. Huang has been the chairman of the board of directors of PEC International Group Limited (智贏國際(集團)有限公司) since February 2024. Since December 2023, he founded NEXX Global, a company focusing on logistics technology, and has been its chairman of the board. Previously, he was a senior advisor at J.P. Morgan (Asia Pacific) Limited (摩根大通證券(亞太)有限公司) from January 2023 to December 2023. From February 2021 to December 2022, Mr. Huang was the legal representative, chief executive officer and head of investment banking at J.P. Morgan Securities (China) Co., Ltd. (摩根大通證券(中國)有限公司). Mr. Huang was chief executive officer of global investment banking for China at J.P. Morgan from December 2015 to January 2021. From September 2011 to December 2015, he was head of the China Industrials Group for Goldman Sachs. Prior to that, he was managing director of China International Capital Corporation from October 1999 to June 2011.

Mr. Huang has been an independent non-executive director of Zoomlion Heavy Industry Science and Technology Co., Ltd. (中聯重科股份有限公司) (a company listed on the Stock Exchange (stock code: 1157) and the Shenzhen Stock Exchange (stock code: 000157)) since June 2023, director of UCloud Technology Co., Ltd. (優刻得科技股份有限公司) (a company listed on the Shanghai Stock Exchange (stock code: 688158)) since September 2024 and independent non-executive director of RemeGen Co., Ltd. (榮昌生物製藥(煙台)股份有限公司) (a company listed on the Stock Exchange (stock code: 9995) and the Shanghai Stock Exchange (stock code: 688331)) since January 2025.

Mr. Huang obtained a bachelor's degree in inorganic non-metallic materials science from Tongji University (同濟大學) in the PRC in September 1991 and master's degree in business administration from Lancaster University Management School in the United Kingdom in December 1997. Mr. Huang was awarded the Overseas Financial Top Talent by the Shanghai Municipal Government in November 2021. Mr. Huang has also obtained a Hong Kong corporate finance qualification from the SFC in 1999, securities qualification certificate from the Securities Association of China in August 2012, securities industry executive qualifications from the CSRC in 2021 and independent director qualification from the Shenzhen Stock Exchange in May 2023. Mr. Huang is the board member of Tongji University.

Dr. Wang Pengcheng (王鵬程), aged 54, was appointed as an independent non-executive Director with effect from Listing Date. He is responsible for providing independent opinion and judgment to the Board.

– 354 –

DIRECTORS AND SENIOR MANAGEMENT

Dr. Wang has been a professor at Beijing Technology and Business University (北京工商大學) since July 2022. Prior to that, he was the managing partner of assurance services of Greater China of Ernst & Young from December 2014 to June 2022. From June 2005 to May 2014, Dr. Wang served as managing partner of global financial services industry in Greater China of Deloitte Touche Tohmatsu. Prior to that, he was partner of Pan-China Certified Public Accountants from September 2000 to May 2005. From April 1994 to September 2000, he served as associate professor and assistant director of the school of accountancy of the Central University of Finance and Economics (中央財經大學). Dr. Wang became a certified public accountant in the PRC in August 2000.

Dr. Wang has been an independent non-executive director of The People's Insurance Company (Group) of China Limited, a company listed on the Stock Exchange (stock code: 1339) and the Shanghai Stock Exchange (stock code: 601319), since August 2023 and Sinopec Oilfield Service Corporation, a company listed on the Stock Exchange (stock code: 1033) and the Shanghai Stock Exchange (stock code: 600871), since June 2024.

Dr. Wang obtained a bachelor's degree in accounting from the Anshan Institute of Iron and Steel (鞍山鋼鐵學院) (currently known as the University of Science and Technology Liaoning (遼寧科技大學)) in the PRC in July 1991, a master's degree in accounting from the Dongbei University of Finance and Economics (東北財經大學) in the PRC in April 1994 and a doctorate degree in accounting from the Chinese Academy of Fiscal Sciences (中國財政科學研究院) in the PRC in March 2000.

Dr. Zhu Huaxing (朱華星), aged 41, was appointed as an independent non-executive Director with effect from Listing Date. He is responsible for providing independent opinion and judgment to the Board.

Dr. Zhu has contributed extensively to theoretical physics, with an emphasis on their practical applications. Dr. Zhu has worked at Peking University (北京大學) since July 2023, where he first acted as a tenured associate professor and later a Boya distinguished professor. Prior to that, he was a researcher at the Hundred Talents Program of Zhejiang University (浙江大學) from April 2017 to June 2023. From September 2015 to April 2017, Dr. Zhu was a postdoctoral fellow at the Massachusetts Institute of Technology. Prior to that, he was a research assistant at the SLAC National Accelerator Laboratory from October 2012 to August 2015.

Dr. Zhu obtained a bachelor's degree in physics and a doctorate degree in physics from Peking University (北京大學) in the PRC in July 2007 and July 2012, respectively. Dr. Zhu was awarded the Qiushi Young Scholar Award by the Hong Kong Qiushi Foundation in 2020, the Asian Young Scientist Project in 2023 and the National Outstanding Youth Fund Project in 2024.

– 355 –

DIRECTORS AND SENIOR MANAGEMENT

SENIOR MANAGEMENT

The following table sets forth the key information about our senior management.

| Name | Age |

Position/Title | Time of Joining our Group | Time of Appointment as our Senior Management | Responsibilities

Dr. Yan Junjie (閆俊傑), aged 36 | January 2022 | October 2023 | Chairman of the Board, executive Director, chief executive officer and chief technology officer — Oversee the overall management and business operation, board affairs, financial affairs, formulate strategies and operation plans particularly on AI research and development, make major business decisions of our Group

Ms. Yun Yeyi (貟燁禕), aged 31 | March 2022 | December 2022 | Executive Director and chief operating officer — Oversee the overall management and business operation, board affairs, formulate strategies and operation plans particularly on product and commercialization, make major business decisions of our Group

Mr. Zhao Pengyu (趙鵬宇), aged 29 | August 2023 | June 2025 | Executive Director and large language model research and engineering leader — Research and development of large language models

Mr. Zhou Yucong (周彧聰), aged 32 | March 2022 | June 2025 | Executive Director and visual model research and engineering leader — Research and development of visual models

For the biographical details of Dr. Yan, Ms. Yun, Mr. Zhao Pengyu and Mr. Zhou Yucong, see "— Board of Directors — Executive Directors"

– 356 –

DIRECTORS AND SENIOR MANAGEMENT

GENERAL

Save as disclosed above, none of the Directors or members of senior management of our Company has been a director of any public company the securities of which are listed on any securities market in Hong Kong or overseas in the three years immediately preceding the date of this Prospectus.

None of the Directors or members of the senior management of our Company is related to any other Directors and members of the senior management of our Company.

Save as disclosed above, to the best knowledge, information and belief of our Directors having made all reasonable inquiries, there was no other matter with respect to the appointment of our Directors that needs to be brought to the attention of the Shareholders and there was no information relating to our Directors that is required to be disclosed pursuant to Rule 13.51(2)(h) to (v) of the Listing Rules as of the Latest Practicable Date.

CONFIRMATION FROM OUR DIRECTORS

Rule 8.10 of the Listing Rules

Each of our Directors confirms that as of the Latest Practicable Date, he or she did not have any interest in a business which competes or is likely to compete, either directly or indirectly, with our Company's business which would require disclosure under Rule 8.10 of the Listing Rules.

Rule 3.09D of the Listing Rules

Each of our Directors confirms that he or she (i) has obtained the legal advice referred to under Rule 3.09D of the Listing Rules in June 2025, and (ii) understands his or her obligations as a director of a listed issuer under the Listing Rules.

Rule 3.13 of the Listing Rules

Each of the independent non-executive Directors has confirmed (i) his/her independence as regards each of the factors referred to in Rules 3.13(1) to (8) of the Listing Rules, (ii) he/she has no past or present financial or other interest in the business of the Company or its subsidiaries or any connection with any core connected person of the Company under the Listing Rules as of the Latest Practicable Date, and (iii) that there are no other factors that may affect his/her independence at the time of his/her appointments.

– 357 –

DIRECTORS AND SENIOR MANAGEMENT

JOINT COMPANY SECRETARIES

Mr. Xue Zizhao (薛子釗), aged 33, is our vice president of capital markets and investments since September 2023 and a joint company secretary. He is primarily responsible for capital market operations and financial affairs.

Prior to joining our Company, Mr. Xue served as an investment professional in Hillhouse Investment from September 2019 to September 2023 where he was responsible for assessing investment opportunities in the primary market. Mr. Xue worked as an investment professional in Silver Lake Capital from August 2017 to August 2019 where he was responsible for assessing investment opportunities in the primary market.

Mr. Xue received his bachelor's degrees in micro-electronics and in economics and his Master's degree in finance from Peking University (北京大學) in the PRC in July 2015 and in June 2017, respectively. Mr. Xue has been a Chartered Financial Analyst since June 2025.

Ms. Chan Sau Ling (陳秀玲) was appointed as a joint company secretary in June 2025. Ms. Chan is a director of company secretarial services of Tricor Services Limited and she has over 25 years of experience in the corporate secretarial field. Ms. Chan has been providing professional corporate services to and acting as the company secretary or joint company secretary of several companies listed on the Stock Exchange.

Ms. Chan is a Chartered Secretary, a Chartered Governance Professional and a fellow of both The Hong Kong Chartered Governance Institute and The Chartered Governance Institute in the United Kingdom.

BOARD COMMITTEES

Our Board delegates certain responsibilities to various committees. In accordance with the relevant laws and regulations and the Corporate Governance Code, our Company has formed four Board committees, namely the Audit Committee, the Nomination Committee, the Remuneration Committee and the Corporate Governance Committee.

Audit Committee

We have established an Audit Committee with written terms of reference in compliance with Rule 3.21 of the Listing Rules and paragraph D.3 of the Corporate Governance Code. The Audit Committee consists of three Directors, namely Dr. Wang Pengcheng, Mr. Huang Guobin and Mr. Liu Wei. Dr. Wang Pengcheng has the appropriate professional qualifications or accounting or related financial management expertise as required under Rules 3.10(2) and 3.21 of the Listing Rules. Dr. Wang Pengcheng serves as the chairman of the Audit Committee. The primary duties of the Audit Committee include, but not limited to, the following:

•

proposing the appointment or change of external auditors to our Board, and monitoring the independence of external auditors and evaluating their performance; – 358 –

DIRECTORS AND SENIOR MANAGEMENT •

examining the financial information of our Company and reviewing financial reports and statements of our Company;

•

examining the financial reporting system, the risk management and internal control system of our Company, overseeing their rationality, efficiency and implementation and making recommendations to our Board; and

•

dealing with other matters that are authorized by our Board.

Remuneration Committee We have established a Remuneration Committee with written terms of reference in compliance with paragraph E.1 of the Corporate Governance Code. The Remuneration Committee consists of three Directors, namely Mr. Huang Guobin, Dr. Yan Junjie and Dr. Wang Pengcheng. Mr. Huang Guobin serves as the chairman of the Remuneration Committee. The primary duties of the Remuneration Committee include, but not limited to, the following: •

making recommendations to the Board on the Company's policy and structure for all Directors' and senior managements' remuneration and on the establishment of a formal and transparent procedure for developing remuneration policy;

•

monitoring the implementation of remuneration system of our Company;

•

making recommendations on the remuneration packages of our Directors and senior management; and

•

dealing with other matters that are authorized by our Board.

Nomination Committee We have established a Nomination Committee with written terms of reference in compliance with paragraph B.3 of the Corporate Governance Code. The Nomination Committee consists of three Directors, namely Mr. Huang Guobin, Ms. Yun Yeyi and Dr. Zhu Huaxing. Mr. Huang Guobin serves as the chairman of the Nomination Committee. The primary duties of the Nomination Committee include, but not limited to, the following: •

conducting extensive search and providing to our Board suitable candidates for our Directors, chief executive officer and other members of the senior management;

•

reviewing the structure, size and composition (including the skills, knowledge and experience) of our Board at least annually, assisting our Board in maintaining a board skills matrix and making recommendations on any proposed changes to our Board to complement the Company's corporate strategy;

– 359 –

DIRECTORS AND SENIOR MANAGEMENT •

researching and developing standards and procedures for the election of our Board members, chief executive officer and members of the senior management, and making recommendations to our Board;

•

assess the independence of independent non-executive directors;

•

supporting our Company's regular evaluation of our Board's performance; and

•

dealing with other matters that are authorized by our Board.

Corporate Governance Committee We have established a Corporate Governance Committee in compliance with Chapter 8A of the Listing Rules. The Corporate Governance Committee comprises three independent non-executive Directors, namely Dr. Zhu Huaxing, Dr. Wang Pengcheng and Mr. Huang Guobin. Dr. Zhu Huaxing is the chairman of the Corporate Governance Committee. The primary duties of the corporate governance committee are, among other things, to ensure that the Company is operated and managed for the benefit of all Shareholders and to ensure the Company's compliance with the Listing Rules and safeguards relating to the weighted voting right structures of the Company. For details of their experience in corporate governance related matters, see the biographies of the independent non-executive Directors in the section headed "— Independent Non-executive Directors" above. In accordance with Rule 8A.30 of the Listing Rules and the Corporate Governance Code, the work of our corporate governance committee as set out in its terms of reference includes: (a)

to develop and review the Company's policies and practices on corporate governance and make recommendations to the Board;

(b)

to review and monitor the training and continuous professional development of Directors and senior management;

(c)

to review and monitor the Company's policies and practices on compliance with legal and regulatory requirements;

(d)

to develop, review and monitor the code of conduct and compliance manual (if any) applicable to employees and directors;

(e)

to review the Company's compliance with the Corporate Governance Code and disclosure in the Corporate Governance Report;

(f)

to review and monitor whether the Company is operated and managed for the benefit of all its Shareholders;

– 360 –

DIRECTORS AND SENIOR MANAGEMENT (g)

to confirm, on an annual basis, that the beneficiaries of weighted voting rights have been members of the Board throughout the year and that no matters under Rule 8A.17 of the Listing Rules have occurred during the relevant financial year;

(h)

to confirm, on an annual basis, whether or not the beneficiaries of weighted voting rights have complied with Rules 8A.14, 8A.15, 8A.18 and 8A.24 of the Listing Rules throughout the year;

(i)

to review and monitor the management of conflicts of interests and make a recommendation to the Board on any matter where there is a potential conflict of interest between the Company, a subsidiary of the Company and/or Shareholders of the Company (considered as a group) on one hand and any beneficiary of weighted voting rights on the other;

(j)

to review and monitor all risks related to the Company's weighted voting rights structure, including connected transactions between the Company and/or a subsidiary of the Company on one hand and any beneficiary of weighted voting rights on the other and make a recommendation to the Board on any such transaction;

(k)

to make a recommendation to the Board as to the appointment or removal of the Compliance Adviser;

(l)

DIRECTORS AND SENIOR MANAGEMENT

ROLE OF OUR INDEPENDENT NON-EXECUTIVE DIRECTORS

根据上市规则第8A.26条，具有加权投票权架构的上市公司之独立非执行董事的职责须包括（但不限于）上市规则附录C1所载《企业管治守则》第2部分守则条文C.1.2、C.1.6及C.1.7所述的职能。独立非执行董事的职能包括：

• 参与董事会会议，就策略、政策、表现、问责、资源、主要任命及行为准则等事宜作出独立判断；

• 在出现潜在利益冲突时带头处理；

• 如获邀请，出任审计委员会、薪酬委员会、提名委员会及企业管治委员会成员；

• 审查本公司在实现议定企业目标方面的表现，并监察表现报告；

• 透过定期出席及积极参与，向董事会及其所服务的任何委员会提供其技能、专业知识及不同背景与资历方面的裨益；

• 透过独立、建设性及有据可查的意见，对本公司战略及政策的制定作出积极贡献；及

• 出席股东大会，并全面了解本公司股东的意见。

REMUNERATION OF DIRECTORS AND SENIOR MANAGEMENT

本公司董事以费用、薪金、津贴、酌情花红、股份酬劳、退休福利计划供款及其他实物福利的形式收取酬劳。

截至2022年、2023年及2024年12月31日止年度以及截至2025年9月30日止九个月，向本公司董事支付或应付的酬劳总额分别为1,460,000美元、2,170,000美元、3,630,000美元及2,280,000美元。

根据现行薪酬安排，我们估计截至2025年12月31日止年度应计予本公司董事的税前薪酬总额约为6,030,000美元。

– 362 –

DIRECTORS AND SENIOR MANAGEMENT

截至2022年、2023年及2024年12月31日止年度以及截至2025年9月30日止九个月，五名最高薪酬人士中其余人士的酬劳总额分别为540,000美元、1,200,000美元、1,220,000美元及4,700,000美元。

于记录期间，本公司并无向本公司董事或五名最高薪酬人士支付或其应收取任何酬劳，作为邀请其加入或于加入本公司时的诱因，或作为其因离任而与本公司或本公司任何附属公司的管理职位相关的补偿。

于记录期间，本公司董事概无放弃任何酬劳。除上文所披露者外，于记录期间，本公司或本公司任何附属公司并无向本公司董事或五名最高薪酬人士支付或应付任何其他款项。

CORPORATE GOVERNANCE

根据上市规则附录C1所载《企业管治守则》第2部分守则条文C.2.1，在联交所上市的公司须遵守（但可选择偏离）主席与行政总裁的职责须分开并不得由同一人士担任的规定。本公司并无分设主席及行政总裁，上述两个职位现由严博士出任。董事会认为，由同一人士担任主席及行政总裁职务，有助确保本集团领导层的一贯性，并能更有效地为本集团进行整体战略规划。董事会认为，现有安排的权力及权威制衡不会受损，且该架构将使本公司能够迅速有效地作出及执行决策。

BOARD DIVERSITY POLICY

为提升本公司董事会的效能及维持高标准的企业管治，我们已采纳董事会多元化政策，该政策订明实现及维持本公司董事会多元化的目标及方法。根据董事会多元化政策，我们在甄选候选人加入本公司董事会时，会考虑多项因素以实现董事会多元化，包括但不限于性别、技能、年龄、专业经验、知识、文化及教育背景以及任职年期。最终委任决定将以候选人的才干及其对本公司董事会的贡献为基础。

本公司董事具备均衡的知识和技能组合，包括整体管理及战略发展、会计及企业管治，以及行业经验。本公司设有三名具备不同行业背景的独立非执行董事，占本公司董事会成员总数的三分之一。

– 363 –

DIRECTORS AND SENIOR MANAGEMENT

evaluated the structure, size and composition of our Board, and is of the opinion that the structure of our Board is reasonable, and the experience and skills of the Directors in various aspects and fields can enable our Company to maintain a high standard of operations.

Besides, we particularly recognize the importance of gender diversity. Our Board currently consists one female Directors and eight male Directors. We have taken, and will continue to take, steps to promote gender diversity at all levels of our Company, including but without limitation to our Board and senior management levels. Going forward, we will continue to work to enhance gender diversity of our Board when selecting and recommending suitable candidates for Board appointments and will maintain at least one female Director on the Board. Our Company also intends to promote gender diversity at the mid to senior level so that our Company can maintain a balanced gender ratio at different levels. Taking into account our existing business model and specific needs as well as the different background of our Directors, the composition of our Board satisfies our board diversity policy.

Our Nomination Committee is responsible for ensuring the diversity of our Board members. After the Listing, our Nomination Committee will examine the board diversity policy from time to time to ensure its continued effectiveness and we will disclose in our corporate governance report about the implementation of the board diversity policy on an annual basis.

COMPLIANCE ADVISER

We have appointed Somerley Capital Limited as our Compliance Adviser pursuant to Rule 3A.19 of the Listing Rules. Our Compliance Adviser will provide us with guidance and advice as to compliance with the Listing Rules and applicable Hong Kong laws.

Pursuant to Rule 3A.23 and 8A.34 of the Listing Rules, our Compliance Adviser will advise our Company, among others, in the following circumstances:

(a) before the publication of any regulatory announcement, circular or financial report;

(b) where a transaction, which might be a notifiable or connected transaction, is contemplated, including share issues, sales or transfers of treasury shares and share repurchases;

(c) where we propose to use the proceeds from the Global Offering in a manner different from that detailed in this Prospectus or where our business activities, developments or results deviate from any forecast, estimate or other information in this Prospectus;

(d) where the Stock Exchange makes an inquiry to our Company regarding unusual movements in the price or trading volume of its listed securities or any other matters in accordance with Rule 13.10 of the Listing Rules;

(e) the WVR Structure;

– 364 –

DIRECTORS AND SENIOR MANAGEMENT

(f) transactions in which any beneficiary of weighted voting rights in the Company has an interest; and

(g) where there is a potential conflict of interest between the Company, its subsidiary and/or Shareholders (considered as a group) on one hand and any beneficiary of weighted voting rights in the Company on the other.

The term of appointment of the Compliance Adviser shall commence on the Listing Date. Pursuant to Rule 8A.33 of the Listing Rules, the Company is required to engage a compliance adviser on a permanent basis.

– 365 –

RELATIONSHIP WITH OUR CONTROLLING SHAREHOLDERS

OUR CONTROLLING SHAREHOLDERS

Immediately following the completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised), an aggregate of 5,000,000 Class A Ordinary Shares and 74,102,534 Class B Ordinary Shares, representing approximately (i) 72.05% of the voting rights in our issued share capital in general meetings (except for resolutions with respect to the Reserved Matters), and (ii) 25.90% of the voting rights in our issued share capital in general meetings for resolutions with respect to the Reserved Matters, will be held by MiniMax Awakening, MiniMax Limited and Alpha EXP as well as MiniMax Matrix. MiniMax Awakening and MiniMax Limited are wholly owned by Dr. Yan through Local Linearity. MiniMax Matrix is also a controlled entity of Dr. Yan through Local Linearity. Alpha EXP is held by Scaling EXP Limited as to 99% and Local Linearity as to 1%. Scaling EXP Limited is wholly-owned by Trident Trust Company (Hong Kong) Limited, which acts as the trustee of Alpha EXP Trust. Alpha EXP Trust is a trust established by Dr. Yan (as settlor) for the benefit of himself. Accordingly, Dr. Yan, Local Linearity Inc., MiniMax Awakening, MiniMax Limited, Alpha EXP, Scaling EXP Limited and MiniMax Matrix together will constitute as a group of Controlling Shareholders of our Company after the Listing.

RULE 8.10 OF THE LISTING RULES

Each of our Controlling Shareholders has confirmed that he or she or it or his or her or its respective close associates do not have any interest in a business, apart from the business of our Group, which competes or is likely to compete, directly or indirectly, with our business, and which would require disclosure under Rule 8.10 of the Listing Rules.

INDEPENDENCE OF OUR BUSINESS

Having considered the following factors, our Directors are satisfied that we are able to carry out our business independently from our Controlling Shareholders and their respective close associates upon and after the Listing.

Management Independence

We are able to carry on our business independently from our Controlling Shareholders from a management perspective. Our Board consists of nine Directors, including four executive Directors, two non-executive Directors and three independent non-executive Directors.

(a) each Director is aware of his/her fiduciary duties as a director which require, among other things, that he/she acts for the benefit and in the interest of our Company and does not allow any conflict between his/her duties as a Director and his/her personal interests;

– 366 –

RELATIONSHIP WITH OUR CONTROLLING SHAREHOLDERS

(b) our daily management and operations are carried out by a senior management team, all of whom have substantial experience in the industry in which our Company is engaged, and will therefore be able to make business decisions that are in the best interests of our Group. For details of the industry experience of our senior management team, see "Directors and Senior Management";

(c) we have three independent non-executive Directors and certain matters of our Company must always be referred to the independent non-executive Directors for review;

(d) in the event that there is a potential conflict of interest arising out of any transaction to be entered into between our Group and a Director and/or his/her associate, he/she shall abstain from voting and shall not be counted towards the quorum for the voting; and

(e) we have adopted a series of corporate governance measures to manage conflicts of interest, if any, between our Group and our Controlling Shareholders which would support our independent management. For details, see "— Corporate Governance."

Based on the above, our Directors believe that our Board as a whole and together with our senior management are able to perform the managerial role in our Group independently from our Controlling Shareholders and their close associates after the Listing.

Operational Independence

We do not rely on our Controlling Shareholders and their close associates for our business development, staffing, logistics, administration, finance, internal audit, information technology, sales and marketing, or company secretarial functions. We have our own departments specializing in these respective areas which have been in operation and are expected to continue to operate separately and independently from our Controlling Shareholders and their close associates. In addition, we have our own headcount of employees for our operations and management for human resources.

We have independent access to suppliers and customers and an independent management team to handle our day-to-day operations. We are also in possession of all relevant licenses, certificates, facilities and intellectual property rights necessary to carry on and operate our principal businesses and we have sufficient operational capacity in terms of capital and employees to operate independently.

Based on the above, our Directors believe that we are able to operate independently of our Controlling Shareholders and their close associates.

– 367 –

RELATIONSHIP WITH OUR CONTROLLING SHAREHOLDERS

Financial Independence

We have an independent financial system and make financial decisions according to our Group's own business needs. We have internal control and accounting systems and an independent finance department for discharging the treasury function. We have sufficient capital to operate our business independently, and have adequate internal resources and working capital to support our daily operations. We do not expect to rely on our Controlling Shareholders and their close associates for financing after the Listing as we expect that our working capital will be funded by cash flows generated from operating activities, equity financing, bank loans as well as the proceeds from the Global Offering.

In addition, we are capable of obtaining financing from independent third parties without relying on any guarantee or security provided by our Controlling Shareholders or their respective associates. As of the Latest Practicable Date, there was no outstanding loans or guarantees provided by or granted to our Controlling Shareholders or their respective associates. During the Track Record Period and as of the Latest Practicable Date, we had also received a series of Pre-IPO Investments from third party investors independently. For details of the Pre-IPO Investments, see "History, Reorganization and Corporate Structure."

Based on the above, our Directors believe that we do not place undue reliance on our Controlling Shareholders upon the Listing.

CORPORATE GOVERNANCE MEASURES

Our Directors recognize the importance of good corporate governance in protecting our Shareholders' interests. We have adopted the following measures to safeguard good corporate governance standards and to avoid potential conflicts of interests between our Group and our Controlling Shareholders:

• where a Shareholders' meeting is to be held for considering proposed transactions in which our Controlling Shareholders or any of their respective close associates has a material interest, our Controlling Shareholders will not vote on the resolutions and shall not be counted in the quorum in the voting;

• our Group has established internal control mechanisms to identify connected transactions. Upon the Listing, if any transaction is proposed between our Group and our Controlling Shareholders and their respective associates, we will comply with the requirements of the Articles of Association and the Listing Rules, including, where appropriate, the reporting, annual review by the independent non-executive Directors, announcement and independent shareholders' approval;

– 368 –

CONNECTED TRANSACTIONS

OVERVIEW

Prior to the Listing, our Company has entered into a number of transactions with our connected persons in our ordinary and usual course of business. Upon the Listing, the transactions disclosed in this section will constitute continuing connected transactions under Chapter 14A of the Listing Rules.

OUR CONNECTED PERSONS

As of the Latest Practicable Date, Alibaba Group Holding Limited ("Alibaba"), through its associates, is indirectly interested in approximately 13.66% of the beneficial interests in the share capital of our Company. Following the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised), Alibaba will hold approximately 12.52% beneficial interests in the share capital of our Company (without taking into account the cornerstone investment to be made by Alisoft China as disclosed in the section headed "Cornerstone Investors") and continue to be our substantial shareholder. Accordingly, Alibaba and its close associates (including Alibaba Cloud Computing Ltd. (阿里雲計算有限公司)) are connected persons of our Company under the Listing Rules. Mr. Chen Yingjie is a non-executive Director appointed by Alibaba on our Board. He joined Alibaba in December 2012 and currently serves as the managing director of strategic investment department of Alibaba.

Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng are controlled by Dr. Yan, our executive Director, as to 99%, accordingly, they are connected persons of our Company under the Listing Rules.

PARTIALLY-EXEMPT CONTINUING CONNECTED TRANSACTIONS

Master API Service Agreement

Principal Terms

On December 29, 2025, the Company entered into an agreement relating to the provision of API interface service with Alibaba Cloud Computing Ltd. (the "Master API Service Agreement"), pursuant to which we agreed to provide API services to businesses operated by Alibaba.

The Master API Service Agreement has a term commencing from the Listing Date and ending on December 31, 2028.

Reasons for the Transaction

Since 2023, we have been providing the API services to Alibaba and we have established compatible systems with the relevant parties. Having considered that Alibaba is a well-known internet platform with large customer base and customer traffic, it is mutually beneficial for our Group and Alibaba to cooperate with each other on the provision and purchase of the API services as each of our Group and Alibaba has competitive advantages in its respective business segment. Also, having considered, among others, (i) the reputation of Alibaba as a leading internet platform in the industry, (ii) the customer traffic in the platforms of Alibaba, and (iii)

– 370 –

CONNECTED TRANSACTIONS

the variety of industries and scenarios we will be introduced to through the Alibaba ecosystem, the transactions with Alibaba enable our Group to expand our customer base and market penetration. Our Directors are of the view that the price of service fees payable by Alibaba is in line with market practice and the transactions contemplated under the Master API Service Agreement will provide us with a steady source of income which is in the interest of our Company and our Shareholders as a whole.

Pricing Policy

The amounts paid and to be paid by Alibaba to our Company under the API Service Agreement are determined based on the standard fee rates of API services as provided by the Group from time to time. The sales price for our API services under the Master API Service Agreement are fair and reasonable, and on normal commercial terms no less favorable to our Company than terms offered to Independent Third Parties.

Historical Transaction Amounts

Our sales of API services to Alibaba amounted to nil, USD41,376, USD33,284 and USD233,009 for the years ended December 31, 2022 and 2023 and 2024 and the nine months ended September 30, 2025, respectively.

Annual Caps and Basis of Caps

Our proposed annual caps of the transactions under the Master API Service Agreement for the years ending December 31, 2026, 2027 and 2028 are USD650,000, USD1,000,000 and USD1,500,000, respectively.

We expect revenue from API services provided to Alibaba Group to increase significantly over the next three years, while the current year remains relatively small because the cooperation is still in the early ramp-up stage. The API usage only started to scale in recent months following initial integration and pilot testing, we have already agreed and planned several new projects to adopt our APIs starting next year, and we expect to expand the use cases and traffic with Alibaba Group, which together are expected to drive a material increase in API consumption going forward. In arriving at the above annual caps, the Directors have considered the following factors: (1) the historical transaction amounts paid by Alibaba to the Group in respect of the API services; (2) the expected increase of purchase amount of API services from Alibaba of no less than 200% in the next three years based on existing leads and our current estimation; and (3) the sales price of the API services agreed between the Group and Alibaba in the previous agreements.

Listing Rules Implications

In respect of the continuing connected transactions as described above, the highest applicable percentage ratio calculated for the purpose of Chapter 14A of the Listing Rules is expected to be above 0.1% but will not exceed 5% on an annual basis. Accordingly, the continuing connected transactions as described above are exempt from the independent shareholders' approval requirement under Chapter 14A of the Listing Rules but will be subject to the annual reporting, annual review and announcement requirements under Chapter 14A of the Listing Rules.

– 371 –

CONNECTED TRANSACTIONS

NON-EXEMPT CONTINUING CONNECTED TRANSACTIONS

1. Alibaba Cloud Services Agreement

Principal Terms

On December 29, 2025, the Company entered into an agreement with Alibaba Cloud Computing Ltd. (the "Alibaba Cloud Services Agreement"), pursuant to which our Group agreed to purchase from Alibaba Cloud Computing Ltd. certain cloud products and services.

The Alibaba Cloud Services Agreement has a term commencing from the Listing Date and ending on December 31, 2028.

The Master API Service Agreement and the Alibaba Cloud Service Agreement were separately negotiated and not bundled with each other.

Reasons for the Transaction

Alibaba Cloud Computing Ltd. is a global leader in cloud computing and AI. The cloud services offered by Alibaba Cloud Computing Ltd. has been used in the operations of the Group since 2022 and the long-time cooperation with Alibaba Cloud Computing Ltd. has proved that it can provide the Group with reliable and secured cloud services. The Company believes that it would be beneficial to continue using the cloud services provided by Alibaba Cloud Computing Ltd. to satisfy the increasing demand on cloud computing and data processing capabilities as a result of the business development of the Group.

Pricing Policy

The prices of transactions contemplated under the Alibaba Cloud Services Agreement are based on the standard fee rates as provided by Alibaba Cloud Computing Ltd. from time to time, which sets out the specific service scope and the corresponding prices. The prices offered by Alibaba Cloud Computing Ltd. are comparable to the prices offered by other third-party cloud services providers.

Historical Transaction Amounts

Our purchases of cloud services from Alibaba Cloud Computing Ltd. amounted to approximately USD0.04 million, USD3.1 million, USD10.0 million and USD58.3 million for the years ended December 31, 2022 and 2023 and 2024 and the nine months ended September 30, 2025, respectively.

– 372 –

CONNECTED TRANSACTIONS

Annual Cap and Basis of Cap

Our proposed annual caps of the transactions under the Alibaba Cloud Services Agreement for the years ending December 31, 2026, 2027 and 2028 are US$115 million, US$125 million, and US$135 million, respectively.

In arriving at the above annual caps, the Directors have considered the following factors: (1) the prices of cloud services as set out in the price catalog as published by Alibaba Cloud Computing Ltd. and agreed by the parties; (2) the historical transaction amounts; and (3) the estimated increase in the demand for cloud services of no less than 15% in the next three years based on our current estimation as a result of the business growth of the Company. The increase in our purchases of cloud services from Alibaba Group was primarily due to: (i) the rapid expansion of our business scale, which has led to a substantial increase in the demand for computing power for model training and inference; and (ii) the competitive advantages of Alibaba Group in terms of pricing, technical compatibility and service stability, as a result of which, after a comprehensive assessment among multiple cloud service providers, we increased the proportion of services procured from it on normal commercial terms.

2.

The Business Cooperation Agreement

Principal Terms

In June 2025 and as amended on December 29, 2025, Shanghai MiniMax, Shanghai Jizhi, Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng entered into a business cooperation agreement (the "Business Cooperation Agreement"). Pursuant to the Business Cooperation Agreement, Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng will provide certain operational supporting services to the Group including but not limited to the operation and management of operational support, routine updates and maintenance, contents uploading, promotion and marketing on the applications and websites of the Group, based on price determined on a cost-plus basis by adding a reasonable profit which is in line with market practice and industry peers.

The Business Cooperation Agreement has a term commencing from the Listing Date and ending on December 31, 2028.

Reasons for the Transaction

As part of our Reorganization to streamline our shareholding and corporate structure, Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng were established to, among others, better facilitate our operations in the PRC. For further details, please refer to the paragraph "Corporate Reorganization" under the section headed "History, Reorganization and Corporate Structure" in this Prospectus. We believe it is in the best interests of the Group and our Shareholders as a whole to continue to utilize relevant supportive technical services including but not limited to operational support, routine maintenance, promotion, marketing and data analysis services provided by Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng after Listing.

– 373 –

CONNECTED TRANSACTIONS

Pricing Policy

The price was determined on a cost-plus basis by adding a reasonable profit which is in line with market practice and industry peers. Relevant costs include but not limited to direct labor costs, costs and expenses associated with cloud services, use of platforms and tools, software subscriptions, technical support, and other reasonable expenses related to such services that are mutually confirmed in writing by both parties. The terms under the Business Cooperation Agreement are fair and reasonable, and on normal commercial terms no less favorable to our Company than terms offered to Independent Third Parties.

Historical Transaction Amounts

Our transaction amounts with Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng amounted to nil for each of the years ended December 31, 2022 and 2023 and 2024 and USD27,225 for the nine months ended September 30, 2025.

Annual Cap and Basis of Cap

Our proposed annual caps of the transactions under the Business Cooperation Agreement for the years ending December 31, 2026, 2027 and 2028 are US$4.0 million, US$4.8 million and US$5.6 million, respectively.

The expected increase in our service procurement from Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng, comparing with the historical transaction amount after Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng were established in 2025, is primarily due to the expansion of our business scale and user base, which will lead to heightened requirements for security and compliance, as well as the continuous iteration of our applications, the rollout of new business modules, all of which will significantly increase the volume of operational support, routine maintenance, promotion and marketing work, and, in turn, drive a gradual rise in the related service procurement amount. In arriving at the above annual caps, the Directors have considered the following factors:

(1)

the expected increase in our demand of the relevant service of no less than 30% in the next three years based on our current estimation due to our business expansion which are primarily based on our internal business plan and financial budget as we projected the growth in our average annual active users, traffic volume and products to be supported by Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng's services over the next three years. With stable unit service pricing, our aggregate service procurement from the next three years would increase by over 30% in the next three years. In addition, the expected increase in our service procurement from Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng, comparing with the historical transaction amount after Shanghai Jizhi Wujie and Shanghai Jizhi Zongheng were established in 2025, is primarily due to the expansion of our business scale and user base, which will lead to heightened requirements for security and compliance, as well as

– 374 –

CONNECTED TRANSACTIONS

the continuous iteration of our applications, the rollout of new business modules, all of which will significantly increase the volume of operational support, routine maintenance, promotion and marketing work, and, in turn, drive a gradual rise in the related service procurement amount; and

(2) the prevailing market price or quotations from other independent service providers.

**Listing Rules Implications**

As one or more of the applicable percentage ratios in respect of the transactions under the Alibaba Cloud Services Agreement and Business Cooperation Agreement are expected to exceed 5% on an annual basis, such transactions will, upon the Listing, be subject to the reporting, annual review, announcement and the independent Shareholders' approval requirements under Chapter 14A of the Listing Rules.

**CONFIRMATION OF OUR DIRECTORS**

Our Directors (including independent non-executive Directors) consider that (i) the partially-exempt and non-exempt continuing connected transactions have been and will be entered into in the ordinary and usual course of business of our Group, on normal commercial terms, are fair and reasonable and in the interests of our Group and Shareholders as a whole; and (ii) the proposed annual caps in respect of the partially-exempt and non-exempt continuing connected transactions are fair and reasonable, and in the interests of our Group and Shareholders as a whole.

**CONFIRMATION OF THE JOINT SPONSORS**

The Joint Sponsors are of the view that (i) the partially-exempt and non-exempt continuing connected transactions as set out above have been and will be entered into in the ordinary and usual course of business of our Group, on normal commercial terms or better, are fair and reasonable and in the interests of our Group and Shareholders as a whole; and (ii) the proposed annual caps are fair and reasonable, and in the interests of our Group and Shareholders as a whole.

**WAIVERS FROM STRICT COMPLIANCE WITH THE LISTING RULES**

In relation to the continuing connected transactions above, we have applied for, and the Stock Exchange has granted to us, a waiver from strict compliance with the announcement, circular and independent Shareholders' approval requirement under Chapter 14A of the Listing Rules pursuant to Rule 14A.105 of the Listing Rules, subject to the condition that the aggregate value of such continuing connected transactions for the years ended December 31, 2026, 2027 and 2028 shall not exceed relevant annual amounts stated above.

– 375 –

SUBSTANTIAL SHAREHOLDERS

**SUBSTANTIAL SHAREHOLDERS**

So far as our Directors are aware, immediately following completion of the Global Offering, assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised, the following persons will have interests and/or short positions in the Shares or underlying shares of our Company which would fall to be disclosed to us pursuant to the provisions of Divisions 2 and 3 of Part XV of the SFO, or, who are directly or indirectly, interested in 10% or more of the nominal value of any class of share capital carrying rights to vote in all circumstances at general meetings of our Company:

| Name of Substantial Shareholder | Capacity/Nature of Interest(1) | Number of Shares | Approximate percentage of shareholding in respective class of Share of our Company upon completion of the Global Offering assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised | Approximate percentage of shareholding in the issued share capital of our Company upon completion of the Global Offering assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised(2) | |---|---|---|---|---| | **Class B Ordinary Shares** | | | | | | Alpha EXP(3) | Beneficial owner | 62,593,180 | 77.18% | 20.49% | | Scaling EXP Limited(3) | Interest in controlled corporations | 62,593,180 | 77.18% | 20.49% | | Trident Trust Company (HK) Limited(3) | Trustee | 62,593,180 | 77.18% | 20.49% | | MiniMax Limited(3) | Beneficial owner | 15 | 0.00002% | 0.000005% | | MiniMax Awakening(3) | Beneficial owner | 11,509,339 | 14.19% | 3.77% | | Local Linearity(3) | Interest in controlled corporations | 11,509,354 | 14.19% | 3.77% | | Dr. Yan | Interest in controlled corporations | 74,102,534 | 91.37% | 24.26% | | Floating Sky | Beneficial Owner | 7,000,000 | 8.63% | 2.29% | | Floating Cloud Limited | Interest in controlled corporations | 7,000,000 | 8.63% | 2.29% | | Trident Trust Company (HK) Limited(3) | Trustee | 7,000,000 | 8.63% | 2.29% | | Ms. Yun | Interest in controlled corporations | 7,000,000 | 8.63% | 2.29% | | **Class A Ordinary Shares** | | | | | | MiniMax Matrix(3) | Beneficial owner | 5,000,000 | 2.23% | 1.64% | | Local Linearity(3) | Interested in controlled corporations | 5,000,000 | 2.23% | 1.64% | | Dr. Yan | Interested in controlled corporations | 5,000,000 | 2.23% | 1.64% | | Alibaba China Holding Limited | Beneficial owner | 38,247,987 | 17.05% | 12.52% |

– 376 –

**SUBSTANTIAL SHAREHOLDERS**

| Name of Substantial Shareholder | Capacity/Nature of Interest(1) | |---|---|

SUBSTANTIAL SHAREHOLDERS

Alisoft Investment Holding Limited(4) · · · · Interest in controlled corporations Alisoft Holding Limited(4) · · · · · · · · · · Interest in controlled corporations Alibaba Group Holding Limited(4) · · · · · · Interest in controlled corporations miHoYo Limited · · · · · · · · · · · · · · · Beneficial owner Shanghai Fanxing Dingchuang Technology Company Limited(5) · · · · · · · · · · · · Interest in controlled corporations Luo Yuhao(5) · · · · · · · · · · · · · · · · · Interest in controlled corporations XAM Holdings Limited(6) · · · · · · · · · · Beneficial owner NVMB IV Holdings Limited(6) · · · · · · · · Interest in controlled corporations BXA Holdings II, L.P.(6) · · · · · · · · · · · Interest in controlled corporations JNR Holdings GP Limited(6) · · · · · · · · · Interest in controlled corporations Mr. Colm O'Connell(1)(7) · · · · · · · · · · Interest in controlled corporations

| | Number of Shares | Approximate percentage of shareholding in respective class of Share of our Company upon completion of the Global Offering assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised | Approximate percentage of shareholding in the issued share capital of our Company upon completion of the Global Offering assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised(2) | |---|---|---|---| | Alisoft Investment Holding Limited | 38,247,987 | 17.05% | 12.52% | | Alisoft Holding Limited | 38,247,987 | 17.05% | 12.52% | | Alibaba Group Holding Limited | 38,247,987 | 17.05% | 12.52% | | miHoYo Limited | 16,015,779 | 7.14% | 5.24% | | Shanghai Fanxing Dingchuang Technology Company Limited | 16,015,779 | 7.14% | 5.24% | | Luo Yuhao | 16,015,779 | 7.14% | 5.24% | | XAM Holdings Limited | 14,201,184 | 6.33% | 4.65% | | NVMB IV Holdings Limited | 14,201,184 | 6.33% | 4.65% | | BXA Holdings II, L.P. | 14,201,184 | 6.33% | 4.65% | | JNR Holdings GP Limited | 14,201,184 | 6.33% | 4.65% | | Mr. Colm O'Connell | 16,544,380 | 7.37% | 5.42% |

Notes:

(1) All interests stated are long positions.

(2) The table above assumes the Preferred Shares will be automatically converted into Class A Ordinary Shares on a 1:1 basis.

(3) MiniMax Awakening and MiniMax Limited are wholly owned by Dr. Yan through Local Linearity. MiniMax Matrix is also a controlled entity of Dr. Yan through Local Linearity. Alpha EXP is held by Scaling EXP Limited as to 99% and Local Linearity as to 1%. Scaling EXP Limited is wholly-owned by Trident Trust Company (Hong Kong) Limited, which acts as the trustee of Alpha EXP Trust. Alpha EXP Trust is a trust established by Dr. Yan (as settlor) for the benefit of himself.

Floating Sky is held by Floating Cloud Limited as to 99% and Apricity Investment Limited as to 1%. Apricity Investment Limited is wholly-owned by Ms. Yun. Floating Cloud Limited is wholly-owned by Trident Trust Company (Hong Kong) Limited, which acts as the trustee of Floating Sky Trust. Floating Sky Trust is a trust established by Ms. Yun (as settlor) for the benefit of herself.

– 377 –

SUBSTANTIAL SHAREHOLDERS

Accordingly, under the SFO, Dr. Yan is deemed to be interested in the Shares held by MiniMax Awakening, MiniMax Limited, MiniMax Matrix and Alpha EXP. Local Linearity is deemed to be interested in the Shares held by MiniMax Awakening, MiniMax Limited and MiniMax Matrix. Scaling EXP Limited and Trident Trust Company (Hong Kong) Limited is deemed to be interested in the Shares held by Alpha EXP. Ms. Yun, Floating Cloud Limited and Trident Trust Company (Hong Kong) Limited is deemed to be interested in the Shares held by Floating Sky.

(4) Alibaba China Holding Limited is controlled by Alisoft Investment Holding Limited, a company controlled by Alisoft Holding Limited, which is in turn controlled by Alibaba Group Holding Limited. Therefore, each of Alisoft Investment Holding Limited, Alisoft Holding Limited, and Alibaba Group Holding Limited is deemed to be interested in the Shares held by Alibaba China Holding Limited. The number of Shares held by Alibaba China Holding Limited upon Listing does not take into account the cornerstone investment to be made by Alisoft China as disclosed in the section headed "Cornerstone Investors".

(5) miHoYo Limited is wholly owned by Shanghai Fanxing Dingchuang Technology Company Limited (上海繁星顶创科技有限公司), which is wholly owned by Luo Yuhao (罗宇昊). Therefore, each of Shanghai Fanxing Dingchuang Technology Company Limited and Luo Yuhao is deemed to be interested in the Shares held by miHoYo Limited.

(6) For further details, please refer to the section headed "History, Reorganization and Corporate Structure — 4. Information relating to our key Pre-IPO Investors — MNM Holdings Limited and XAM Holdings Limited" of this Prospectus.

(7) Mr. Colm O'Connell is the sole shareholder of each of JNR Holdings GP Limited and BXA Holdings II GP Limited, being the general partner of BXA Holdings II, L.P. and BXA Holdings, L.P., respectively, which in turn indirectly held 14,201,184 Shares and 2,343,196 Shares through XAM Holdings Limited and MNM Holdings Limited, respectively. Mr. Colm O'Connell is deemed to be interested in these Shares.

Save as disclosed above and the section headed "Statutory and General Information — C. Further Information about our Directors and Substantial Shareholders" in Appendix IV to this Prospectus, our Directors are not aware of any person who will, immediately following completion of the Global Offering, assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised, have any interest and/or short position in the Shares or underlying Shares of our Company which will be required to be disclosed to our Company and the Stock Exchange pursuant to the provisions of Divisions 2 and 3 of Part XV of the SFO, or, who are directly or indirectly, interested in 10% or more of the nominal value of any class of share capital carrying rights to vote in all circumstances at general meeting of our Company or other members of the Group.

– 378 –

CORNERSTONE INVESTORS

THE CORNERSTONE PLACING

We have entered into cornerstone investment agreements (each a "Cornerstone Investment Agreement", and together the "Cornerstone Investment Agreements") with the cornerstone investors set out below (each a "Cornerstone Investor", and together the "Cornerstone Investors"), pursuant to which the Cornerstone Investors have agreed to, subject to certain conditions, subscribe at the Offer Price for a certain number of Offer Shares that may be purchased for an aggregate amount of approximately US$350 million (approximately HK$2,723 million) (the "Cornerstone Placing"). The calculations in this section, which are based on the exchange rates as disclosed in the section headed "Information about this Prospectus and the Global Offering", are for illustration purpose.

Assuming an Offer Price of HK$151.0, being the low-end of the indicative Offer Price range set out in this Prospectus, the total number of Offer Shares to be subscribed by the Cornerstone Investors would be 18,034,240 Offer Shares. The table below reflects the shareholding percentage immediately after the completion of the Global Offering assuming there is no other change made to the issued share capital of our Company between the Latest Practicable Date and the Listing Date (or the date of exercise of Over-allotment Option (where applicable)).

| | Assuming the Offer Size Adjustment Option is not exercised | | | | Assuming the Offer Size Adjustment Option is exercised in full | | | | |---|---|---|---|---|---|---|---|---| | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | | | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | | | 71.03% | 5.90% | 61.77% | 5.83% | 61.77% | 5.83% | 53.71% | 5.75% |

Assuming an Offer Price of HK$158.0, being the mid-point of the Offer Price range set out in this Prospectus, the total number of Offer Shares to be subscribed by the Cornerstone Investors would be 17,235,120 Offer Shares. The table below reflects the shareholding percentage immediately after the completion of the Global Offering assuming there is no other change made to the issued share capital of our Company between the Latest Practicable Date and the Listing Date (or the date of exercise of Over-allotment Option (where applicable)).

| | Assuming the Offer Size Adjustment Option is not exercised | | | | Assuming the Offer Size Adjustment Option is exercised in full | | | | |---|---|---|---|---|---|---|---|---| | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | | | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | | | 67.88% | 5.64% | 59.03% | 5.57% | 59.03% | 5.57% | 51.33% | 5.50% |

– 379 –

CORNERSTONE INVESTORS

Assuming an Offer Price of HK$165.0, being the high-end of the Offer Price range set out in this Prospectus, the total number of Offer Shares to be subscribed by the Cornerstone Investors would be 16,504,040 Offer Shares. The table below reflects the shareholding percentage immediately after the completion of the Global Offering assuming there is no other change made to the issued share capital of our Company between the Latest Practicable Date and the Listing Date (or the date of exercise of Over-allotment Option (where applicable)).

| | Assuming the Offer Size Adjustment Option is not exercised | | | | Assuming the Offer Size Adjustment Option is exercised in full | | | | |---|---|---|---|---|---|---|---|---| | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | | | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | Approximate % of the Offer Share | Approximate % of the Shares in issue | | | 65.00% | 5.40% | 56.53% | 5.34% | 56.53% | 5.34% | 49.15% | 5.26% |

CORNERSTONE INVESTORS

Our Company is of the view that the Cornerstone Placing will help to raise the profile of our Company and to signify that such investors have confidence in our business and prospect. Our Company became acquainted with each of the Cornerstone Investors through the Group's business network, previous financing, or introduction by the Overall Coordinators and Capital Market Intermediaries in the Global Offering.

To the best knowledge of our Company and save as that certain Cornerstone Investors are our existing Shareholders or close associates of existing Shareholders as disclosed below, each of the Cornerstone Investors and their respective ultimate beneficial owners (i) is an Independent Third Party; (ii) none of the Cornerstone Investors is accustomed to taking instructions from our Company, the Directors, chief executive, our Controlling Shareholders, substantial shareholders, existing Shareholders or any of their respective subsidiaries or their respective close associates in relation to the acquisition, disposal, voting or other disposition of the Offer Shares; (iii) none of the subscription of the relevant Offer Shares by any of the Cornerstone Investors is financed directly or indirectly by our Company, the Directors, chief executive, our Controlling Shareholders, substantial shareholders, existing Shareholders or any of their respective subsidiaries or their respective close associates; (iv) each Cornerstone Investor will be utilizing their internal resources as their source of funding for the subscription of the Offer Shares; and (v) no approval from other stock exchange is required for each Cornerstone Investor's investment in our Company as described in this section.

Among the Cornerstone Investors, Alisoft China (as defined below), Aspex Master Fund, Abstract Enigma Limited, IDG Breyer Fund (as defined below), Janchor Funds (as defined below) and MPC VII (as defined below) are our existing Shareholders or close associates of existing Shareholders. The Stock Exchange has granted us a waiver from strict compliance with the requirements under Rule under Rule 10.04 of the Listing Rules and consent under paragraph 1C of Appendix F1 to the Listing Rules to permit Offer Shares in the International Offering to be placed to certain existing Shareholders and/or their close associates. For further details, please see the section headed "Waivers and Exemption".

– 380 –

CORNERSTONE INVESTORS

The Cornerstone Placing will form part of the International Offering and the Cornerstone Investors will not subscribe for any Offer Shares under the Global Offering other than pursuant to the Cornerstone Investment Agreements. The Offer Shares to be subscribed by the Cornerstone Investors will rank pari passu in all respect with the fully paid Shares in issue and will be counted towards the public float of our Company under Rule 8.08 of the Listing Rules (except for Alisoft China). Except for Alisoft China, immediately following the completion of the Global Offering, none of the Cornerstone Investors will become a substantial shareholder of the Company, and the Cornerstone Investors will not have any Board representation in our Company. Other than a guaranteed allocation of the relevant Offer Shares at the final Offer Price, the Cornerstone Investors do not have any preferential rights in the Cornerstone Investment Agreements compared with other public Shareholders. There are no side arrangements between our Company and the Cornerstone Investors or any benefit, direct or indirect, conferred on the Cornerstone Investors by virtue of or in relation to the Cornerstone Placing.

The total number of Offer Shares to be subscribed by the Cornerstone Investors may be affected by reallocation of the Offer Shares between the International Offering and the Hong Kong Public Offering in the event of over-subscription under the Hong Kong Public Offering as described in the paragraph headed "Structure of the Global Offering — The Hong Kong Public Offering — Reallocation" in this Prospectus. The number of Offer Shares to be acquired by each Cornerstone Investor may be reduced on a pro rata basis in accordance with the terms of the Cornerstone Investment Agreement to satisfy the short fall, after taking into account the requirements under Appendix F1 to the Listing Rules as well as the discretion of the Joint Global Coordinators and the Overall Coordinators (for themselves and on behalf of the International Underwriters) to exercise the Over-allotment Option.

There will be no delayed delivery or deferred settlement of Offer Shares to be subscribed by the Cornerstone Investors and the consideration will be settled by the Cornerstone Investors before the Listing Date. The Offer Shares to be subscribed by the Cornerstone Investors may be affected by reallocation in the event of over-subscription under the Hong Kong Public Offering, as described in "Structure of the Global Offering — The Hong Kong Public Offering — Reallocation". Details of the actual number of Offer Shares to be allocated to the Cornerstone Investors will be disclosed in the allotment results announcement to be issued by us on or around January 8, 2026.

– 381 –

CORNERSTONE INVESTORS

OUR CORNERSTONE INVESTORS

Set out below in the aggregate number of Offer Shares, and the corresponding percentages to the Offer Shares and our Company's total issued share capital under the Cornerstone Placing:

Based on the Offer Price of HK$151.0 (being the low-end of the indicative Offer Price range)

Assuming the Offer Size Adjustment Option is not exercised

| Cornerstone Investor | Investment amount(1) (US$) | Number of Offer Shares(2) | Approximate % of the Offer Shares | |---|---|---|---|

CORNERSTONE INVESTORS

Based on the Offer Price of HK$158.0 (being the mid-point of the indicative Offer Price range)

Assuming the Offer Size Adjustment Option is not exercised

| Cornerstone Investor | Investment amount(1) (US$) | Number of Offer Shares(2) | Approximate % of the Offer Share | Assuming the Over-allotment Option is not exercised Approximate % of the Shares in issue | Assuming the Offer Size Adjustment Option is exercised in full Approximate % of the Offer Share | Assuming the Offer Size Adjustment Option is exercised in full — Assuming the Over-allotment Option is not exercised Approximate % of the Shares in issue | Assuming the Offer Size Adjustment Option is exercised in full — Assuming the Over-allotment Option is exercised in full Approximate % of the Offer Share | Assuming the Offer Size Adjustment Option is exercised in full — Assuming the Over-allotment Option is exercised in full Approximate % of the Shares in issue | Assuming the Over-allotment Option is exercised in full Approximate % of the Offer Share | Assuming the Over-allotment Option is exercised in full Approximate % of the Shares in issue | |---|---|---|---|---|---|---|---|---|---|---| | ADIA | 65,000,000 | 3,200,840 | 12.61% | 1.05% | 10.96% | 1.04% | 10.96% | 1.04% | 9.53% | 1.02% | | Alisoft China | 30,000,000 | 1,477,300 | 5.82% | 0.48% | 5.06% | 0.48% | 5.06% | 0.48% | 4.40% | 0.47% | | Aspex Master Fund | 35,000,000 | 1,723,520 | 6.79% | 0.56% | 5.90% | 0.56% | 5.90% | 0.56% | 5.13% | 0.55% | | Boyu | 35,000,000 | 1,723,520 | 6.79% | 0.56% | 5.90% | 0.56% | 5.90% | 0.56% | 5.13% | 0.55% | | China Universal (HK) | 15,000,000 | 738,640 | 2.91% | 0.24% | 2.53% | 0.24% | 2.53% | 0.24% | 2.20% | 0.24% | | Eastspring | 15,000,000 | 738,640 | 2.91% | 0.24% | 2.53% | 0.24% | 2.53% | 0.24% | 2.20% | 0.24% | | E Fund Management | 10,000,000 | 492,420 | 1.94% | 0.16% | 1.69% | 0.16% | 1.69% | 0.16% | 1.47% | 0.16% | | IDG Breyer Fund | 15,000,000 | 738,640 | 2.91% | 0.24% | 2.53% | 0.24% | 2.53% | 0.24% | 2.20% | 0.24% | | Janchor Funds | 35,000,000 | 1,723,520 | 6.79% | 0.56% | 5.90% | 0.56% | 5.90% | 0.56% | 5.13% | 0.55% | | Martis Fund, L.P. | 15,000,000 | 738,640 | 2.91% | 0.24% | 2.53% | 0.24% | 2.53% | 0.24% | 2.20% | 0.24% | | Mirae Asset Securities | 20,000,000 | 984,860 | 3.88% | 0.32% | 3.37% | 0.32% | 3.37% | 0.32% | 2.93% | 0.31% | | MPC VII | 15,000,000 | 738,640 | 2.91% | 0.24% | 2.53% | 0.24% | 2.53% | 0.24% | 2.20% | 0.24% | | Perseverance Asset Management | 25,000,000 | 1,231,080 | 4.85% | 0.40% | 4.22% | 0.40% | 4.22% | 0.40% | 3.67% | 0.39% | | Taikang Life | 20,000,000 | 984,860 | 3.88% | 0.32% | 3.37% | 0.32% | 3.37% | 0.32% | 2.93% | 0.31% | | Total | 350,000,000 | 17,235,120 | 67.88% | 5.64% | 59.03% | 5.57% | 59.03% | 5.57% | 51.33% | 5.50% |

Notes: 1. The investment amount excludes brokerage, SFC transaction levy, AFRC transaction levy and Stock Exchange trading fee, and is calculated based on the exchange rate set out in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion" in this Prospectus. The number of Offer Shares to be subscribed by the Cornerstone Investors are subject to the exchange rate to be determined in accordance with each relevant Cornerstone Investment Agreement.

2. Rounded down to the nearest whole board lot of 20 Shares, and is calculated based on the exchange rate set out in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion" in this Prospectus.

– 383 –

CORNERSTONE INVESTORS

Based on the Offer Price of HK$165.0 (being the high-end of the indicative Offer Price range)

Assuming the Offer Size Adjustment Option is not exercised

| Cornerstone Investor | Investment amount(1) (US$) | Number of Offer Shares(2) | Approximate % of the Offer Share | Assuming the Over-allotment Option is not exercised Approximate % of the Shares in issue | Assuming the Offer Size Adjustment Option is exercised in full Approximate % of the Offer Share | Assuming the Offer Size Adjustment Option is exercised in full — Assuming the Over-allotment Option is not exercised Approximate % of the Shares in issue | Assuming the Offer Size Adjustment Option is exercised in full — Assuming the Over-allotment Option is exercised in full Approximate % of the Offer Share | Assuming the Offer Size Adjustment Option is exercised in full — Assuming the Over-allotment Option is exercised in full Approximate % of the Shares in issue | Assuming the Over-allotment Option is exercised in full Approximate % of the Offer Share | Assuming the Over-allotment Option is exercised in full Approximate % of the Shares in issue | |---|---|---|---|---|---|---|---|---|---|---| | ADIA | 65,000,000 | 3,349,220 | 13.19% | 1.10% | 11.47% | 1.08% | 11.47% | 1.08% | 9.97% | 1.07% | | Alisoft China | 30,000,000 | 1,545,800 | 6.09% | 0.51% | 5.29% | 0.50% | 5.29% | 0.50% | 4.60% | 0.49% | | Aspex Master Fund | 35,000,000 | 1,803,420 | 7.10% | 0.59% | 6.18% | 0.58% | 6.18% | 0.58% | 5.37% | 0.58% | | Boyu | 35,000,000 | 1,803,420 | 7.10% | 0.59% | 6.18% | 0.58% | 6.18% | 0.58% | 5.37% | 0.58% | | China Universal (HK) | 15,000,000 | 772,900 | 3.04% | 0.25% | 2.65% | 0.25% | 2.65% | 0.25% | 2.30% | 0.25% | | Eastspring | 15,000,000 | 772,900 | 3.04% | 0.25% | 2.65% | 0.25% | 2.65% | 0.25% | 2.30% | 0.25% | | E Fund Management | 10,000,000 | 515,260 | 2.03% | 0.17% | 1.76% | 0.17% | 1.76% | 0.17% | 1.53% | 0.16% | | IDG Breyer Fund | 15,000,000 | 772,900 | 3.04% | 0.25% | 2.65% | 0.25% | 2.65% | 0.25% | 2.30% | 0.25% | | Janchor Funds | 35,000,000 | 1,803,420 | 7.10% | 0.59% | 6.18% | 0.58% | 6.18% | 0.58% | 5.37% | 0.58% | | Martis Fund, L.P. | 15,000,000 | 772,900 | 3.04% | 0.25% | 2.65% | 0.25% | 2.65% | 0.25% | 2.30% | 0.25% | | Mirae Asset Securities | 20,000,000 | 1,030,520 | 4.06% | 0.34% | 3.53% | 0.33% | 3.53% | 0.33% | 3.07% | 0.33% | | MPC VII | 15,000,000 | 772,900 | 3.04% | 0.25% | 2.65% | 0.25% | 2.65% | 0.25% | 2.30% | 0.25% | | Perseverance Asset Management | 25,000,000 | 1,288,160 | 5.07% | 0.42% | 4.41% | 0.42% | 4.41% | 0.42% | 3.84% | 0.41% | | Taikang Life | 20,000,000 | 1,030,520 | 4.06% | 0.34% | 3.53% | 0.33% | 3.53% | 0.33% | 3.07% | 0.33% | | Total | 350,000,000 | 18,034,240 | 71.03% | 5.90% | 61.77% | 5.83% | 61.77% | 5.83% | 53.71% | 5.75% |

Notes: 1. The investment amount excludes brokerage, SFC transaction levy, AFRC transaction levy and Stock Exchange trading fee, and is calculated based on the exchange rate set out in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion" in this Prospectus. The number of Offer Shares to be subscribed by the Cornerstone Investors are subject to the exchange rate to be determined in accordance with each relevant Cornerstone Investment Agreement.

2. Rounded down to the nearest whole board lot of 20 Shares, and is calculated based on the exchange rate set out in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion" in this Prospectus.

– 382 –

CORNERSTONE INVESTORS

The following information about the other Cornerstone Investors was provided to our Company by the Cornerstone Investors in relation to the Cornerstone Placing.

| Cornerstone Investor | Assuming the Over-allotment Option is not exercised | | | Assuming the Offer Size Adjustment Option is exercised in full | | | | Assuming the Over-allotment Option is not exercised | | Assuming the Over-allotment Option is exercised in full | | |---|---|---|---|---|---|---|---|---|---|---|---| | | Shares in issue | Approximate % of the Offer Share | Approximate % of the issue | Approximate % of the Offer Share | Approximate % of the issue | Approximate % of the Offer Share | Approximate % of the issue | Approximate % of the Offer Share | Approximate % of the issue | Approximate % of the Offer Share | Approximate % of the issue |

| Cornerstone Investor | Shares | Approx. Shares (lots) | % of Offer Share (no OA) | % of issue (no OA) | % of Offer Share (OA adj., no OA) | % of issue (OA adj., no OA) | % of Offer Share (OA adj., OA exercised) | % of issue (OA adj., OA exercised) | % of Offer Share (OA exercised) | % of issue (OA exercised) | |---|---|---|---|---|---|---|---|---|---|---| | ADIA | 65,000,000 | 3,065,040 | 12.07% | 1.00% | 10.50% | 0.99% | 10.50% | 0.99% | 9.13% | 0.98% | | Alisoft China | 30,000,000 | 1,414,640 | 5.57% | 0.46% | 4.85% | 0.46% | 4.85% | 0.46% | 4.21% | 0.45% | | Aspex Master Fund | 35,000,000 | 1,650,400 | 6.50% | 0.54% | 5.65% | 0.53% | 5.65% | 0.53% | 4.92% | 0.53% | | Boyu | 35,000,000 | 1,650,400 | 6.50% | 0.54% | 5.65% | 0.53% | 5.65% | 0.53% | 4.92% | 0.53% | | China Universal (HK) | 15,000,000 | 707,320 | 2.79% | 0.23% | 2.42% | 0.23% | 2.42% | 0.23% | 2.11% | 0.23% | | Eastspring | 15,000,000 | 707,320 | 2.79% | 0.23% | 2.42% | 0.23% | 2.42% | 0.23% | 2.11% | 0.23% | | E Fund Management | 10,000,000 | 471,540 | 1.86% | 0.15% | 1.61% | 0.15% | 1.61% | 0.15% | 1.40% | 0.15% | | IDG Breyer Fund | 15,000,000 | 707,320 | 2.79% | 0.23% | 2.42% | 0.23% | 2.42% | 0.23% | 2.11% | 0.23% | | Janchor Funds | 35,000,000 | 1,650,400 | 6.50% | 0.54% | 5.65% | 0.53% | 5.65% | 0.53% | 4.92% | 0.53% | | Martis Fund, L.P. | 15,000,000 | 707,320 | 2.79% | 0.23% | 2.42% | 0.23% | 2.42% | 0.23% | 2.11% | 0.23% | | Mirae Asset Securities | 20,000,000 | 943,080 | 3.71% | 0.31% | 3.23% | 0.30% | 3.23% | 0.30% | 2.81% | 0.30% | | MPC VII | 15,000,000 | 707,320 | 2.79% | 0.23% | 2.42% | 0.23% | 2.42% | 0.23% | 2.11% | 0.23% | | Perseverance Asset Management | 25,000,000 | 1,178,860 | 4.64% | 0.39% | 4.04% | 0.38% | 4.04% | 0.38% | 3.51% | 0.38% | | Taikang Life | 20,000,000 | 943,080 | 3.71% | 0.31% | 3.23% | 0.30% | 3.23% | 0.30% | 2.81% | 0.30% | | **Total** | **350,000,000** | **16,504,040** | **65.00%** | **5.40%** | **56.53%** | **5.34%** | **56.53%** | **5.34%** | **49.15%** | **5.26%** |

Notes:

1. The investment amount excludes brokerage, SFC transaction levy, AFRC transaction levy and Stock Exchange trading fee, and is calculated based on the exchange rate set out in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion" in this Prospectus. The number of Offer Shares to be subscribed by the Cornerstone Investors are subject to the exchange rate to be determined in accordance with each relevant Cornerstone Investment Agreement.

2. Rounded down to the nearest whole board lot of 20 Shares, and is calculated based on the exchange rate set out in the section headed "Information about this Prospectus and the Global Offering — Exchange Rate Conversion" in this Prospectus.

– 384 –

CORNERSTONE INVESTORS

The following information about the other Cornerstone Investors was provided to our Company by the Cornerstone Investors in relation to the Cornerstone Placing.

**ADIA**

Abu Dhabi Investment Authority ("ADIA") is a public institution established by the Government of the Emirate of Abu Dhabi in 1976 as an independent investment institution. ADIA's objective is to receive funds of the Government of Abu Dhabi allocated for investment, and invest and reinvest those funds, for the general benefit of the Emirate of Abu Dhabi.

ADIA manages a global investment portfolio that is diversified across more than two dozen asset classes and sub-categories including developed equities, emerging market equities, small cap equities, government bonds, credit, fixed income, real estate, infrastructure, private equity, cash and alternatives.

**Alisoft China**

Alisoft China Holding Limited ("Alisoft China") is a limited liability company incorporated in Hong Kong and an indirect wholly-owned subsidiary of Alibaba Group Holding Limited ("Alibaba Group"). Alisoft China is the holding company of certain PRC subsidiaries of Alibaba Group primarily involved in the operation of cloud computing business. Alibaba Group is a company incorporated in the Cayman Islands, with its American depositary shares, each representing eight ordinary shares, listed on the New York Stock Exchange (Stock Symbol: BABA), and its ordinary shares listed on the Main Board of the Stock Exchange (Stock Code: 9988). Alibaba Group's mission is to make it easy to do business anywhere. Alibaba Group aims to build the future infrastructure of commerce and envisions that its customers will meet, work and live at Alibaba, and that it aspires to be a good company that will last for 102 years. Alibaba Group's core businesses are comprised of e-commerce and cloud computing.

**Aspex Master Fund**

Aspex Master Fund ("AMF") is a company incorporated and registered as a mutual fund in the Cayman Islands. AMF is managed by Aspex Management (HK) Limited ("Aspex Management"), a company incorporated in Hong Kong and licensed by the Securities and Futures Commission of Hong Kong to carry out type 9 (asset management) regulated activities in Hong Kong. Mr. Li Ho Kei is the ultimate beneficial owner of Aspex Management and controls the voting rights of AMF, in each case through a holding entity. Mr. Li Ho Kei is an Independent Third Party to the Company. No other investor holds an ultimate beneficial ownership of 30% or more in AMF or Aspex Management.

– 385 –

CORNERSTONE INVESTORS

**Boyu**

Abstract Enigma Limited is a company incorporated under the laws of the Cayman Islands and a controlled subsidiary of Boyu Capital Offshore Fund. Boyu Capital Offshore Fund is an exempted company incorporated under the laws of the Cayman Islands and an investment fund managed by Boyu Capital Management (Singapore) Pte. Ltd. ("Boyu"). Boyu holds a capital markets services license and is regulated by the Monetary Authority of Singapore. Boyu provides catalytic capital and strategic support for leading companies in sectors including technology, healthcare, consumer and sustainable energy. Boyu is 100% indirectly owned by Boyu Group, LLC, which is in turn ultimately controlled by Mr. Xiaomeng Tong, an Independent Third Party. There is no single investor holding 30% or more interest in Abstract Enigma Limited through Boyu Capital Offshore Fund.

**China Universal (HK)**

China Universal Asset Management (Hong Kong) Company Limited ("China Universal (HK)"), founded in November 2009, is a wholly owned subsidiary of China Universal Asset Management Co., Ltd, an asset management company with assets under management of over RMB1,100 billion as of 31 December 2024. China Universal (HK) is among the first group of Chinese fund management company subsidiaries established outside of Mainland China. China Universal (HK) is licensed by the Hong Kong Securities and Futures Commission to carry on Type 1 (Dealing in Securities), Type 4 (Advising on Securities) and Type 9 (Asset Management) regulated activities under Part V of the Securities and Futures Ordinance. China Universal (HK) manages investment funds, provides investment advisory services, and manages discretionary accounts.

The subscription of the Offer Shares as a cornerstone investor will be made by China Universal (HK) in its capacity as the investment manager on a discretionary basis for and on behalf of Better Supply Chain (HK) Holdings Co., Limited and Seraphim Advantage Inc. Zimei PENG and Jun WANG holds 30% or more interest in Better Supply Chain (HK) Holdings Co., Limited and Seraphim Advantage Inc. respectively.

**Eastspring**

Eastspring Investments (Singapore) Limited ("Eastspring"), established in 1994 and headquartered in Singapore, brings over 30 years of investment expertise in Asia. Eastspring is ultimately 100% held by Prudential plc, a publicly listed company, which has dual primary listings on the Stock Exchange of Hong Kong (HKEX: 2378) and the London Stock Exchange (LSE: PRU), and a secondary listing on the Singapore Stock Exchange (SGX: K6S) and a listing on the New York Stock Exchange (NYSE: PUK) in the form of American Depositary Receipts.

As of September 30, 2025, Eastspring manages US$286 billion in assets. Eastspring offers a diverse range of investment strategies for both Asian and non-Asian institutions, working closely with its local offices to deliver tailored solutions to institutional clients.

– 386 –

CORNERSTONE INVESTORS

Eastspring, acting as the discretionary investment manager for and on behalf of two discretionary funds (the "ESI Managed Funds"), has agreed to participate in the Global Offering and for such ESI Managed Funds to invest as Cornerstone Investor. The ESI Managed Funds comprise an open-end mutual fund (namely EASTSPRING INVESTMENTS — ASIA OPPORTUNITIES EQUITY FUND) and a segregated mandate (namely AHAPAG — ASIA PACIFIC ACTIVE GROWTH EQUITY PORTFOLIO) established under various jurisdictions and have multiple holders, who together with their ultimate beneficial owners are, to the best of the knowledge, information and belief of the Company, Independent Third Parties. The only ultimate beneficial owner for each of EASTSPRING INVESTMENTS — ASIA OPPORTUNITIES EQUITY FUND and AHAPAG — ASIA PACIFIC ACTIVE GROWTH EQUITY PORTFOLIO is Prudential plc.

**E Fund Management**

E Fund Management Co., Ltd. ("E Fund Management") is a leading comprehensive asset management company in the PRC. E Fund Management is a QDII approved by the relevant PRC authority and targets at companies with competitive edge over its competitors. E Fund Management is a fund manager managing assets on behalf of its underlying clients. The shareholders of E Fund Management include (1) Guangdong Finance Trust Co., Ltd. (广东粤财信托有限公司), which is ultimately owned by The People's Government of Guangzhou Municipality (广东省人民政府), (2) GF Securities Co., Ltd. (广发证券股份有限公司) ("GF Securities"), which is listed on the Stock Exchange (stock code: 1776) and the Shenzhen Stock Exchange (stock code: 000776), and (3) Infore Group Co., Ltd (盈峰集团有限公司), which is ultimately owned by He Jianfeng (何剑锋), each holding 22.65% in E Fund Management and an Independent Third Party. None of the remaining shareholders of E Fund Management owns 30% or more equity interest therein.

**IDG Breyer Fund**

IDG Breyer Capital Fund L.P. ("IDG Breyer Fund") is an exempted limited partnership established under the laws of the Cayman Islands. IDG Breyer Fund is a capital fund with a primary purpose of making equity and equity-related investments in the next-generation technology and technology-driven sectors, including, without limitation, artificial intelligence, autonomous driving, intelligent manufacturing, genome technology, fintech and 5G enabled next generation cloud services. It is ultimately controlled by Chi Sing HO and Fei YANG, both being Independent Third Parties. The investor holding 30% or more stake in IDG Breyer Fund is a listed company after equity penetration; save as disclosed above, none of the other ultimate beneficial owners of IDG Breyer Fund is interested in it as to 30% or more.

**Janchor Funds**

Janchor Partners Pan-Asian Master Fund and Janchor Partners Opportunities Master Fund III (together "Janchor Funds") are investment funds established in the Cayman Islands. Janchor Partners Limited ("Janchor Partners") serves as investment manager of the Janchor

– 387 –

CORNERSTONE INVESTORS

Funds. Established in 2009, Janchor Partners is a long-term industrialist investor, partnering with companies that have superior business models, favourable growth prospects and the potential to be part of long-term positive structural dynamics of Asian countries and economies.

Janchor Partners is licensed by the SFC to conduct asset management and is an experienced institutional investor with a track record of investing in technology companies. None of the participating shareholders or limited partners of the Janchor Funds or their feeder funds holds an interest of 30% or more of the Janchor Funds' total capital.

Martis Fund, L.P.

Martis Fund, L.P. 是一家依据开曼群岛法律注册成立的豁免有限合伙企业，专注于医疗健康、电信、媒体、科技及消费行业的投资。Martis Fund, L.P. 的普通合伙人为 Pulsating Star GP Limited，该公司由李先生（Mr. Eric Li）最终全资控制，其为本公司的独立第三方。Martis Fund, L.P. 中没有任何有限合伙人持有30%或以上的合伙权益。李先生（Mr. Eric Li）为香港公民，在投资行业拥有丰富经验。通过其最终控制的多只投资基金，李先生专注于医疗健康、电信、媒体、科技及消费行业的投资，并已成功投资多家在香港上市的公司，包括：作为IPO前投资者投资于巨子生物（股票代码：02367）、味丰天地（股票代码：09985）及顺丰同城（股票代码：09699），以及作为基础投资者投资于古茗（股票代码：01364）、三花智控（股票代码：02050）、奇瑞汽车（股票代码：09973）及中国工业气体（股票代码：06166）。

Mirae Asset Securities（未来资产证券）

Mirae Asset Securities Co., Ltd.（"未来资产证券"）是大韩民国规模最大的投资银行之一，提供全面的金融服务，包括经纪、财富管理、投资银行、销售与交易及自营投资。其最终由大韩民国金融投资公司 Mirae Asset Capital Co., Ltd. 控制。未来资产证券在韩国证券交易所上市，股票代码为 006800.KS。

MPC VII

MPC VII Pte. Ltd.（"MPC VII"）是一家在新加坡注册成立并以新加坡为主要营业地的有限公司，分别由 MPC VII L.P. 及 MPC VII-A L.P. 持有93.97%及6.03%的权益。MPC VII L.P. 及 MPC VII-A L.P. 均为依据开曼群岛法律注册成立的豁免有限合伙企业，两者的普通合伙人均为 MPC Management VII L.P.。MPC Management VII L.P. 的普通合伙人为 MPC GPGP VII Ltd.。David Su 为 MPC GPGP VII Ltd. 的控股股东。MPC VII L.P. 及 MPC VII-A L.P. 中均无单一有限合伙人持有30%或以上的权益。就 MPC VII 所知，David Su 为独立第三方。

– 388 –

CORNERSTONE INVESTORS

Perseverance Asset Management（毅持资产管理）

Perseverance Asset Management International (Singapore) Pte. Ltd.（"毅持资产管理"）以全权委托方式担任不超过六只投资基金及/或独立管理账户（统称"毅持基金"）的投资顾问或投资管理人。毅持基金中每只基金均无单一最终实益拥有人持有30%或以上的权益。毅持资产管理为一家于2018年10月在新加坡注册成立的私人有限公司，持有新加坡金融管理局颁发的基金管理资本市场服务牌照。毅持资产管理由 Perseverance Asset Management International 全资拥有，后者主要从事投资管理及投资顾问服务，为独立第三方。毅持资产管理担任投资顾问或投资管理人的若干投资基金曾作为基础投资者投资于紫金黄金国际有限公司（ZIJIN GOLD INTERNATIONAL COMPANY LIMITED）（股票代码：2259.HK）、宁德时代新能源科技股份有限公司（Contemporary Amperex Technology Co., Limited）（股票代码：3750.HK）及先瑞达医疗科技控股有限公司（Acotec Scientific Holdings Limited）（股票代码：6669.HK）。毅持资产管理以投资顾问或投资管理人的身份代表毅持基金与本公司订立基础投资协议。

Taikang Life（泰康人寿）

泰康人寿保险股份有限公司（"泰康人寿"）是一家在中国注册成立的公司，为泰康保险集团股份有限公司的全资子公司。泰康保险集团股份有限公司中没有任何股东持有30%或以上的股份。泰康人寿为个人及家庭提供全面的个人保障及投资理财产品与服务。所提供的产品对应客户在儿童及青少年、女性及高收入人群等细分市场的不同需求，同时满足医疗健康与意外保障、养老及财富管理等多维度需求。泰康保险集团股份有限公司是一家以保险、资产管理及健康养老为主营业务的保险金融服务集团。该集团总部位于北京，旗下拥有多家子公司，包括泰康人寿、泰康资产、泰康养老、泰康健投、泰康健康及泰康在线。其产品覆盖人寿保险、互联网金融保险、企业年金、资产管理、健康养老、健康管理及商业地产等领域。

– 389 –

CORNERSTONE INVESTORS

交割条件

各基础投资者根据各自基础投资协议认购发售股份的义务须满足（除其他条件外）以下交割条件：

(i)

承销协议已订立并已生效且无条件（按照各自的原始条款，或经各方事后放弃或协议变更）的时间及日期不迟于承销协议中规定的时间及日期，且承销协议均未被终止；

（ii）要约价格已根据承销协议及各方就全球发售签订的价格确定协议达成一致；

（iii）上市委员会已批准A类普通股（包括投资者股份及其他适用的豁免及批准）的上市及买卖，且该等批准、许可或豁免在A类普通股于联交所开始买卖前未被撤销；

（iv）任何政府机构均未颁布或公布任何法律，禁止完成全球发售或基石投资协议所拟进行的交易，且不存在任何有管辖权的法院所作出的、阻止或禁止完成该等交易的命令或禁令；以及

（v）相关基石投资者根据相关基石投资协议所作出的各项陈述、保证、承诺及确认在所有方面均属且将属准确及真实，并无误导性，且相关基石投资者在基石投资协议方面不存在任何重大违约。

基石投资者的限制

各基石投资者已同意，在上市日期起计（包括上市日期）六个月的期间内（「禁售期」），无论直接或间接，均不会出售其根据相关基石投资协议所购买的任何发售股份，惟若干有限情况除外，例如转让予其任何全资附属公司，而该等全资附属公司须受该基石投资者相同义务的约束，包括禁售期限制。

－390－

AUTHORIZED AND ISSUED SHARE CAPITAL

The following is a description of the authorized and issued share capital of our Company in issue and to be issued as fully paid or credited as fully paid upon Listing, assuming the Presumptions.

Share capital as of the date of this Prospectus

(i) Authorized share capital

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 221,311,196 | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$22,131.1196 | | 106,650,075 | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$10,665.0075 | | 172,038,729 | Preferred Shares with a nominal value of US$0.0001 each in issue | US$17,203.8729 | | 500,000,000 | Total | US$50,000 |

(ii) Issued and to be issued, fully paid or credited to be fully paid

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 22,890,736 (1) | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$2,289.0736 | | 85,759,339 (2) | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$8,575.9339 | | 171,407,993 | Preferred Shares with a nominal value of US$0.0001 each in issue | US$17,140.7993 | | 280,058,068 | Total | US$28,005.8068 |

Notes:

(1) representing 343,195 Class A Ordinary Shares, 20,890,736 Class A Ordinary Shares and 1,656,805 Class A Ordinary Shares held by Alpha EXP, MiniMax Gene and Himalia Holding Limited, respectively, as of the date of this Prospectus.

(2) representing 15 Class B Ordinary Shares, 5,000,000 Class B Ordinary Shares, 11,509,339 Class B Ordinary Shares, 62,249,985 Class B Ordinary Shares and 7,000,000 Class B Ordinary Shares held by MiniMax Limited, MiniMax Matrix, MiniMax Awakening, Alpha EXP and Floating Sky, respectively, as of the date of this Prospectus.

– 391 –

SHARE CAPITAL

Share capital immediately following the completion of the Global Offering

(i) Authorized share capital

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 393,349,925 | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$39,334.9925 | | 106,650,075 | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$10,665.0075 | | 500,000,000 | Total | US$50,000 |

(ii) Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised)

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 22,547,541 (1) | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$2,254.7541 | | 80,759,339 (2) | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$8,075.9339 | | 343,195 (3) | Class A Ordinary Shares to be converted into Class B Ordinary Shares | US$34.3195 | | 5,000,000 (4) | Class B Ordinary Shares to be converted into Class A Ordinary Shares | US$500 | | 171,407,993 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued on conversion of Preferred Shares | US$17,140.7993 | | 25,389,220 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Global Offering | US$2,538.9220 | | 305,447,288 | Total | US$30,544.7288 |

– 392 –

SHARE CAPITAL

Notes: (1)

representing 20,890,736 Class A Ordinary Shares and 1,656,805 Class A Ordinary Shares held by MiniMax Gene and Himalia Holding Limited, respectively, upon Listing.

(2)

representing 15 Class B Ordinary Shares, 11,509,339 Class B Ordinary Shares, 62,249,985 Class B Ordinary Shares and 7,000,000 Class B Ordinary Shares held by MiniMax Limited, MiniMax Awakening, Alpha EXP and Floating Sky, respectively, upon Listing.

(3)

representing 343,195 Class A Ordinary Shares held by Alpha EXP to be converted into Class B Ordinary Shares upon Listing.

(4)

representing 5,000,000 Class B Ordinary Shares held by MiniMax Matrix to be converted into Class A Ordinary Shares upon Listing.

(iii) Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option is fully exercised and the Over-allotment Option is not exercised)

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 22,547,541(1) | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$2,254.7541 | | 80,759,339(2) | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$8,075.9339 | | 343,195(3) | Class A Ordinary Shares to be converted into Class B Ordinary Shares | US$34.3195 | | 5,000,000(4) | Class B Ordinary Shares to be converted into Class A Ordinary Shares | US$500 | | 171,407,993 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued on conversion of Preferred Shares | US$17,140.7993 | | 25,389,220 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Global Offering | US$2,538.9220 | | 3,808,380 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Offer Size Adjustment Option | US$380.8380 | | 309,255,668 | Total | US$30,925.5668 |

Note:

please refer to the section headed "— (ii) Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised)" above.

– 393 –

SHARE CAPITAL (iv) Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option is not exercised and the Over-allotment Option is fully exercised)

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 22,547,541 (1) | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$2,254.7541 | | 80,759,339 (2) | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$8,075.9339 | | 343,195 (3) | Class A Ordinary Shares to be converted into Class B Ordinary Shares | US$34.3195 | | 5,000,000 (4) | Class B Ordinary Shares to be converted into Class A Ordinary Shares | US$500 | | 171,407,993 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued on conversion of Preferred Shares | US$17,140.7993 | | 25,389,220 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Global Offering | US$2,538.9220 | | 3,808,380 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Over-allotment Option | US$380.8380 | | 309,255,668 | Total | US$30,925.5668 |

Note:

please refer to the section headed "— (ii) Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised)" above.

– 394 –

SHARE CAPITAL (v)

Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option and the Over-allotment Option are fully exercised)

| Number | Description of Shares | Aggregate Nominal Value | |---|---|---| | 22,547,541 (1) | Class A Ordinary Shares with a nominal value of US$0.0001 each in issue | US$2,254.7541 | | 80,759,339 (2) | Class B Ordinary Shares with a nominal value of US$0.0001 each in issue | US$8,075.9339 | | 343,195 (3) | Class A Ordinary Shares to be converted into Class B Ordinary Shares | US$34.3195 | | 5,000,000 (4) | Class B Ordinary Shares to be converted into Class A Ordinary Shares | US$500 | | 171,407,993 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued on conversion of Preferred Shares | US$17,140.7993 | | 25,389,220 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Global Offering | US$2,538.9220 | | 3,808,380 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Offer Size Adjustment Option | US$380.8380 | | 4,379,640 | Class A Ordinary Shares with a nominal value of US$0.0001 to be issued pursuant to the Over-allotment Option | US$437.9640 | | 313,635,308 | Total | US$31,363.5308 |

Note:

please refer to the section headed "— (ii) Issued and to be issued, fully paid or credited to be fully paid (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised)" above.

– 395 –

股本

加权投票权架构

本公司设有加权投票权架构。根据本公司的加权投票权架构，本公司股本由A类普通股及B类普通股组成。就本公司股东大会上须予表决的任何事项，每股B类普通股赋予持有人行使十票的投票权，每股A类普通股赋予持有人行使一票的投票权，但须遵守《上市规则》第8A.24条关于保留事项须按每股一票基准进行表决的规定。

保留事项为：

(i) 对本备忘录及章程细则的任何修订；

(ii) 对任何类别股份所附权利的变更；

(iii) 任何独立非执行董事的委任、选举或罢免；

(iv) 本公司核数师的委任或罢免；及

(v) 本公司的自愿清盘或结业清理。

详情请参阅本招股说明书附录三"本公司章程及开曼群岛公司法概要——2 章程细则"。

B类普通股可按一比一的基准转换为A类普通股。于所有已发行及流通B类普通股悉数转换为A类普通股后，本公司将发行81,102,534股A类普通股，占紧随上市后（假设发行规模调整选项及超额配股权均未获行使）已发行A类普通股总数约26.55%。

根据《上市规则》第8A.22条，当加权投票权受益人不再实益拥有本公司任何B类普通股时，附于本公司B类普通股的加权投票权将告终止。此情况可能于以下情形发生：

(i) 出现《上市规则》第8A.17条所列任何情形时，尤其是加权投票权受益人：(1) 身故；(2) 不再担任本公司董事会成员；(3) 被联交所认定为丧失行使其董事职责的行为能力；或(4) 被联交所认定为不再符合《上市规则》所订明的董事资格要求；

(ii) 当B类普通股持有人已将其对B类普通股的实益拥有权或经济利益，或对其所附投票权的控制权转让予另一人，惟根据《上市规则》第8A.18条所允许的情形除外；

– 396 –

SHARE CAPITAL

(iii) where a vehicle holding Class B Ordinary Shares on behalf of a WVR Beneficiary no longer complies with Rule 8A.18(2) of the Listing Rules; or

(iv) when all of the Class B Ordinary Shares have been converted to Class A Ordinary Shares.

Shareholding Structure of the WVR Beneficiaries

The table below sets out the beneficial interests entitled to and voting rights to be held by the WVR Beneficiaries upon the completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised):

| | Number of Class B Ordinary Shares held | Number of Class A Ordinary Shares interested in (3) | Approximate percentage of beneficial interests in the issued share capital | Approximate percentage of voting rights controlled (1) | |---|---|---|---|---| | Dr. Yan (2) | 74,102,534 | 3,355,030 | 25.36% | 72.05% | | Ms. Yun (2) | 7,000,000 | 1,644,970 | 2.83% | 6.76% |

Notes:

(1) On the basis that each Class A Ordinary Share entitles the Shareholder to one vote per Share and each Class B Ordinary Share entitles the Shareholder to ten votes per Share.

(2) For details of the shareholding structure of our WVR Beneficiaries, please refer to the section headed "History, Reorganization and Corporate Structure."

(3) Dr. Yan and Ms. Yun are interested in MiniMax Matrix as to 67.1% and 32.9%.

The Company confirms that the holding arrangement through which the WVR Beneficiaries hold the Class B Ordinary Shares as described above meets the requirements in Rule 8A.18 of the Listing Rules and the holding arrangement is permitted under the "Consultation Conclusions — a listing regime for companies from emerging and innovative sectors" issued by the Stock Exchange in April 2018, namely: (a) a partnership of which the WVR Beneficiary is a partner and the terms of which must expressly specify that the voting rights attached to any and all of the Class B Ordinary Shares held by such partnership are solely dictated by the WVR Beneficiary; (b) a trust of which the WVR Beneficiary is a beneficiary and that meets the following conditions: (i) the WVR Beneficiary must in substance retain an element of control of the trust and any immediate holding companies of, or, if not permitted in the relevant tax jurisdiction, retain a beneficial interest in any and all of the Class B Ordinary Shares held by such trust; and (ii) the purpose of the trust must be for estate planning and/or tax planning purposes; or (c) a private company or other vehicle wholly owned and wholly controlled by the WVR Beneficiary or by a trust referred to in paragraph (b) above.

– 397 –

SHARE CAPITAL

To ensure that there will not be any circumvention of Rule 8A.18(1), each of the Company, Dr. Yan and Ms. Yun undertakes that so long there is any weighted voting rights attached to the Shares held by Alpha EXP, MiniMax Gene, Floating Sky, MiniMax Awakening, MiniMax Limited (the "WVR Management Shareholders"), respectively, Dr. Yan and Ms. Yun will not transfer any beneficial ownership of or economic interest in the WVR Management Shareholders or the control over the voting rights attached to the Shares held by WVR Management Shareholders to another person. In the event that there is any change in the beneficial ownership of or economic interest in the Shares held by the WVR Management Shareholders or the control over the voting rights attached to the Shares held by the WVR Management Shareholders to another person, the Company, Dr. Yan and Ms. Yun will notify the Stock Exchange pursuant to Rule 8A.19 of the Listing Rules and comply with the relevant statutory obligations including obligations of disclosure of interests under the SFO, and the weighted voting rights attached to the Class B Ordinary Shares held by WVR Management Shareholders shall cease upon such transfer accordingly. The Company will also comply with Rule 8A.30 of the Listing Rules to confirm, on an annual basis, that the WVR Beneficiary has complied with Rule 8A.18 of the Listing Rules.

Contribution of the WVR Beneficiaries

Dr. Yan and Ms. Yun, being the WVR beneficiaries, have been materially responsible for the growth of the Company's business during the Track Record Period by way of their respective skills, knowledge and/or insights to the industry. As the core of the Group's leadership team and leveraging their professional experience in the industry, each of Dr. Yan and Ms. Yun is pivotal to the success of the Group and has made significant contributions to the Group from strategic, technological and operational perspectives.

We set forth below the academic background, work experience and contribution of the proposed WVR beneficiaries to the success of the Company:

Dr. Yan

Dr. Yan is the founder, the chairman of the board of directors, chief executive officer and chief technology officer of the Company. As the chief executive officer and chief technology officer of the Company, Dr. Yan has been integral to the success of the Company and has been materially responsible for the founding and growth of the Company during the Track Record Period. Dr. Yan, with profound technical insight and deep understanding and knowledge of AI technology, laid the foundation for the Company and was critical in shaping the Group's mission, vision and values, and devising long-term strategies for the R&D and operations of the Group over the years. During the Track Record Period, Dr. Yan had spearheaded the team in developing a trimodal large model that integrates text, audio, and visual capabilities and have led our Group to achieve its key milestones. For example, he led the launch of our first text model abab1 in 2022, our text model abab5.5 and speech model MiniMax-Speech-01 in 2023, our MoE text model abab6, visual generation platform Hailuo AI and video-generation model Hailuo-01 and music model Music-01 in 2024 as well as our open-source text model MiniMax-Text-01, MiniMax-M1 and MiniMax-M2 in 2025. In addition, leveraging the – 398 –

SHARE CAPITAL

reputation and experience of Dr. Yan in the industry, the Company has been able to secure investments from numerous investors at a relatively early stage and before its significant commercialization. Dr. Yan has also led the Company to develop a suite of AI-native products that serve a broad range of user scenarios and lay a solid foundation for the Company's path to commercialization.

Ms. Yun

As the chief operating officer of the Company, Ms. Yun has been integral to the success of the Company and has been materially responsible for the founding and growth of the Company's business and guiding its development since she co-founded the Group. With profound industry insight and deep understanding and knowledge of AI technology as well as her global vision, Ms. Yun led the rapid global business expansion of the Company and was critical in shaping the Group's mission, vision and values. She is also instrumental in devising long-term R&D strategies, product development, commercialization and operations of the Group over the years. With the deep insights in the global AI industry, she played a pivotal role in designing and developing the Company's foundation models and devising long-term strategies and R&D focuses for the Company, with an emphasis on developing AI-native products. In particular, leveraging her expertise and knowledge of the AI ecosystem, Ms. Yun is able to envisage market needs and drive the Company's creative product development efforts and create AI-native offerings products that serve a broad range of user scenarios. For example, for individual users, the Company has launched (i) MiniMax, its intelligent chat agent application, (ii) Hailuo AI, its flagship artificial intelligence visual generation platform, and (iii) Xingye/Talkie AI-powered Multi-modal Entertainment Platform. For enterprise customers, the Company offers an open platform, which provides API access to its self-developed multimodal models. As the chief operating officer, Ms. Yun led the day-to-day operation of the Company and coordinated major matters of the Company, including but not limited to corporate strategy, overall operational management, corporate governance and investor relationship. Under the leadership of Ms. Yun, the Company was able to overcome challenges and has achieved rapid growth in the past few years. Since the Company's inception, Ms. Yun has been overseeing talent acquisition and human resources functions of the Company, playing a pivotal role in establishing a competitive organization from the ground up. Leveraging her expertise in investment and financing, Ms. Yun is able to secure investments for the Company from numerous investors at a relatively early stage and before its significant commercialization. With a global vision, Ms. Yun has led the Company to achieve its rapid global expansion. During the early stages of the Company's development, Ms. Yun has been instrumental in building the Company's department focusing on products commercialization from the ground up. These strategic initiatives have laid the groundwork for the Company's forthcoming technological innovations and accelerated global expansion. With the contribution from Ms. Yun, the Company's open platform currently provides scalable and customizable AI services to enterprise customers across more than 100 countries and regions, and is one of the top-ranking open platforms in Asia in terms of daily token volume. The Company's service has reached more than 100 thousand registered enterprise customers and developers, including a range of well-known enterprise customers.

– 399 –

股本

排名

发售股份在各方面与本招股说明书中提及的所有现已发行或将予发行的A类普通股享有同等权利，并有资格就本招股说明书日期之后的记录日期所宣派、作出或支付的所有股息或其他分派，与该等股份平等参与。

加权投票权受益人的承诺

根据上市规则第8A.43条，每位加权投票权受益人须向本公司作出具有法律约束力的承诺，承诺其将遵守第8A.43条所载的相关规定，该等规定旨在为股东的利益而设，并可由股东强制执行。于2025年12月22日，严博士及云女士各自向本公司作出承诺（"承诺"），表明在其为加权投票权受益人期间：

(a) 其将遵守（若其实益拥有加权投票权所附股份系透过有限合伙、信托、私人公司或其他载体持有，则将尽最大努力促使该有限合伙、信托、私人公司或其他载体遵守）上市规则第8A.09、8A.14、8A.15、8A.17、8A.18及8A.24条不时生效的所有适用规定（"规定"）；及

(b) 其将尽最大努力促使本公司遵守所有适用规定。

为免生疑问，该等规定须受上市规则第2.04条规限。加权投票权受益人确认并同意，股东在购买及持有其股份时依赖该等承诺。加权投票权受益人确认并同意，该承诺旨在赋予本公司及全体股东利益，并可由本公司及╱或任何股东对加权投票权受益人强制执行。

承诺须于以下较早者发生时自动终止：(i) 本公司从联交所退市之日；及(ii) 相关加权投票权受益人不再为本公司加权投票权受益人之日。为免生疑问，承诺的终止不影响本公司及╱或任何股东及╱或加权投票权受益人本人于终止日期之前已累积的任何权利、补救措施、义务或责任，包括就于终止日期当日或之前存在的任何违反承诺行为索赔损害赔偿及╱或申请任何禁令的权利。

承诺受香港法律管辖，所有因承诺而产生或与之相关的事宜、索赔或争议须受香港法院专属司法管辖权约束。

– 400 –

股本的潜在变动

须召开股东大会的情形

根据开曼群岛公司法及章程细则的条款，本公司可不时通过股东普通决议（i）增加股本；（ii）合并股本并将其划分为较大面值的股份；（iii）将股份划分为数个类别；及（iv）注销任何尚未被认购或同意认购的股份。此外，本公司可在符合开曼群岛公司法规定的前提下，由股东通过特别决议削减股本或资本赎回储备。详情请参阅本招股说明书附录三"本公司章程及开曼群岛公司法概要——2. 章程细则——2.5 股本变更"。

（i）发行股份及（ii）出售及/或转让库存股的一般授权

在全球发售成为无条件的前提下，本公司董事获授一般授权，以（i）配发、发行及处理任何A类普通股或可转换为股份的证券，及（ii）出售及/或转让作为库存股持有的股份，数量不超过以下两项之和：

• 全球发售完成后即时已发行股份总数的20%（不包括（i）根据行使超额配股权可能发行的额外A类普通股，（ii）根据上市后股份激励计划拟发行的A类普通股，（iii）可因B类普通股转换而发行的A类普通股，及（iv）库存股（如有））；及

• 本公司根据本节"——回购股份的一般授权"段所述授权回购的股份面值总额。

此项发行A类普通股及出售及/或转让库存股的一般授权将于以下最早发生者届满时失效：

• 本公司下届股东周年大会结束时，除非在股东大会上由股东以普通决议续期，无论是否附带条件；或

• 本公司章程细则或任何其他适用法律规定须召开下届股东周年大会的期限届满时；或

• 在股东大会上由股东以普通决议对该授权作出更改或予以撤销之日。

— 401 —

SHARE CAPITAL General mandate to repurchase shares Subject to the Global Offering becoming unconditional, our Directors have been granted a general unconditional mandate, to exercise all the powers of our Company to repurchase our own securities with nominal value of up to 10% of the total number of Shares in issue immediately following the completion of the Global Offering (excluding (i) the additional Class A Ordinary Shares which may be issued pursuant to the exercise of the Over-allotment Option, (ii) the Class A Ordinary Shares to be issued pursuant to the Post-IPO Share Incentive Plan, (iii) the Class A Ordinary Shares that are issuable upon conversion of the Class B Ordinary Shares, and (iv) treasury shares, if any). The repurchase mandate only relates to repurchases made on the Stock Exchange, or on any other stock exchange on which our Shares are listed (and which are recognized by the SFC and the Stock Exchange for this purpose), and which are in accordance with the Listing Rules. A summary of the relevant Listing Rules is set out in "Statutory and General Information — A. Further Information about our Group — 5. Repurchases of Our Own Securities" in Appendix IV. This general mandate to repurchase Shares will expire at the earliest of: •

the conclusion of the next annual general meeting of our Company unless otherwise renewed by an ordinary resolution of our Shareholders in a general meeting, either unconditionally or subject to conditions; or

•

the expiration of the period within which our Company's next annual general meeting is required by Articles of Association or any other applicable laws to be held; or

•

the date on which it is varied or revoked by an ordinary resolution of our Shareholders in a general meeting.

See "Statutory and General Information — A. Further Information about Our Group — 4. Resolutions of Our Shareholders" in Appendix IV to this Prospectus for further details of the repurchase mandate. SHARE INCENTIVE PLAN The Company has adopted the Pre-IPO Share Incentive Plan and the Post-IPO Share Incentive Plan. See "Statutory and General Information — D. Share Incentive Plans" in Appendix IV to this Prospectus for further details.

– 402 –

FINANCIAL INFORMATION

You should read the following discussion and analysis in conjunction with our consolidated financial statements and the accompanying notes included in the Accountants' Report set forth in Appendix I to this Prospectus. Our consolidated financial statements have been prepared in accordance with IFRSs, which may differ in material aspects from generally accepted accounting principles in other jurisdictions. You should read the entire Accountants' Report and not merely rely on the information contained in this section.

The following discussion and analysis contain forward-looking statements that reflect the current views with respect to future events and financial performance. These statements are based on assumptions and analysis made by us in light of our experience and perception of historical trends, current conditions and expected future developments, as well as other factors that we believe are appropriate under the circumstances. However, whether the actual outcome and developments will meet our expectations and predictions depends on a number of risks and uncertainties over which we do not have control. In evaluating our business, you should carefully consider all of the information provided in this Prospectus.

For the purpose of this section, unless the context otherwise requires, references to 2022, 2023 and 2024 refer to our financial year ended December 31 of such year. Unless the context otherwise requires, financial information described in this section is described on a consolidated basis.

OVERVIEW

MiniMax is a global AI foundation model company. Founded by a group of forward-thinking engineers, we are committed to driving AI innovation towards performing the full range of human intellectual tasks, from learning and reasoning to planning and generalizing knowledge across diverse domains.

We have been consistently iterating our models to higher intelligence levels. Today, our proprietary foundation model suite, led by MiniMax-M2, Hailuo-02, and Speech-02, has long context processing capacity and can understand, generate, and integrate a wide range of modalities, including text, video and audio. These models power our major AI-native products — including MiniMax, Hailuo AI, MiniMax Audio, Talkie/Xingye, and our enterprise and developer-facing Open Platform, delivering intelligent and dynamic experiences to users globally.

– 403 –

财务信息

报告编制基础

本历史财务信息乃根据国际会计准则委员会（"IASB"）颁布的所有适用国际财务报告准则会计准则编制。所采纳的重要会计政策信息的进一步详情载于本招股说明书附录一所载会计师报告附注2。

IASB已颁布多项新订及经修订的国际财务报告准则会计准则。为编制本历史财务信息，本集团已在整个往绩记录期间采纳所有适用的新订及经修订国际财务报告准则会计准则，惟任何于2025年1月1日起开始的会计期间尚未生效的新准则或诠释除外。已颁布但于2025年1月1日起开始的会计期间尚未生效的经修订及新订会计准则及诠释，已载于本招股说明书附录一所载会计师报告附注2。

影响本公司经营业绩的主要因素

本公司的业务及经营业绩受到各种影响基础模型及人工智能原生产品整体终端用户需求及市场状况的一般因素所影响。该等因素包括宏观经济趋势、行业动态、技术进步与创新、人工智能在各行业的渗透速度及用户适应性、政府政策及法规（包括专门针对人工智能的政策法规），以及竞争格局。上述任何条件的不利变化均可能对本公司的经营业绩产生不利影响。

除上述一般因素外，以下特定因素对本公司经营业绩的影响更为直接。

维持模型智能技术领先地位的能力

本公司基础模型的智能水平、性能及竞争力是影响本公司业务及经营业绩的最关键因素。上述竞争力取决于本公司对技术发展趋势的判断、持续创新的能力以及模型研发流程的效率。具体而言，本公司模型的智能水平及实力直接影响本公司产品的用户采纳率、市场需求、产品渗透率及定价，进而影响本公司的收入增长及盈利能力。

— 404 —

财务信息

多元化产品供应、拓宽变现渠道及提升可及性的能力

我们的收入增长主要得益于人工智能原生产品矩阵的快速扩张以及多元化的变现策略。自成立以来，我们遵循以产品为导向的增长路径，依托自研基础模型开发智能化、场景驱动型应用程序。我们多元化的产品供应服务于个人用户和企业客户，覆盖广泛的应用场景，包括视频生成、通用智能体服务、语音与音乐合成以及多模态对话界面。随着越来越多的用户采用我们的旗舰产品——包括MiniMax、海螺AI、MiniMax Audio、Talkie/星野——我们已累计服务超过2.12亿个人用户，覆盖200多个国家和地区，以及超过10万家企业客户和开发者，覆盖100多个国家和地区。

我们已建立多元化的变现渠道，包括订阅服务、基于代币的应用内购买、在线营销服务以及基于用量的企业级API。对于面向消费者的产品，如MiniMax、海螺AI、MiniMax Audio及Talkie/星野，变现主要通过高级功能订阅服务、基于代币的应用内购买及在线营销服务实现。对于我们的开放平台，变现主要来源于按代币量计费的API调用（即用户请求调用本公司模型）。

我们已实现用户增长及早期变现成效。我们的平均月活跃用户数（MAU）从2023年的约310万增长至2024年的约1,910万，并进一步增长至截至2025年9月30日止九个月的约2,760万。我们人工智能原生产品的付费用户数从2023年的约119,700人增长至2024年的约650,300人，并进一步增长至截至2025年9月30日止九个月的约1,771,600人。展望未来，我们计划拓宽变现渠道并提升各产品线的每用户收入。具体而言，我们旨在丰富面向消费者产品的增值功能，并强化API分层体系，以支持各行业的大规模使用场景。我们还计划深化模型与产品之间的整合，从而在各模态中释放更具规模化的商业机遇。

拓展全球用户基础及提升品牌知名度的能力

我们的全球化战略支持我们在各市场同步推出产品，并推动了国际业务的快速增长。截至2025年9月30日，我们的产品和服务已部署于200多个国家和地区，且在整个业绩记录期内，来自国际市场的收入在我们总收入中占据重要比例。截至2025年9月30日止九个月，来自中国大陆以外地区的收入约占我们总收入的73.1%。

我们相信，进一步增长是高竞争力产品与高品牌知名度的自然结果，而品牌知名度又取决于模型智能能力的持续提升。随着我们不断提升模型智能水平，我们预期将以有机用户获取方式扩大全球用户基础，而无需依赖大规模品牌推广及用户获取投入。

— 405 —

FINANCIAL INFORMATION

通过提升模型计算效率来优化成本的能力

我们认为，在支持日益复杂的人工智能模型的同时提升模型计算效率的能力，是实现盈利能力的关键驱动因素。在业绩记录期间，与模型推理活动相关的成本被记录在销售成本项下与推理活动相关的云服务成本中。在业绩记录期的每一年，上述成本均占我们销售成本总额的90.0%以上。

通过在模型架构和基础设施方面的持续创新，我们提升了与推理活动相关的成本效率。我们通过动态资源分配和统一的训练推理框架，保持较高的算力利用率。我们的专有人工智能基础设施能够动态分配计算资源，确保服务可用性，并支持高性能基础模型的大规模可持续交付。就占收入的百分比而言，我们的销售成本从2023年的124.7%下降至2024年的87.8%，并进一步从截至2024年9月30日止九个月的97.4%下降至截至2025年9月30日止九个月的76.7%。这一改善反映了推理效率的提升以及因更智能化的模型和更高的基础设施利用率所带来的规模经济效益。

我们认为，有效训练人工智能模型对我们的长期成功至关重要。与模型训练、微调及实验相关的成本被确认为研究与开发费用。为提升训练效率，我们组建了内部人工智能基础设施团队，并独立开发了一套专为大规模第三方计算集群量身定制的高性能训练框架。我们的人工智能基础设施采用整体化设计——从算子层面到跨集群资源调度——从而实现模型训练的高效执行。

尽管研发仍是我们最大的投入领域，但与训练相关的云服务费用占收入的百分比有所下降，从2023年的逾1,300%降至2024年的460.8%，并进一步从截至2024年9月30日止九个月的530.0%降至截至2025年9月30日止九个月的266.5%。这一降幅反映了训练效率的提升及我们基础设施的可扩展性，也体现了我们的业务正从研究密集型开发阶段向规模化商业部署阶段转型。

持续提升研究开发及管理效率的能力

我们维持着一支精简、灵活的研究与开发团队及运作机制。我们的研究与开发团队在精简、扁平化且高度协同的组织架构下运作。为进一步提升研发及管理效率，我们正将自有模型技术整合至内部运营之中。这包括部署内部开发的大语言模型（LLM）智能体，以支持软件开发、实现工作流程自动化及处理常规任务。我们期望通过充分发挥自身的人工智能能力，提升人员效能，以更少的投入实现更大的产出。

– 406 –

财务信息 重要会计政策信息及估计

我们的某些会计政策要求我们对会计项目应用估计、假设及复杂判断。这些估计、假设及判断对我们的财务状况及经营业绩具有重大影响。我们的管理层根据过去经验、行业惯例及在当时情况下被认为合理的对未来事件的预期，持续评估此类估计、假设及判断。于业绩记录期间，我们管理层的估计或假设与实际结果之间不存在任何重大偏差，且我们未对上述估计或假设作出任何重大变更。我们预计在可见未来不会对上述估计及假设作出任何重大变更。

对于理解我们财务状况及经营业绩而言重要的重要会计政策信息、估计及判断，已在本招股说明书附录一所载会计师报告附注2及附注3中作进一步详细阐述。

以下列示我们认为对我们而言属重要或涉及编制财务报表所采用的最重大估计、假设及判断的会计政策。

收入确认

来自客户合同的收入

来自客户合同的收入于货物或服务的控制权转移至客户时予以确认，确认金额反映本集团预期就该等货物或服务而有权收取的代价。

(i) 人工智能原生产品收入

• 会员订阅

本集团向个人用户提供会员订阅服务，订阅会员可享有本集团人工智能原生产品高级功能的访问权限。会员订阅费须预先支付，且不予退还。收入按提供服务的会员期限内均匀确认。

– 407 –

FINANCIAL INFORMATION

•

Virtual items

Our Group also offers individual users with virtual items in its AI-native Products to enhance the using experience. Users have option to pre-purchase additional credits to recharge their accounts and buy these virtual items. For consumable virtual items, our Group's performance obligation is to provide one-off services to users. This performance obligation is satisfied when the virtual items are consumed. Accordingly, our Group recognises the revenues at the point in time. For non-consumable virtual items, our Group's performance obligation is to provide on-going services to users who purchased virtual items. This performance obligation is satisfied over the acting period of the paying users. Accordingly, our Group recognises the revenues ratably over the estimated average acting period of these paying users.

•

Online marketing service

In addition, our Group provides performance-based online marketing service to enterprise customers on certain of its AI-native applications, including through a mediation platform. Revenues from online marketing service are primarily recognised at a point in time when users view or click on the advertisement.

(ii) Revenue from Open Platform and other AI-based enterprise services

Our Group provides enterprise customers with access to its core AI models through its Open Platform. The performance obligation of such services is satisfied at a point in time when the customers call APIs with tokens. At the end of each month, the consideration is fixed based on tokens consumed and no variable consideration exists.

Our Group also provides enterprise customers with other AI-based enterprise services, mainly consists of arrangements customized to enterprise requirements and licensed deliverables. For customised arrangements, we work with enterprise customers to set up dedicated inference resource pools tailored to their needs, helping ensure stable and predictable model inference performance. For licensed deliverables, we license our foundation models to enable customers to deploy and operate such models in their own systems. Consideration for such services is fixed and revenue from other AI-based enterprise services is typically recognised at a point in time when the service is accepted by the customers.

Other income

Interest income is recognised on an accrual basis using the effective interest method by applying the rate that exactly discounts the estimated future cash receipts over the expected life of the financial instrument or a shorter period, when appropriate, to the net carrying amount of the financial asset.

– 408 –

FINANCIAL INFORMATION

合约负债

当本集团向客户转让相关商品或服务之前，收到客户付款或付款到期（以较早者为准）时，确认合约负债。当本集团履行合同义务（即将相关商品或服务的控制权转移给客户）时，合约负债确认为收入。

以股份为基础的付款

本公司设有购股权计划。本集团的雇员（包括董事）以股份支付的形式获得报酬，即雇员以提供服务换取权益工具（"以权益结算的交易"）。与雇员进行以权益结算的交易的成本，按授予日期的公平值计量。公平值由外部估值师采用二项式模型厘定，详情载于财务报表附注26。

以权益结算的交易成本于绩效条件及╱或服务条件获满足的期间内，连同权益的相应增加，在雇员福利开支中予以确认。于相关期间届满日止各期间末，就以权益结算的交易所确认的累计开支，反映归属期届满的程度及本集团对最终将归属的权益工具数量的最佳估计。当期损益表的支出或抵免，代表于该期间期初及期末所确认的累计开支的变动。

在釐定奖励的授予日公平值时，服务条件及非市场绩效条件不予考虑，惟有关条件获满足的可能性，将作为本集团对最终将归属的权益工具数量的最佳估计的一部分予以评估。附于奖励但无相关服务规定的任何其他条件，被视为非归属条件。非归属条件反映于奖励的公平值中，并导致即时将奖励列支为开支，除非同时存在服务条件及╱或绩效条件。

对于因未能满足非市场绩效条件及╱或服务条件而最终未能归属的奖励，不确认任何开支。

倘以权益结算的奖励条款经修改，则至少须按原条款获满足的情况确认开支，即按条款未经修改处理。此外，对于任何导致股份支付总公平值增加的修改，或于修改日计量对雇员而言在其他方面属有利的修改，须确认相关开支。倘以权益结算的奖励被取消，则视同于取消日归属处理，且就该奖励尚未确认的任何开支须即时予以确认。

– 409 –

FINANCIAL INFORMATION

Financial liabilities

Initial recognition and measurement

Financial liabilities are classified, at initial recognition, as financial liabilities at fair value through profit or loss, loans and borrowings, or payables, as appropriate.

All financial liabilities are recognised initially at fair value and, in the case of loans and borrowings and payables, net of directly attributable transaction costs.

Our Group's financial liabilities include trade and bills payables, other payables, accruals and other liabilities, convertible redeemable preferred shares, interest-bearing bank borrowings and lease liabilities.

Subsequent measurement

The subsequent measurement of financial liabilities depends on their classification as follows:

(i) Financial liabilities at amortised cost (trade and bills payables, other payables, accruals and other liabilities excluding convertible bonds, interest-bearing bank borrowings and lease liabilities)

After initial recognition, other payables and accruals, and lease liabilities are subsequently measured at amortised cost, using the effective interest rate method unless the effect of discounting would be immaterial, in which case they are stated at cost. Gains and losses are recognised in the statement of profit or loss when the liabilities are derecognised as well as through the effective interest rate amortisation process.

Amortised cost is calculated by taking into account any discount or premium on acquisition and fees or costs that are an integral part of the effective interest rate. The effective interest rate amortisation is included in finance costs in the statement of profit or loss.

(ii) Financial liabilities at fair value through profit or loss (convertible redeemable preferred shares and convertible bonds)

Financial liabilities at fair value through profit or loss include financial liabilities designated upon initial recognition as at fair value through profit or loss. The convertible redeemable preferred shares and convertible bonds issued by the Company were designated upon initial recognition at fair value through profit or loss. They are initially recognised at fair value. Any directly attributable transaction costs are recognised as finance costs in profit or loss. Gains or losses on them are recognised in the statements of profit or loss, except for the gains or losses arising from the Company's own credit risk which are presented in other comprehensive income with no subsequent reclassification to the statements of profit or loss. The net fair value gain or loss recognised in the statements of profit or loss does not include any interest charged on these financial liabilities.

– 410 –

FINANCIAL INFORMATION

Fair value measurement

Our Group measures its financial assets at fair value through profit or loss, financial assets at fair value through other comprehensive income, convertible redeemable preferred shares and convertible bonds at the end of each of the Relevant Periods. Fair value is the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date. The fair value measurement is based on the presumption that the transaction to sell the asset or transfer the liability takes place either in the principal market for the asset or liability, or in the absence of a principal market, in the most advantageous market for the asset or liability. The principal or the most advantageous market must be accessible by our Group. The fair value of an asset or a liability is measured using the assumptions that market participants would use when pricing the asset or liability, assuming that market participants act in their economic best interest.

A fair value measurement of a non-financial asset takes into account a market participant's ability to generate economic benefits by using the asset in its highest and best use or by selling it to another market participant that would use the asset in its highest and best use.

Our Group uses valuation techniques that are appropriate in the circumstances and for which sufficient data are available to measure fair value, maximising the use of relevant observable inputs and minimising the use of unobservable inputs.

All assets and liabilities for which fair value is measured or disclosed in the Historical Financial Information are categorised within the fair value hierarchy, described as follows, based on the lowest level input that is significant to the fair value measurement as a whole:

• Level 1 – based on quoted prices (unadjusted) in active markets for identical assets or liabilities

• Level 2 – based on valuation techniques for which the lowest level input that is significant to the fair value measurement is observable, either directly or indirectly

• Level 3 – based on valuation techniques for which the lowest level input that is significant to the fair value measurement is unobservable

For assets and liabilities that are recognised in the Historical Financial Information on a recurring basis, our Group determines whether transfers have occurred between levels in the hierarchy by reassessing categorisation (based on the lowest level input that is significant to the fair value measurement as a whole) at the end of each of the Relevant Periods.

– 411 –

FINANCIAL INFORMATION

CONSOLIDATED STATEMENTS OF PROFIT OR LOSS

The following table sets forth a summary of our consolidated statements of profit or loss, in absolute amounts and as a percentage of our total revenue, for the periods indicated.

| | For the year ended December 31, | | | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | |

| | 2022 US$ | % | 2023 US$ | % | 2024 US$ | % | 2024 (unaudited) US$ | % | 2025 US$ | % | |---|---|---|---|---|---|---|---|---|---|---| | Revenue | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 | | Cost of sales | – | – | (4,314) | (124.7) | (26,785) | (87.8) | (18,944) | (97.4) | (40,961) | (76.7) | | Gross (loss)/profit | – | – | (854) | (24.7) | 3,738 | 12.2 | 510 | 2.6 | 12,476 | 23.3 | | Other income and gains, net | – | – | 8,942 | 258.4 | 36,151 | 118.4 | 25,278 | 129.9 | 31,232 | 58.4 | | Selling and distribution expenses | – | – | (22,827) | (659.7) | (86,995) | (285.0) | (53,389) | (274.4) | (39,325) | (73.6) | | Administrative expenses | – | – | (7,615) | (220.1) | (14,384) | (47.1) | (9,610) | (49.4) | (22,074) | (41.3) | | Research and development expenses | (10,560) | – | (70,002) | (2,023.2) | (188,979) | (619.1) | (138,684) | (712.9) | (180,312) | (337.4) | | Fair value loss on financial liabilities | (60,509) | – | (176,826) | (5,110.6) | (214,172) | (701.7) | (128,063) | (658.3) | (313,477) | (586.6) | | Finance costs | (14) | – | (61) | (1.8) | (509) | (1.7) | (316) | (1.6) | (511) | (1.0) | | Impairment losses on financial assets, net | – | – | (3) | (0.1) | (88) | (0.3) | (68) | (0.3) | (22) | – | | Loss before tax | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) | | Income tax expense | – | – | – | – | – | – | – | – | – | – | | Loss for the year/period | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) |

**Attributable to:**

| | 2022 US$ | % | 2023 US$ | % | 2024 US$ | % | 2024 (unaudited) US$ | % | 2025 US$ | % | |---|---|---|---|---|---|---|---|---|---|---| | Owners of the parent | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) | | Non-controlling interests | – | – | – | – | – | – | – | – | – | – | | Loss and total comprehensive income for the year | (73,728) | – | (269,246) | (7,781.7) | (465,238) | (1,524.2) | (304,342) | (1,564.4) | (512,013) | (958.2) |

**Loss per share attributable to ordinary equity holders of the parent**

**Basic and diluted** – For loss for the year/period (US$): (0.74) | – | (2.56) | (4.28) | (2.80) | (4.71)

– 412 –

FINANCIAL INFORMATION NON-IFRS FINANCIAL MEASURE

We use adjusted net loss (non-IFRS measure), which is a non-IFRS financial measure, in evaluating our operating results and for financial and operational decision-making purposes. We believe that adjusted net loss (non-IFRS measure) helps identify underlying trends in our business that could otherwise be distorted by the effect of certain expenses that we include in our net loss. We believe that adjusted net loss (non-IFRS measure) provides useful information about our results of operations, enhances the overall understanding of our past performance and future prospects and allows for greater visibility with respect to key metrics used by our management in its financial and operational decision-making.

Adjusted net loss (non-IFRS measure) should not be considered in isolation or construed as an alternative to net loss or any other measure of performance or as an indicator of our operating performance. Investors are encouraged to review adjusted net loss (non-IFRS measure) and the reconciliation to its most directly comparable IFRS measure. Adjusted net loss (non-IFRS measure) presented here may not be comparable to similarly titled measures presented by other companies. Other companies may calculate similarly titled measures differently, limiting their usefulness as comparative measures to our data. We encourage investors and others to review our financial information in its entirety and not rely on a single financial measure.

We define our adjusted net loss (non-IFRS measure) as net loss adjusted by adding back (i) share-based payment expenses that are included in cost of sales, general administrative, research and development, and sales and marketing expenses, relates to the share-based awards that we grant to participants of our share incentive schemes and is a non-cash expense, (ii) fair value losses on financial liabilities, comprising fair value changes of convertible redeemable preferred shares which will be re-designated from liabilities to equity as a result of the automatic conversion into ordinary shares upon Listing, and convertible bonds, which have subsequently been repaid in full as of the Latest Practicable Date, and (iii) listing expenses.

– 413 –

FINANCIAL INFORMATION

The following table presents our non-IFRS financial measure for the years ended December 31, 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025.

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 US$ | 2023 US$ | 2024 US$ | 2024 US$ | 2025 US$ | | | | | | (unaudited) | | | | (in thousands) | | | | | | Loss for the year/period | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | Add: | | | | | | | Share-based payment expenses | 1,069 | 3,346 | 6,823 | 6,100 | 8,581 | | Fair value loss on financial liabilities | 60,509 | 176,826 | 214,172 | 128,063 | 313,477 | | Listing expenses | – | – | – | – | 3,675 | | Adjusted net loss for the year/period (non-IFRS measure) | (12,150) | (89,074) | (244,243) | (170,179) | (186,280) |

DESCRIPTION OF MAJOR COMPONENTS OF OUR RESULTS OF OPERATIONS

Revenue

Our revenue is derived from two primary sources — (i) AI-native products and (ii) Open Platform and other AI-based enterprise services, mainly consists of API usage as well as arrangements customized to enterprise requirements and licensed deliverables. For customised arrangements, we work with enterprise customers to set up dedicated inference resource pools tailored to their needs, helping ensure stable and predictable model inference performance. For licensed deliverables, we license our foundation models to enable customers to deploy and operate such models in their own systems. Each revenue stream reflects a distinct monetization pathway aligned with our product and platform strategies. The following table sets forth the breakdown of our revenue by nature, in absolute amounts and as a percentage of our total revenue, for the periods indicated.

– 414 –

FINANCIAL INFORMATION

For the nine months ended September 30,

For the year ended December 31, | | 2022 US$ | 2022 % | 2023 US$ | 2023 % | 2024 US$ | 2024 % | 2024 US$ | 2024 % | 2025 US$ | 2025 % | |---|---|---|---|---|---|---|---|---|---|---| | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | | |

AI原生产品 (AI-native products)　　　　　 开放平台及其他基于AI的企业服务 (Open Platform and other AI-based enterprise services)

| | – | – | 758 | 21.9 | 21,805 | 71.4 | 13,529 | 69.5 | 38,020 | 71.1 | | | – | – | 2,702 | 78.1 | 8,718 | 28.6 | 5,925 | 30.5 | 15,417 | 28.9 |

总收入 (Total revenue)

| | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 |

**AI原生产品 (AI-native products).** We generate revenue from individual users through subscription-based access to our monetized AI-native consumer applications, such as MiniMax, Hailuo AI, MiniMax Audio, and Talkie/Xingye. Subscriptions provide users with premium functionality across multi-modal generation, intelligent interaction, and personalized experiences. Revenue is recognised ratably over the subscription period, as we fulfill a stand-ready performance obligation to provide continuous access to content and services throughout the term. Users have the option to pre-purchase additional credits to recharge their accounts and buy these virtual items. For consumable virtual items, revenue is recognised when the virtual items are consumed. For non-consumable virtual items, revenue is recognised over the estimated average acting period of the paying users. In addition, we generate online marketing service revenue by providing marketing services to mediation platforms on certain of our AI-native applications. Revenue is recognised at a point in time, when a user views or clicks on an advertisement, thereby fulfilling our performance obligation. These services enable mediation platforms to engage with end users in a contextually relevant and measurable manner. As our user base and engagement levels expand, this revenue stream is expected to continue contributing to our overall monetization.

**开放平台及其他基于AI的企业服务 (Open Platform and other AI-based enterprise services).** We provide enterprise customers with access to our usage-based Open Platform and other AI-based enterprise services. Revenue from API usage is recognised at a point in time when the customers call APIs with tokens, which are billed under certain agreed fee schedule or usage-based structure. Revenue from other AI-based enterprise services, mainly consisting of arrangements customized to enterprise requirements and licensed deliverables, is typically recognised at a point in time, when control is transferred or acceptance is confirmed. Specifically, for customised arrangements, we work with enterprise customers to set up dedicated inference resource pools tailored to their needs, helping ensure stable and predictable model inference performance. For licensed deliverables, we license our foundation models to enable customers to deploy and operate such models in their own systems. These services support enterprise use cases across sectors such as smart devices, healthcare, tourism, and finance.

— 415 —

FINANCIAL INFORMATION

The tables below set forth breakdowns of revenue by product and further by monetization method:

For the year ended December 31, / For the nine months ended September 30,

| | 2022 US$ | 2022 % | 2023 US$ | 2023 % | 2024 US$ | 2024 % | 2024 US$ | 2024 % | 2025 US$ | 2025 % | |---|---|---|---|---|---|---|---|---|---|---| | | | | | | | | (unaudited) | | | | | (in thousands, except for percentages) | | | | | | | | | | | | **AI-native products** | | | | | | | | | | | | MiniMax | – | – | – | – | – | – | – | – | 756 | 1.4 | | Hailuo AI | – | – | – | – | 2,347 | 7.7 | – | – | 17,464 | 32.6 | | MiniMax Audio | – | – | – | – | – | – | – | – | 1,050 | 2.0 | | Talkie/Xingye | – | – | 758 | 21.9 | 19,458 | 63.7 | 13,529 | 69.5 | 18,750 | 35.1 | | **Open Platform and other AI-based enterprise services** | – | – | 2,702 | 78.1 | 8,718 | 28.6 | 5,925 | 30.5 | 15,417 | 28.9 | | **Total revenue** | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 |

For the year ended December 31, / For the nine months ended September 30,

| | 2022 US$ | 2022 % | 2023 US$ | 2023 % | 2024 US$ | 2024 % | 2024 US$ | 2024 % | 2025 US$ | 2025 % | |---|---|---|---|---|---|---|---|---|---|---| | | | | | | | | (unaudited) | | | | | (in thousands, except for percentages) | | | | | | | | | | | | **AI-native products** | | | | | | | | | | | | MiniMax | | | | | | | | | | | | — In-app top-up | – | – | – | – | – | – | – | – | 204.0 | 0.4 | | — Subscriptions | – | – | – | – | – | – | – | – | 552.0 | 1.0 | | Hailuo AI | | | | | | | | | | | | — In-app top-up | – | – | – | – | 527 | 1.7 | – | – | 3,317 | 6.2 | | — Subscriptions | – | – | – | – | 1,820 | 6.0 | – | – | 14,147 | 26.4 | | MiniMax Audio | | | | | | | | | | | | — In-app top-up | – | – | – | – | – | – | – | – | 196 | 0.4 | | — Subscriptions | – | – | – | – | – | – | – | – | 854 | 1.6 | | Talkie/Xingye | | | | | | | | | | | | — In-app top-up | – | – | 164 | 4.8 | 897 | 3.0 | 712 | 3.7 | 958 | 1.8 | | — Subscriptions | – | – | 594 | 17.1 | 3,960 | 12.9 | 2,917 | 14.9 | 6,604 | 12.4 | | — Online marketing service | – | – | – | – | 14,601 | 47.8 | 9,900 | 50.9 | 11,188 | 20.9 | | **Open Platform and other AI-based enterprise services** | – | – | 2,702 | 78.1 | 8,718 | 28.6 | 5,925 | 30.5 | 15,417 | 28.9 | | **Total revenue** | – | – | 3,460 | 100.0 | 30,523 | 100.0 | 19,454 | 100.0 | 53,437 | 100.0 |

– 416 –

FINANCIAL INFORMATION

In terms of geographic coverage, we generate revenue mainly in APAC, Americas, and EMEA. The following tables set out a breakdown of our revenue by geographical locations, in absolute amounts and as a percentage of our total revenue, for the periods indicated. For our AI-native products, revenue is based on users' billing address. For Open Platform, revenue is based on customers' jurisdiction of incorporation.

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | | | |---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | | | | APAC | – | – | 2,822 | 81.5 | 21,631 | 70.9 | 14,739 | 75.7 | 32,676 | 61.1 | | Americas | – | – | 598 | 17.3 | 5,405 | 17.7 | 3,012 | 15.5 | 12,658 | 23.7 | | EMEA | – | – | 40 | 1.2 | 3,487 | 11.4 | 1,703 | 8.8 | 8,103 | 15.2 | | **Total** | **–** | **–** | **3,460** | **100.0** | **30,523** | **100.0** | **19,454** | **100.0** | **53,437** | **100.0** |

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | | | |---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | | | | Mainland China | – | – | 2,797 | 80.8 | 9,217 | 30.2 | 6,768 | 34.8 | 14,400 | 26.9 | | Singapore | – | – | 1 | 0.0 | 11,455 | 37.5 | 7,664 | 39.4 | 12,980 | 24.3 | | United States | – | – | 575 | 16.6 | 4,999 | 16.4 | 2,871 | 14.8 | 10,913 | 20.4 | | Others | – | – | 87 | 2.6 | 4,852 | 15.9 | 2,151 | 11.0 | 15,144 | 28.4 | | **Total** | **–** | **–** | **3,460** | **100.0** | **30,523** | **100.0** | **19,454** | **100.0** | **53,437** | **100.0** |

Note: Others included nil, 88, 189, 134, and 221 jurisdictions during 2022, 2023, 2024, and the nine months ended September 30, 2024 and 2025, respectively. Revenue contribution from others primarily includes Israel, South Korea, the United Kingdom, Spain, Germany, Ireland, Australia, India, Hong Kong, Turkey, Brazil, Canada, France, and other countries and regions.

– 417 –

FINANCIAL INFORMATION

Cost of Sales

Our cost of sales primarily consists of expenses associated with cloud services related to inference, as well as platform commission fees.

• Cloud service costs related to inference activities refer to real-time computation and inference activities that directly support the delivery of our commercialized products and services, such as real-time interactions through our applications and open platform. These costs include expenditures on third-party cloud platforms to handle user prompts, generate outputs, and serve results to end-users. In contrast, cloud service expenses related to training are recognised under research and development expenses.

• In addition, commission fees paid to third-party distribution channels are recorded as cost of sales. As our user activity and enterprise usage volumes increased, these variable costs grew in absolute terms.

• However, as a percentage of revenue, our cost of sales decreased from 124.7% in 2023 to 87.8% in 2024 and also decreased from 97.4% in the nine months ended September 30, 2024 to 76.7% for the same period in 2025, reflecting increasing inference cost efficiency.

The following table sets out a breakdown of our cost of sales by nature, in absolute amounts and as a percentage of our total cost of sales, for the periods indicated.

| | For the year ended December 31, | | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | | | Cloud services costs related to inference activities | – | – | 4,097 | 95.0 | 25,966 | 96.9 | 18,370 | 97.0 | 37,988 | 92.7 | | Platform commission fees | – | – | 217 | 5.0 | 819 | 3.1 | 574 | 3.0 | 2,360 | 5.8 | | Labor costs | – | – | – | – | – | – | – | – | 592 | 1.4 | | Share-based payment expenses | – | – | – | – | – | – | – | – | 21 | 0.1 | | Total | – | – | 4,314 | 100.0 | 26,785 | 100.0 | 18,944 | 100.0 | 40,961 | 100.0 |

– 418 –

FINANCIAL INFORMATION

The following table sets forth a breakdown of our cost of sales by revenue source, in absolute amounts and as a percentage of our total cost of sales, for the periods indicated.

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | (in thousands, except for percentages) |

| | 2022 US$ | % | 2023 US$ | % | 2024 US$ | % | 2024 US$ | % | 2025 US$ | % | |---|---|---|---|---|---|---|---|---|---|---| | AI-native products | – | – | 3,640 | 84.4 | 23,581 | 88.0 | 16,711 | 88.2 | 36,246 | 88.5 | | Open Platform and other AI-based enterprise services | – | – | 674 | 15.6 | 3,204 | 12.0 | 2,233 | 11.8 | 4,715 | 11.5 | | Total | – | – | 4,314 | 100.0 | 26,785 | 100.0 | 18,944 | 100.0 | 40,961 | 100.0 |

(unaudited) (in thousands, except for percentages)

Gross Profit and Gross Profit Margin

As a result of the foregoing, we recorded gross profit of nil, negative US$0.9 million, US$3.7 million, US$0.5 million and US$12.5 million in 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively, representing gross profit margins of nil, negative 24.7%, 12.2%, 2.6% and 23.3%, respectively, during the same periods.

The following table sets forth a breakdown of our gross profit and gross profit margin by revenue source for the periods indicated. Our gross profit and gross profit margin have been and will continue to be affected by a number of factors, including the intelligence level of our foundation models, our revenue mix, pricing strategies, and inference cost efficiency.

| | For the year ended December 31, | | | | | | | | For the nine months ended September 30, | | | | |---|---|---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | Gross Profit US$ | Gross Profit Margin % | Gross Profit US$ | Gross Profit Margin % | Gross Profit US$ | Gross Profit Margin % | Gross Profit US$ | Gross Profit Margin % | Gross Profit US$ | Gross Profit Margin % | | | (unaudited) | | | (in thousands, except for percentages) |

| | 2022 Gross Profit US$ | Gross Profit Margin % | 2023 Gross Profit US$ | Gross Profit Margin % | 2024 Gross Profit US$ | Gross Profit Margin % | 2024 Gross Profit US$ | Gross Profit Margin % | 2025 Gross Profit US$ | Gross Profit Margin % | |---|---|---|---|---|---|---|---|---|---|---| | AI-native products | – | – | (2,882) | (380.2) | (1,776) | (8.1) | (3,182) | (23.5) | 1,774 | 4.7 | | Open Platform and other AI-based enterprise services | – | – | 2,028 | 75.1 | 5,514 | 63.2 | 3,692 | 62.3 | 10,702 | 69.4 | | Total | – | – | (854) | (24.7) | 3,738 | 12.2 | 510 | 2.6 | 12,476 | 23.3 |

(unaudited) (in thousands, except for percentages)

– 419 –

FINANCIAL INFORMATION

Other Income and Gains, Net

Our other income and gains, net consist of (i) interest income, (ii) foreign exchange gains, net, (iii) fair value gain on financial assets at fair value through profit or loss, representing gains arising from the remeasurement of financial assets, including structured wealth management products designated at fair value through profit or loss, and (iv) others, which mainly consist of gain or loss on disposal of long-term assets and early termination of right-of-use assets and lease liabilities.

The following table sets forth the breakdown of our income and gains, net for the periods indicated.

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | (in thousands, except for percentages) |

| Other income and gains, net | | | | | | | | | | | | Interest income | 39 | 3.4 | 7,785 | 87.1 | 20,448 | 56.6 | 17,199 | 68.0 | 7,876 | 25.2 | | Foreign exchange gains, net | 175 | 15.2 | 311 | 3.5 | 2 | 0.0 | 1,415 | 5.7 | 1,600 | 5.1 | | Fair value gain on financial assets at fair value through profit or loss | 941 | 81.4 | 788 | 8.8 | 15,710 | 43.4 | 6,682 | 26.4 | 20,414 | 65.4 | | Others | – | – | 58 | 0.6 | (9) | 0.0 | (18) | (0.1) | 1,342 | 4.3 | | Total | 1,155 | 100.0 | 8,942 | 100.0 | 36,151 | 100.0 | 25,278 | 100.0 | 31,232 | 100.0 |

Selling and Distribution Expenses

Our selling and distribution expenses primarily consist of (i) business promotion expenses, including spending on brand campaigns, online and offline marketing, and user growth initiatives intended to enhance awareness and adoption of our AI-native products, (ii) staff costs, including salaries, bonuses, and social insurance contributions paid to personnel engaged in marketing and business development, (iii) share-based payment expenses, which includes the amortization of employee stock options granted to selling and distribution personnel, and (iv) others, including travel costs and public relations fees.

– 420 –

FINANCIAL INFORMATION

The following table sets forth a breakdown of our selling and distribution expenses, in absolute amounts and as a percentage of our total selling and distribution expenses, for the periods indicated.

| | For the year ended December 31, | | | | For the nine months ended September 30, | | | | |---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | | | | | | | (unaudited) | | | | | | (in thousands, except for percentages) | | Business promotion expenses | 345 | 58.8 | 22,035 | 96.5 | 84,861 | 97.5 | 52,122 | 97.6 | 36,190 | 92.0 | | Staff costs | 240 | 40.9 | 698 | 3.1 | 1,869 | 2.1 | 1,118 | 2.1 | 2,774 | 7.1 | | Share-based payment expenses | – | – | 4 | – | 61 | 0.1 | 20 | 0.0 | 104 | 0.3 | | Others | 2 | 0.3 | 90 | 0.4 | 204 | 0.3 | 129 | 0.3 | 257 | 0.6 | | Total | 587 | 100.0 | 22,827 | 100.0 | 86,995 | 100.0 | 53,389 | 100.0 | 39,325 | 100.0 |

These expenses support the commercial adoption of our product offerings across individual and enterprise users and have increased in absolute amount in line with our growth in active user base and market presence from 2022 to 2024, and have decreased in absolute amount in the nine months ended September 30, 2025 due to the adoption of organic user growth strategy and the enhanced model intelligence. As a percentage of our total revenue, our selling and distribution expenses were 659.7%, 285.0%, 274.4% and 73.6% in 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively. This decrease was primarily driven by the success of our organic user acquisition strategy.

Looking ahead, we expect our selling and distribution expenses to continue to decrease as a percentage of revenue as we focus more on organic growth of our user and customer base. Further information about the movement of our selling and distribution expenses during the Track Record Period is set forth in "— Period-to-Period Comparison of Results of Operations."

Administrative Expenses

Our administrative expenses primarily consist of (i) staff costs, including salaries, bonuses, and social insurance contributions, for personnel engaging in administrative function, (ii) listing expenses, (iii) professional service fees, mainly for professional services and IT services, (iv) depreciation and amortization, relating to periodic expense recognition of right-of-use asset, office equipment and leasehold improvements, used in administrative functions, (v) share-based payment expenses, which includes the amortization of employee stock options granted to management personnel, (vi) travel and office expenses, comprising office operations and business travel expenses, and (vii) premise expenses, comprising short-term rental expense, premises maintenance fees and utilities fee related to core administrative activities, and (viii) others, mainly residual administrative items.

– 421 –

FINANCIAL INFORMATION

The following table sets forth a breakdown of our administrative expenses, in absolute amounts and as a percentage of our total administrative expenses, for the periods indicated.

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | 2025 | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) |

| | | | | | | | | | | | |---|---|---|---|---|---|---|---|---|---|---| | Staff costs | 661 | 20.6 | 2,392 | 31.3 | 5,502 | 38.3 | 3,710 | 38.6 | 9,997 | 45.3 | | Listing expenses | – | – | – | – | – | – | – | – | 3,675 | 16.6 | | Professional service fees | 719 | 22.4 | 1,931 | 25.4 | 3,483 | 24.1 | 2,438 | 25.4 | 2,116 | 9.6 | | Depreciation and amortization | 207 | 6.4 | 773 | 10.2 | 1,852 | 12.9 | 1,397 | 14.5 | 2,060 | 9.3 | | Share-based payment expenses | 981 | 30.6 | 1,516 | 19.9 | 2,059 | 14.3 | 1,358 | 14.1 | 2,109 | 9.6 | | Travel and office expenses | 181 | 5.6 | 546 | 7.2 | 556 | 3.9 | 183 | 1.9 | 609 | 2.8 | | Premise expenses | 72 | 2.2 | 365 | 4.8 | 374 | 2.6 | 255 | 2.7 | 375 | 1.7 | | Others | 392 | 12.2 | 92 | 1.2 | 558 | 3.9 | 269 | 2.8 | 1,133 | 5.1 | | **Total** | **3,213** | **100.0** | **7,615** | **100.0** | **14,384** | **100.0** | **9,610** | **100.0** | **22,074** | **100.0** |

Our administrative expenses increased in absolute amount during the Track Record Period as we expanded operations. As a percentage of our total revenue, our administrative expenses were 220.1%, 47.1%, 49.4% and 41.3% in 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively. We expect our administrative expenses to continue increasing in absolute terms as we expand our operations, organization, and compliance framework. However, we expect that our administrative expenses as a percentage of total revenue will decrease as we improve our operational efficiency and benefit from economies of scale. Further information about the movement of our administrative expenses during the Track Record Period is set forth in "— Period-to-Period Comparison of Results of Operations."

**Research and Development Expenses**

Our research and development expenses consist primarily of: (i) cloud services expenses related to training activities including foundational model training, architectural experimentation, large-scale evaluation, and early-stage prototyping; (ii) staff costs, including salaries, bonuses, and social insurance contributions for personnel engaging in research and development function; (iii) share-based payment expenses, which includes the amortization of employee stock options granted to research and development personnel; (iv) travel and professional expenses, outsourced technical services and mobility-related expenses incurred by our research and development teams; and (v) others, including general support expenses for research and development activities such as depreciation of office equipment.

– 422 –

FINANCIAL INFORMATION

The following table sets forth the breakdown of our research and development expenses, in absolute amounts and as a percentage of our total research and development expenses, for the periods indicated.

| | For the year ended December 31, | | | | For the nine months ended September 30, | | |---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | (in thousands, except for percentages) |

| | | | | | | | | | | | |---|---|---|---|---|---|---|---|---|---|---| | Cloud services expenses related to training | 4,149 | 39.4 | 47,229 | 67.6 | 140,642 | 74.4 | 103,114 | 74.4 | 142,397 | 79.0 | | Staff costs | 5,461 | 51.7 | 18,674 | 26.7 | 40,420 | 21.4 | 28,514 | 20.6 | 28,531 | 15.8 | | Share-based payment expenses | 88 | 0.8 | 1,826 | 2.6 | 4,703 | 2.5 | 4,722 | 3.4 | 6,347 | 3.5 | | Travel and professional expenses | 457 | 4.3 | 1,467 | 2.1 | 2,842 | 1.5 | 2,066 | 1.5 | 2,670 | 1.5 | | Others | 405 | 3.8 | 806 | 1.0 | 372 | 0.2 | 268 | 0.1 | 367 | 0.2 | | Total | 10,560 | 100.0 | 70,002 | 100.0 | 180,312 | 100.0 | 138,684 | 100.0 | 188,979 | 100.0 |

Our research and development expenses increased in absolute amount throughout the Track Record Period, reflecting our strategic focus on advancing foundational AI model capabilities. As a percentage of total revenue, our research and development expenses decreased from over 2,000% in 2023 to 619.1% in 2024, and further decreased from 712.9% in the nine months ended September 30, 2024 to 337.4% in the nine months ended September 30, 2025, reflecting revenue scale-up while maintaining a high level of technology investment. We expect our research and development expenses to continue increasing in absolute terms as we expand our model capabilities and pursue further innovation. As a percentage of our total revenue, however, our research and development expenses are expected to decrease over time as we continue to enhance commercialization and improve our research and development efficiency, especially on foundational model training activities and organizational efficiency. Further information about the movement of our research and development expenses during the Track Record Period is set forth in "— Period-to-Period Comparison of Results of Operations."

Fair Value Loss on Financial Liabilities

Our fair value loss on financial liabilities represents changes in the carrying amount of our convertible redeemable preferred shares and other financial liabilities. These fair value changes are non-cash in nature. We recorded fair value loss on financial liabilities of US$60.5 million, US$176.8 million, US$214.2 million, US$128.1 million and US$313.5 million in 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively.

– 423 –

FINANCIAL INFORMATION

Finance Costs

Our finance costs consist of (i) interest on bank and other borrowings, and (ii) interest on lease liabilities. We recorded finance costs of US$14.0 thousand, US$61.0 thousand, US$0.5 million, US$0.3 million and US$0.5 million in 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively.

The following table sets forth the breakdown of our finance costs, in absolute amounts and as a percentage of our total finance costs, for the periods indicated.

| | For the year ended December 31, | | | | | | For the nine months ended September 30, | | | | |---|---|---|---|---|---|---|---|---|---|---| | | 2022 | | 2023 | | 2024 | | 2024 | | 2025 | | | | US$ | % | US$ | % | US$ | % | US$ | % | US$ | % | | | (unaudited) | | | | | | | | | | | | (in thousands, except for percentages) | | | | | | | | | | | Interest on bank and other borrowings | – | – | – | – | 355 | 69.7 | 199 | 63.0 | 404 | 79.1 | | Interest on lease liabilities | 14 | 100.0 | 61 | 100.0 | 154 | 30.3 | 117 | 37.0 | 107 | 20.9 | | Total | 14 | 100.0 | 61 | 100.0 | 509 | 100.0 | 316 | 100.0 | 511 | 100.0 |

Impairment Losses on Financial Assets, Net

Our impairment losses on financial assets, net represent the expected credit losses or reversal of the expected credit losses on our trade receivables and other receivables. We recorded impairment losses on financial assets, net of nil, US$3.0 thousand, US$88.0 thousand, US$68.0 thousand and US$22.0 thousand in 2022, 2023, 2024 and the nine months ended September 30, 2024 and 2025, respectively.

Income Tax Expense

Income tax expense refers to the aggregate amount of taxes credited or incurred in a given period, calculated in accordance with the applicable laws and regulations. It consists of current income tax, the tax payable on taxable profits for the current period, and deferred income tax, which arises from temporary differences between the accounting and tax treatment of certain items, recognised in accordance with applicable accounting standards. We are subject to various rates of income tax under different jurisdictions. The following summarizes major factors affecting our applicable tax rates in Mainland China, the Cayman Islands, Hong Kong and Singapore.

– 424 –

FINANCIAL INFORMATION

Mainland China

Entities located in the PRC are subject to a statutory income tax rate of 25%, in accordance with the PRC Corporate Income Tax Law. During the Track Record Period, certain subsidiaries within our Group qualified for preferential income tax treatment as High and New Technology Enterprises. Specifically, Beijing Jizhi qualified for a preferential tax rate of 15% from 2023 to 2025, and Shanghai Jizhi qualified for a 15% rate from 2024 to 2026, subject to review by the relevant PRC tax authorities every three years. We did not record any income tax expense during the Track Record Period. Our effective tax rate (calculated as income tax expense divided by profit before tax) was 0% for all periods presented, as we incurred losses before tax during the Track Record Period and did not record any material current or deferred income tax expense.

开曼群岛

根据开曼群岛现行法律，我们及我们的子公司无需就收入或资本收益缴纳税款。

香港

在香港注册成立的子公司须就其在相关期间及截至2024年及2025年9月30日止九个月内于香港产生的应评税利润，按16.5%的税率缴纳香港利得税。在相关期间及截至2024年及2025年9月30日止九个月内，各子公司首200万港元的应评税利润按8.25%税率征税，其余应评税利润按16.5%税率征税。

新加坡

在新加坡注册成立的子公司须就其在所示期间内于新加坡产生的应评税利润，按17%的税率缴纳新加坡利得税。

我们在追踪记录期间录得零所得税支出。在追踪记录期间及截至最后实际可行日期，我们与相关税务机关之间不存在任何争议或未解决的税务问题。

年度/期间亏损

综上所述，我们于2022年、2023年、2024年及截至2024年及2025年9月30日止九个月分别录得年度/期间亏损7,370万美元、2.692亿美元、4.652亿美元、3.043亿美元及5.120亿美元。

年度/期间经调整净亏损（非国际财务报告准则计量）

我们于2022年、2023年、2024年及截至2024年及2025年9月30日止九个月分别录得年度/期间经调整净亏损（非国际财务报告准则计量）1,220万美元、8,910万美元、2.442亿美元、1.702亿美元及1.863亿美元。

– 425 –

FINANCIAL INFORMATION PERIOD-TO-PERIOD COMPARISON OF RESULTS OF OPERATIONS Nine Months Ended September 30, 2025 Compared with Nine Months Ended September 30, 2024

Revenue Our revenue increased by 174.7% from US$19.5 million for the nine months ended September 30, 2024 to US$53.4 million during the same period in 2025. This was primarily driven by the advancement in intelligence level of our foundation models, continued expansion of both of our monetization channels — AI-native products, as well as Open Platform and other AI-based enterprise services, — as we released our models, advanced the commercial rollout of our product suite and broadened our individual users and enterprise customers.

AI-native products. Revenue from our AI-native products increased by 181.0% from US$13.5 million for the nine months ended September 30, 2024 to US$38.0 million during the same period in 2025, primarily driven by higher user engagement and increased customer willingness to pay for our products, as well as the successful launch of products such as Hailuo AI. Average MAUs grew substantially from approximately 14.6 million in the nine months ended September 30, 2024 to approximately 27.6 million during the same period in 2025. Paying users also rose significantly from approximately 489,100 in the nine months ended September 30, 2024 to approximately 1,771,600 during the same period in 2025. These growth figures reflect the development of our business and the expanding customer base, signaling the successful execution of our strategy to drive user engagement and retention.

Open Platform and other AI-based enterprise services. Revenue generated from Open Platform and other AI-based enterprise services increased by 160.2% from US$5.9 million for the nine months ended September 30, 2024 to US$15.4 million during the same period in 2025, primarily fueled by a notable increase in paying users, defined as users who individually consumed no less than US$50 worth of API calls (or its equivalent in other currencies) from approximately 400 in the nine months ended September 30, 2024 to approximately 2,500 during the same period in 2025. Moreover, revenue from overseas markets witnessed significant growth from US$0.1 million in the nine months ended September 30, 2024 to US$7.8 million during the same period in 2025, contributing to the overall revenue increase from Open Platform and other AI-based enterprise services, reflecting the broader global market acceptance and demand for our products.

Cost of Sales Our cost of sales increased by 116.2% from US$18.9 million for the nine months ended September 30, 2024 to US$41.0 million during the same period in 2025, primarily attributable to a 116.9% increase in costs associated to our AI-native products, which rose from US$16.7 million to US$36.2 million over the same period, consistent with our rapid business expansion.

– 426 –

FINANCIAL INFORMATION

Gross Profit and Gross Profit Margin

As a result of the foregoing, our gross profit improved from US$0.5 million for the nine months ended September 30, 2024 to a gross profit of US$12.5 million during the same period in 2025. Our gross profit margin increased from 2.6% for the nine months ended September 30, 2024 to 23.3% during the same period in 2025.

The increase in our overall gross profit margin was primarily driven by the continuous advancement of our foundation models' intelligence and the significant improvements in the efficiency of our inference processes. In particular, the gross profit margin for AI-native products, which was the most significant contributor to our revenue during the Track Record Period, significantly improved from a gross profit margin of negative 23.5% for the nine months ended September 30, 2024 to a gross profit margin of 4.7% during the same period in 2025.

Other Income and Gains, Net

Our other income and gains, net increased by 23.6% from US$25.3 million for the nine months ended September 30, 2024 to US$31.2 million during the same period in 2025, primarily due to an increase in fair value gain on financial assets at fair value through profit or loss from US$6.7 million in the nine months ended September 30, 2024 to US$20.4 million during the same period in 2025, driven by the rise in returns on financial investments resulting from a larger investment allocation.

Selling and Distribution Expenses

Our selling and distribution expenses decreased by 26.3% from US$53.4 million for the nine months ended September 30, 2024 to US$39.3 million during the same period in 2025. This decrease was primarily attributable to a 30.6% decrease in business promotion expenses from US$52.1 million in the nine months ended September 30, 2024 to US$36.2 million during the same period in 2025, as we have been adjusting our marketing strategies with focused efforts on an organic user acquisition approach, without relying heavily upon brand promotion and user acquisition spending.

Administrative Expenses

Our administrative expenses increased by 129.7% from US$9.6 million for the nine months ended September 30, 2024 to US$22.1 million during the same period in 2025, mainly driven by an increase in staff costs during the same period due to increasing headcount and share-based payment expenses for administrative personnel. Moreover, the listing expenses incurred in the nine months ended September 30, 2025 also contributed to the overall increase in administrative expenses, which did not occur during the same period in 2024.

– 427 –

FINANCIAL INFORMATION

Research and Development Expenses

Our research and development expenses increased by 30.0% from US$138.7 million for the nine months ended September 30, 2024 to US$180.3 million during the same period in 2025, mainly attributed to an increase in cloud services expenses related to training activities, driven by the increased model iteration and upgrades as we continued to develop and refine our foundation models and multi-modal capabilities. The year over year growth rate of our research and development expenses in the nine months ended September 30, 2025 was 30.0%, significantly lower than our revenue growth rate of 174.7% during the same period, demonstrating our improved research and development efficiency.

Fair Value Loss on Financial Liabilities

Our fair value loss on financial liabilities increased from US$128.1 million for the nine months ended September 30, 2024 to US$313.5 million during the same period in 2025, mainly driven by significant remeasurement losses on our preferred shares due to continued increases in our valuation.

Finance Costs

Our finance costs increased by 61.7% from US$0.3 million for the nine months ended September 30, 2024 to US$0.5 million during the same period in 2025, primarily due to an increase in interest on bank and other borrowings from US$0.2 million to US$0.4 million over the period.

Impairment Losses on Financial Assets, Net

We recorded impairment losses on financial assets, net of US$68.0 thousand for the nine months ended September 30, 2024 and US$22.0 thousand during the same period in 2025. The improvement over impairment losses on financial assets, net was primarily attributable to our effective collection efforts, which resulted in the release of previously recognised expected credit loss provisions.

Loss for the Period

As a result of the foregoing, our loss for the period increased by 68.2% from US$304.3 million for the nine months ended September 30, 2024 to US$512.0 million during the same period in 2025.

Year Ended December 31, 2024 Compared with Year Ended December 31, 2023

Revenue

Our revenue increased significantly by 782.2% from US$3.5 million in 2023 to US$30.5 million in 2024. This increase was primarily driven by the advancement in intelligence level of our foundation model, which resulted in rapid growth across both of our monetization channels — Open Platform and other AI-based enterprise services and AI-native products — as we scaled up commercialization of our AI-native product suite and user base.

– 428 –

FINANCIAL INFORMATION

AI-native products. Revenue from AI-native products increased by 2,776.6% from US$0.8 million in 2023 to US$21.8 million in 2024, as we began ramping up commercialization mainly through online marketing services and value-added premium features. Such initial monetization success was fueled by continued expansion in our consumer-facing product suite. Specifically, we expanded value-added premium features in core monetized products such as Hailuo AI and Talkie/Xingye, and actively optimized pricing tiers to enhance monetization efficiency. Average MAUs grew from approximately 3.1 million in 2023 to 19.1 million in 2024, and paying users for AI-native products rose from approximately 119,700 in 2023 to approximately 650,300 in 2024, reflecting increased user willingness to pay for premium and intelligent experiences. With an increasingly engaged user base, we were able to grow our online marketing services associated with certain AI-native products.

开放平台及其他基于AI的企业服务（Open Platform and other AI-based enterprise services）。开放平台及其他基于AI的企业服务的收入从2023年的270万美元增加222.6%至2024年的870万美元。这主要由我们开放平台的采用率提升所驱动，该平台在代币（token）使用量及企业开发者订阅方面均实现增长。关键付费用户数量（定义为单独消费API调用金额不低于50美元或其他货币等值金额的用户）从2023年的约100名增长至2024年的700名。

销售成本（Cost of Sales）

我们的销售成本从2023年的430万美元增加520.9%至2024年的2,680万美元，主要归因于用于支持我们AI原生消费者应用程序及开放平台推理工作负载的云服务费用增加533.8%，从同期的410万美元上升至2,600万美元。此增长由更大规模的基础设施使用所驱动，以支持我们AI原生消费者应用程序及面向企业的开放平台的推理工作负载，因为年内累计用户交互及API代币消耗量大幅激增。

毛利及毛利率（Gross Profit and Gross Profit Margin）

由于上述原因，我们的毛利从2023年的负90万美元增加至2024年的370万美元。我们的毛利率从2023年的负24.7%上升至2024年的12.2%，这是由于我们收入来源的构成及其各自毛利率发生变化所致。

我们整体毛利率的提升主要由基础模型智能水平的提高及模型推理效率的优化所驱动。特别是，AI原生产品的毛利率——该类别为追踪记录期间我们收入的最主要贡献来源——从2023年的负380.2%显著改善至2024年的负8.1%，主要归因于我们基础模型智能水平的提升、用户参与度及变现能力的增强，以及新变现功能的引入。

— 429 —

FINANCIAL INFORMATION

Other Income and Gains, Net

Our other income and gains, net increased by 304.3% from US$8.9 million in 2023 to US$36.2 million in 2024, respectively. This increase was primarily attributable to a significant rise in fair value gains on financial assets at fair value through profit or loss, which increased by 1,893.7%, from US$0.8 million in 2023 to US$15.7 million in 2024. These gains primarily reflected unrealized mark-to-market increases in the carrying value of certain financial instruments, including investments designated at fair value. To a lesser extent, interest income increased by 162.7% from US$7.8 million in 2023 to US$20.4 million in 2024, mainly due to higher average cash balances.

Selling and Distribution Expenses

Our selling and distribution expenses increased by 281.1% from US$22.8 million in 2023 to US$87.0 million in 2024. This increase was primarily driven by a 285.1% increase in business promotion expenses, which rose from US$22.0 million in 2023 to US$84.9 million in 2024. The increase was due to our exploration of various user growth channels during our initial period of commercialization. To a lesser extent, staff costs increased by 168.0% from US$0.7 million in 2023 to US$1.9 million in 2024, attributable to increases in headcount for personnel engaged in sales and marketing.

Administrative Expenses

Our administrative expenses increased by 88.9% from US$7.6 million in 2023 to US$14.4 million in 2024, mainly driven by (i) a 130.0% increase in staff costs from US$2.4 million in 2023 to US$5.5 million in 2024 due to higher headcount for administrative personnel and (ii) a 80.4% increase in professional service fees from US$1.9 million in 2023 to US$3.5 million in 2024, mainly due to increased cost for legal, audit and advisory expenses and hiring expenses in connection with our growing operations.

Research and Development Expenses

Our research and development expenses increased by 170.0% from US$70.0 million in 2023 to US$189.0 million in 2024, mainly attributed to (i) a 197.8% increase in cloud services expenses related to training activities from US$47.2 million in 2023 to US$140.6 million in 2024 due to increased model training, evaluation, and architecture experimentation activities as we continued to develop our foundation models and multi-modal capabilities and (ii) a 116.5% increase in staff costs, from US$18.7 million to US$40.4 million over the same period, attributable to higher headcount for our in-house research and engineering teams. The year over year growth rate of our research and development expense in 2024 was 170.0%, significantly lower than our revenue growth rate of 782.2% during the same period, demonstrating our improved research and development efficiency.

– 430 –

FINANCIAL INFORMATION

Fair Value Loss on Financial Liabilities

Our fair value loss on financial liabilities increased by 21.1% from US$176.8 million in 2023 to US$214.2 million in 2024, primarily due to continued remeasurement losses on our preferred shares in 2024, as our valuation increased during the year.

Finance Costs

Our finance costs increased by 734.4% from US$61 thousand in 2023 to US$0.5 million in 2024. This increase was primarily attributable to the incurrence of interest expenses of US$0.4 million in 2024 on bank loans and other borrowings, as we initiated financing arrangements to support our working capital and operating needs. Interest on lease liabilities also increased by 152.5%, from US$61.0 thousand in 2023 to US$0.2 million in 2024, as we expanded our office footprint and entered into new lease agreements to accommodate headcount growth across functions.

Impairment Losses on Financial Assets, Net

We recorded impairment losses on financial assets, net of US$3.0 thousand and US$88.0 thousand in 2023 and 2024, respectively. The increase was primarily attributable to provisions recognised on trade receivables and contract assets in connection with our expanding revenue base and customer coverage. As we scaled up monetization activities, we implemented expected credit loss assessments across a broader set of accounts to align with our credit risk management policies.

Loss for the Year

As a result of the foregoing, our loss for the year increased by 72.8% from US$269.2 million in 2023 to US$465.2 million in 2024.

Year Ended December 31, 2023 Compared with Year Ended December 31, 2022

Revenue

Our revenue increased from nil in 2022 to US$3.5 million in 2023. This increase was primarily driven by initial commercialization across both monetization channels as we launched our AI-native products and began scaling user and enterprise adoption.

AI-native products. We did not generate any revenue from our AI-native products in 2022. In 2023, we recorded US$0.8 million revenue generated by our AI-native products, attributable to the introduction of paid tiers across our consumer-facing applications. Our average MAUs reached approximately 3.1 million during the year and our paying users for AI-native products reached approximately 119,700, as early monetization efforts, including for Talkie/Xingye, gained initial traction.

– 431 –

FINANCIAL INFORMATION

Open Platform. Revenue from Open Platform mainly consists of API calls and other AI-based enterprise services, which includes arrangements customized to enterprise requirements and licensed deliverables. For customised arrangements, we work with enterprise customers to set up dedicated inference resource pools tailored to their needs, helping ensure stable and predictable model inference performance. For licensed deliverables, we license our foundation models to enable customers to deploy and operate such models in their own systems. We did not generate enterprise service revenue in 2022. In 2023, revenue from API and other AI-based enterprise services was US$2.7 million, primarily driven by the launch and adoption of our Open Platform by enterprise developers across initial use cases such as customer service, smart devices, and education. Our paying users, defined as users who have individually consumed no less than US$50 worth of API calls (or its equivalent in other currencies) reached approximately 100 in 2023.

Cost of Sales

Our cost of sales increased from nil in 2022 to US$4.3 million in 2023. The increase was primarily attributable to cloud service costs of US$4.1 million incurred to support inference workloads as we began commercializing our API services and consumer products, including expenditures on third-party cloud services.

Gross Profit and Gross Profit Margin

As a result of the foregoing, our gross profit amounted to nil in 2022 and negative US$0.9 million in 2023. Our gross profit margin was nil in 2022 and negative 24.7% in 2023 (when our monetization was still at a nascent stage).

Our overall gross profit margin was primarily driven by initial commercialization across both monetization channels as we launched our AI-native products suites and began scaling user and enterprise adoption.

Other Income and Gains, Net

Our other income and gains, net increased by 674.2% from US$1.2 million in 2022 to US$8.9 million in 2023. The increase was primarily attributable to a rise in interest income from US$39.0 thousand in 2022 to US$7.8 million in 2023, reflecting higher average cash balances.

Selling and Distribution Expenses

We did not record material selling and distribution expenses in 2022. In 2023, these expenses amounted to US$22.8 million, primarily driven by business promotion expenses of US$22.0 million, reflecting investments in user growth initiatives and brand campaigns to support our commercial launch.

– 432 –

FINANCIAL INFORMATION

Administrative Expenses

Our administrative expenses increased by 137.0% from US$3.2 million in 2022 to US$7.6 million in 2023, mainly driven by (i) an increase in staff costs from US$0.7 million to US$2.4 million due to headcount growth across administrative functions and (ii) a rise in professional service fees from US$0.7 million in 2022 to US$1.9 million in 2023, as we engaged legal, financial, and compliance consultants in connection with operational expansion and financing activities.

Research and Development Expenses

Our research and development expenses increased by 562.9% from US$10.6 million in 2022 to US$70.0 million in 2023. This increase was primarily driven by (i) cloud services expenses related to training activities, which rose by from US$4.1 million in 2022 to US$47.2 million in 2023, due to intensified model training and evaluation activity, and (ii) staff costs, which increased from US$5.5 million to US$18.7 million, due to growth in our research and engineering team headcount and compensation levels.

Fair Value Loss on Financial Liabilities

Our fair value loss on financial liabilities increased by 192.2% from US$60.5 million in 2022 to US$176.8 million in 2023, mainly driven by higher remeasurement losses on our preferred shares due to valuation appreciation during the period.

Finance Costs

Our finance costs increased by 335.7% from US$14.0 thousand in 2022 to US$61.0 thousand in 2023, primarily due to higher interest on lease liabilities, reflecting additional lease arrangements entered into to support team expansion and infrastructure needs.

Impairment Losses on Financial and Contract Assets, Net

We did not record impairment losses on financial and contract assets in 2022. In 2023, we recognised US$3.0 thousand in impairment losses, primarily reflecting expected credit losses assessed on a limited set of customer accounts in our early revenue-generating activities.

Loss for the Year

As a result of the foregoing, our loss for the year increased by 265.2% from US$73.7 million in 2022 to US$269.2 million in 2023.

– 433 –

FINANCIAL INFORMATION

DISCUSSION OF CERTAIN KEY ITEMS FROM OUR CONSOLIDATED STATEMENTS OF FINANCIAL POSITION

The table below sets forth selected information from our consolidated statements of financial position as of the dates indicated, which has been extracted from our consolidated financial statements included in Appendix I to this Prospectus.

| | As of December 31, | As of December 31, | As of December 31, | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | | |

**NON-CURRENT ASSETS**

| | 2022 | 2023 | 2024 | 2025 | |---|---|---|---|---| | Property, plant and equipment | 231 | 709 | 1,093 | 1,134 | | Right-of-use assets | 458 | 3,313 | 3,077 | 2,746 | | Prepayments, other receivables and other assets | – | 435 | 561 | 731 | | Financial assets at fair value through profit or loss | – | – | 95,331 | 70,228 | | Financial assets at fair value through other comprehensive income | – | – | – | – | | Restricted cash | – | 39 | 4,836 | 6,440 | | **Total non-current assets** | **689** | **4,496** | **104,936** | **81,320** |

| | 2022 | 2023 | 2024 | 2025 | |---|---|---|---|---| | | – | 38 | 38 | 41 |

**CURRENT ASSETS**

| | 2022 | 2023 | 2024 | 2025 | |---|---|---|---|---| | Trade receivables | – | 1,338 | 6,982 | 8,063 | | Prepayments, other receivables and other assets | 569 | 4,378 | 13,470 | 11,811 | | Financial assets at amortised costs | – | – | 147,444 | – | | Financial assets at fair value through profit or loss | 65,791 | 15,802 | 295,220 | 644,154 | | Time deposits | – | 91,698 | 26,327 | – | | Restricted cash | 2,221 | – | 27,293 | 25,097 | | Cash and cash equivalents | 4,691 | 206,295 | 288,912 | 362,647 | | **Total current assets** | **73,272** | **319,511** | **805,648** | **1,051,772** |

— 434 —

FINANCIAL INFORMATION

| | As of September 30, | As of December 31, | | | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | | |

**CURRENT LIABILITIES**

| | | | | | |---|---|---|---|---| | Interest-bearing bank borrowings | – | – | 19,455 | 19,102 | | Trade and bills payables | 2,394 | 17,242 | 51,212 | 70,219 | | Other payables, accruals and other liabilities | 2,326 | 14,741 | 51,512 | 17,322 | | Contract liabilities | – | 559 | 1,553 | 4,657 | | Lease liabilities | 349 | 1,248 | 1,964 | 1,694 | | Convertible redeemable preferred shares | 145,175 | 629,001 | 1,581,949 | 2,321,193 | | **Total current liabilities** | **150,244** | **662,791** | **1,707,645** | **2,434,187** |

| | | | | | |---|---|---|---|---| | NET CURRENT LIABILITIES | (76,972) | (343,280) | (901,997) | (1,382,415) | | TOTAL ASSETS LESS CURRENT LIABILITIES | (76,283) | (338,784) | (797,061) | (1,301,095) |

**NON-CURRENT LIABILITIES**

| | | | | | |---|---|---|---|---| | Lease liabilities | 91 | 1,912 | 1,059 | 937 | | Other non-current liabilities | – | 1,218 | 1,200 | 1,467 | | **Total non-current liabilities** | **91** | **3,130** | **2,259** | **2,404** | | **Net liabilities** | **(76,374)** | **(341,914)** | **(799,320)** | **(1,303,499)** |

– 435 –

FINANCIAL INFORMATION

ASSETS

Non-Current Assets

Property, Plant and Equipment

Our property, plant and equipment primarily consist of leasehold improvements and office equipment. Our property, plant and equipment increased from US$0.2 million as of December 31, 2022 to US$0.7 million as of December 31, 2023. The increase was due to US$0.2 million in leasehold improvements and US$0.3 million in office equipment, which supported our infrastructure build-out during early commercialization. Our property, plant and equipment further increased to US$1.1 million as of December 31, 2024, mainly due to new leasehold improvements of US$0.4 million and office equipment of US$0.5 million to support headcount expansion and office upgrades, partially offset by depreciation of US$0.5 million recognised for the year. Our property, plant and equipment remained relatively stable at US$1.1 million as of September 30, 2025.

The following table sets forth a breakdown of our property, plant and equipment as of the dates indicated.

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | | | | Leasehold improvements | – | 210 | 394 | 526 | | Office equipment | 231 | 499 | 699 | 608 | | Total | 231 | 709 | 1,093 | 1,134 |

Right-of-Use Assets

Our right-of-use assets primarily consist of leased office premises. Our right-of-use assets increased from US$0.5 million as of December 31, 2022 to US$3.3 million as of December 31, 2023, mainly due to new office lease agreements of US$3.7 million entered into during the year to support business expansion and headcount growth. Our right-of-use assets decreased slightly to US$3.1 million as of December 31, 2024, primarily as a result of depreciation and lease amortization charges of US$1.5 million, partially offset by new office lease agreements of US$1.2 million. Our right-of-use assets decreased to US$2.7 million as of September 30, 2025, primarily due to the amortization of right-of-use assets and the early disposal of right-of-use assets being greater than the new additions.

As at December 31, 2022, 2023, 2024 and September 30, 2025, no indicators of impairment for our non-financial assets were identified because (i) our non-financial assets were not obsolete or physically damaged, and (ii) our actual losses for the years ended December 31, 2022, 2023, 2024 and the nine months ended September 30, 2025 did not exceed the estimated losses.

– 436 –

FINANCIAL INFORMATION

Prepayments, Other Receivables and Other Assets

The non-current asset portion of our prepayments, other receivables and other assets consist of long-term deposits, which are not expected to be utilized or settled within one year. Our prepayments, other receivables and other assets increased from nil as of December 31, 2022 to US$0.4 million as of December 31, 2023. This increase was primarily due to a rise in deposits of US$0.4 million during the period. Our prepayments, other receivables and other assets further increased to US$0.6 million as of December 31, 2024, primarily driven by an increase in deposits from US$0.4 million to US$0.6 million. Our prepayments, other receivables and other assets increased to US$0.7 million as of September 30, 2025, primarily due to the increased rental deposits and property guarantee deposits resulting from the expansion of our leased properties.

Financial Assets at Fair Value through Profit or Loss

The non-current asset portion of our financial assets at fair value through profit or loss primarily consist of long-term investments in structured wealth management products. We did not record any such long-term assets as of December 31, 2022 and 2023. Our financial assets at fair value increased to US$95.3 million as of December 31, 2024, mainly driven by larger cash reserves being deployed into wealth management products purchased from counterparties of well-known financial institutions to enhance returns on idle funds. Our financial assets at fair value through profit or loss decreased to US$70.2 million as of September 30, 2025, primarily due to the maturity of wealth management products with a term exceeding one year. We have established the treasury management policy, in which Chapter 8 has clarified the principles of our investment, the scope of investment and the decision-making process of investment.

According to our treasury management policy, our investment scope includes, but is not limited to, wealth management products investment, equity investment, etc. The investment shall meet the requirements of our development strategy and business objectives. Sufficient market research, risk assessment and due diligence shall be conducted prior to any investment, and a detailed investment plan shall be formed before the investment is made, which shall be approved by the person in charge of the financial department, the person in charge of the investment and financing department and our COO and CEO for approval. Both the financial department and the investment and financing department carry out regular evaluation and preparation of investment reports and reported to our COO and CEO for final approval. Investment in financial assets at fair value through profit or loss after the listing will be subject to the compliance with Chapter 14 of the Listing Rules. We have also established the information disclosure in compliance with the Listing Rules, and we will make corresponding disclosure of all the investments to be disclosed in accordance with the requirements of the policy and the Listing Rules.

Financial Assets at Fair Value through Other Comprehensive Income

Our financial assets at fair value through other comprehensive income primarily consist of long-term investments in equity instruments for non-trading purposes. We did not record any such assets as of December 31, 2022 and 2023. As of December 31, 2024, we recorded US$4.8 million in financial assets at fair value through other comprehensive income, primarily due to equity investments made during the year. Our financial assets at fair value through other comprehensive income increased to US$6.4 million as of September 30, 2025, primarily due to the share price appreciation of the equity investments during the period.

– 437 –

FINANCIAL INFORMATION

Current Assets

Trade Receivables

Our trade receivables represent amounts due from customers for services rendered for which payment has not yet been received.

We did not record any trade receivables as of December 31, 2022. As of December 31, 2023, our trade receivables increased to US$1.3 million, primarily attributable to initial monetization activities during the year as we began commercialization. Our trade receivables further increased to US$7.0 million as of December 31, 2024, driven by continued business expansion and increased customer adoption of our API products. Our trade receivables increased to US$8.1 million as of September 30, 2025, which was consistent with our revenue growth and business expansion.

We typically grant credit periods of 15 to 60 days, and we seek to maintain strict control over our outstanding receivables and has a credit control process to minimize the credit risks.

The following table sets forth an aging analysis of our trade receivables, based on the invoice date and net of loss allowance, as of the dates indicated.

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | | | | Within one year | – | 1,338 | 6,982 | 8,063 |

The following table sets forth our trade receivables turnover days during the periods indicated.

| | For the year ended December 31, | | | For the nine months ended September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (days) | | | | | Trade receivables turnover days (1) | N/A | 41 | 49 | 38 |

Note:

(1) Trade receivables turnover days for a period are calculated as the average of the opening and closing trade receivables balances divided by the revenue for the relevant period, and then multiplied by the number of days in that period.

– 438 –

FINANCIAL INFORMATION

Our trade receivables turnover days increased from 41 days in 2023 to 49 days in 2024, mainly driven by rapid growth of revenue, which increased by 782.2% from US$3.5 million in 2023 to US$30.5 million in 2024. Our trade receivables turnover days decreased from 49 days in 2024 to 38 days for the nine months ended September 30, 2025, primarily due to our effective collection of trade receivables from key customers.

As of December 21, 2025, US$4.2 million, or 52.5% of our trade receivables outstanding as of September 30, 2025 had been subsequently settled. Our operational plan focuses on business expansion in the first three quarters and collection management in the fourth quarter. We do not anticipate any recoverability issues with trade receivables primarily because (i) most of our customers with granted credit days are well-known enterprises with strong credit records in credit assessments and past transactions, (ii) our receivables management policy includes rigorous credit assessment and an aging warning and collection process, (iii) our Management regularly reviews the recoverability of our outstanding balances and when appropriate, provides for impairment of these trade receivables, and (vi) as of November 30, 2025, US$6.0 million, or 84.7% of our trade receivables outstanding as of December 31, 2024 had been subsequently settled. We believe there is no recoverability issue as of the Latest Practicable Date and the provision to our trade receivables as of September 30, 2025 is sufficient.

Prepayments, Other Receivables and Other Assets

The current asset portion of our prepayments, other receivables and other assets primarily consist of prepayments, VAT recoverable, and deposits and other receivables expected to be utilized or settled within one year. The following table sets forth the details of the current asset portion of our prepayments, other receivables and other assets as of the dates indicated.

| | As of December 31, | As of December 31, | As of December 31, | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | | | | Prepayments | 56 | 119 | 394 | 451 | | Value-added tax recoverable | 387 | 3,854 | 7,144 | 1,500 | | Deposits and other receivables | 126 | 405 | 5,932 | 9,438 | | Deferred listing expenses | – | – | – | 422 | | **Total** | **569** | **4,378** | **13,470** | **11,811** |

– 439 –

FINANCIAL INFORMATION

Our current prepayments, other receivables and other assets increased from US$0.6 million as of December 31, 2022 to US$4.4 million as of December 31, 2023, primarily due to (i) an increase in VAT recoverable from US$0.4 million to US$3.9 million, reflecting higher input VAT credits in line with increased procurement activities, and (ii) a rise in deposits and other receivables from US$0.1 million to US$0.4 million. This balance further increased to US$13.5 million as of December 31, 2024, driven by (i) an increase in deposits and other receivables expected to be settled within one year from US$0.4 million to US$5.9 million as of December 31, 2024, and (ii) an increase in VAT recoverable from US$3.9 million to US$7.1 million as our VAT credits continued to increase commensurate to our increased procurement. The current portion of our prepayments, other receivables and other assets decreased to US$11.8 million as of September 30, 2025, primarily due to a decrease in VAT recoverable from US$7.1 million to US$1.5 million as our VAT refund.

As of December 21, 2025, US$3.5 million, or 29.4% of our prepayments, other receivables and other assets outstanding as of September 30, 2025 had been subsequently settled.

Financial Assets at Amortised Costs

Our financial assets at amortised cost primarily consist of short- to medium-term investments in debt instruments, such as fixed-income products, that are held to collect contractual cash flows. We did not record any financial assets at amortised cost as of December 31, 2022 or 2023. Our financial assets at amortised cost increased to US$147.4 million as of December 31, 2024, as we continued to deploy available funds into stable yield-generating instruments in accordance with our cash management strategy. Our financial assets at amortised cost decreased to nil as of September 30, 2025, primarily due to the maturity and redemption of certain of these instruments during the period.

Financial Assets at Fair Value through Profit or Loss

The current portion of our financial assets at fair value through profit or loss primarily consist of investments in short-term structured wealth management products. We recorded financial assets at fair value through profit or loss of US$65.8 million as of December 31, 2022, mainly due to the initiation of investments in short-term yield-oriented products for cash management purposes. Our financial assets at fair value through profit or loss decreased to US$15.8 million as of December 31, 2023, mainly due to redemption and temporary scaling down of investments in these products. This balance increased to US$295.2 million as of December 31, 2024, reflecting increased cash reserves deployed into wealth management products. Our financial assets at fair value through profit or loss further increased to US$644.2 million as of September 30, 2025, primarily due to our increased investments in wealth management products during the period.

We have established the treasury management policy, in which Chapter 8 has clarified the principles of our investment, the scope of investment and the decision-making process of investment.

– 440 –

FINANCIAL INFORMATION

According to our treasury management policy, our investment scope includes, but is not limited to, wealth management products investment, equity investment, etc. The investment shall meet the requirements of our development strategy and business objectives. Sufficient market research, risk assessment and due diligence shall be conducted prior to any investment, and a detailed investment plan shall be formed before the investment is made, which shall be approved by the person in charge of the financial department, the person in charge of the investment and financing department and our COO and CEO for approval. Both the financial department and the investment and financing department carry out regular evaluation and preparation of investment reports and reported to our COO and CEO for final approval. Investment in financial assets at fair value through profit or loss after the listing will be subject to the compliance with Chapter 14 of the Listing Rules. We have also established the information disclosure in compliance with the Listing Rules, and we will make corresponding disclosure of all the investments to be disclosed in accordance with the requirements of the policy and the Listing Rules.

受限现金（Restricted Cash）

我们的受限现金主要由不可用于一般企业用途的现金余额构成，例如信用卡押金或保证金以及银行承兑汇票押金。我们的受限现金从2022年12月31日的220万美元（US$2.2 million）减少至2023年12月31日的零，主要原因是相关融资安排及银行授信到期或结清后，此前质押的存款获得释放。随后，受限现金增加至2024年12月31日的2,730万美元（US$27.3 million），主要原因是为支持运营及增长需要，公司新增质押存款以担保新的融资安排及银行授信。截至2025年9月30日，受限现金减少至2,510万美元（US$25.1 million），主要原因是质押贷款及票据保证金略有减少。

现金及现金等价物（Cash and Cash Equivalents）

截至2022年12月31日、2023年12月31日、2024年12月31日及2025年9月30日，我们的现金及现金等价物分别为470万美元（US$4.7 million）、20,630万美元（US$206.3 million）、28,890万美元（US$288.9 million）及36,260万美元（US$362.6 million）。各期末现金及现金等价物头寸的波动，主要源于经营活动的现金使用及融资活动所筹集的资金。详见"——现金消耗——现金流量分析"。

– 441 –

FINANCIAL INFORMATION

LIABILITIES

Current Liabilities

Interest-Bearing Bank Borrowings

Our interest-bearing bank borrowings represent short-term loans obtained from commercial banks to fund working capital and operational expenditures. We did not have any interest-bearing bank borrowings as of December 31, 2022 or 2023. As of December 31, 2024, we recorded US$19.5 million in interest-bearing bank borrowings, primarily as a result of our decision to raise external debt financing to support increasing operational scale and infrastructure requirements. Our interest-bearing bank borrowings decreased to US$19.1 million as of September 30, 2025, primarily due to the repayment of certain secured bank loans during the period.

Trade and Bills Payables

Our trade and bills payables primarily represent payments due to third-party vendors and service providers for the procurement of cloud services. These payables are not interest-bearing and are generally settled within standard credit terms of 30 to 90 days. Our trade and bills payables increased from US$2.4 million as of December 31, 2022 to US$17.2 million as of December 31, 2023, primarily due to increased procurement of cloud and technical services as we ramped up operations. Trade and bills payables further increased to US$51.2 million as of December 31, 2024, mainly driven by increased vendor activities supporting our expanded AI development and enterprise service offerings. Our trade and bills payables further increased to US$70.2 million as of September 30, 2025, primarily due to continued procurement of cloud services to support the ramp-up of our AI development and enterprise service offerings as we scaled up our business and continually increased investment in cloud services related to foundation model training.

The following table sets forth the aging analysis of our trade and bills payables based on the invoice date as of the dates indicated.

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | Within 1 year | 2,394 | 17,242 | 51,159 | 70,219 | | Over 1 year | – | – | 53 | – | | Total | 2,394 | 17,242 | 51,212 | 70,219 |

– 442 –

FINANCIAL INFORMATION

The following table sets forth our trade and bills payables turnover days for the periods indicated.

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (days) | | | | | Trade and bills payables turnover days (1) | – | 69 | 74 | 92 |

Note: (1) Trade and bills payables turnover days for a period are calculated as the average of the opening and closing trade and bills payables balances divided by total count of cloud service costs in the cost of sales and cloud service expenses related to training activities in research and development expenses for the relevant period, and then multiplied by the number of days in that period.

Our trade and bills payables turnover days increased from 69 days in 2023 to 74 days in 2024, further increased to 92 days for the nine months ended September 30, 2025, primarily due to improved supply chain management.

As of December 21, 2025, US$61.1 million, or 87.0% of our trade and bills payables outstanding as of September 30, 2025 had been subsequently settled.

Other Payables, Accruals and Other Liabilities

Our other payables, accruals and other liabilities mainly include (i) payroll payables, (ii) other tax payables, (iii) convertible bonds, and (iv) other payables and accruals, which include operational accruals and third-party service-related obligations.

The following table sets forth the details of our other payables, accruals and other liabilities as of the dates indicated.

| | As of December 31, | | | As of September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | (US$ in thousands) | | | | | Payroll payables | 1,976 | 5,469 | 10,596 | 9,052 | | Other tax payables | 82 | 303 | 644 | 1,209 | | Convertible bonds | – | – | 14,722 | – | | Other payables and accruals | 268 | 8,969 | 25,550 | 7,061 | | **Total** | **2,326** | **14,741** | **51,512** | **17,322** |

– 443 –

FINANCIAL INFORMATION

Our other payables, accruals and other liabilities increased from US$2.3 million as of December 31, 2022 to US$14.7 million as of December 31, 2023, primarily due to (i) an increase in other payables and accruals from US$0.3 million as of December 31, 2022 to US$9.0 million as of December 31, 2023, resulting from reflecting expanded operational activities, marketing spending, and professional services usage as our commercial operations scaled and (ii) an increase in payroll payables from US$2.0 million to US$5.5 million due to headcount expansion and year-end bonus accruals.

Our other payables, accruals and other liabilities increased from US$14.7 million as of December 31, 2023 to US$51.5 million as of December 31, 2024, primarily due to (i) an increase in other payables and accruals from US$9.0 million to US$25.6 million, (ii) an increase in payroll payables from US$5.5 million to US$10.6 million, in line with workforce expansion and increased bonus accruals, and (iii) additional financing activities via convertible bonds, amounting to US$14.7 million.

Our other payables, accruals and other liabilities significantly decreased to US$17.3 million as of September 30, 2025, primarily due to (i) a decrease in other payables and accruals from US$25.6 million as of December 31, 2024 to US$7.1 million as of September 30, 2025, mainly due to the reduced investment in business promotion expenses, and (ii) a decrease in convertible bonds from US$14.7 million as of December 31, 2024 to nil as of September 30, 2025, resulting from redemption of convertible bonds.

As of December 21, 2025, US$9.7 million, or 56.2% of our other payables, accruals and other liabilities outstanding as of September 30, 2025 had been subsequently settled.

Contract Liabilities

Our contract liabilities primarily consist of short-term advances received from customers for the provision of enterprise services and membership subscriptions. We did not record any contract liabilities as of December 31, 2022. Our contract liabilities increased to US$0.6 million as of December 31, 2023, primarily due to increased advance payments received from customers for membership subscription services rendered at the beginning of 2024. Our contract liabilities further increased to US$1.6 million as of December 31, 2024, mainly due to continued growth in advance receipts for both enterprise service and membership subscriptions booked near the end of the year.

Our contract liabilities increased from US$1.6 million as of December 31, 2024 to US$4.7 million as of September 30, 2025. The increase in contract liabilities was mainly due to continued expansion of both of our monetization channels — AI-native products, as well as Open Platform and other AI-based enterprise services. The significant increase in the number of paying users has led to an increase in the advance payments we received from our individual users and enterprise customers.

As of December 21, 2025, US$2.0 million, or 43.4% of contract liabilities outstanding as of September 30, 2025 had been subsequently settled.

– 444 –

FINANCIAL INFORMATION

Lease Liabilities

The current portion of our lease liabilities represent payment obligations for office premises due within the next 12 months under lease agreements. Our current lease liabilities increased from US$0.3 million as of December 31, 2022 to US$1.2 million as of December 31, 2023, mainly due to new leases entered into to support our operational expansion. Current lease liabilities further increased to US$2.0 million as of December 31, 2024, primarily due to the reclassification of amounts due under existing leases within one year. Our current lease liabilities decreased to US$1.7 million as of September 30, 2025, primarily due to the repayment of lease liabilities exceeding new lease additions and reallocation of lease obligations from non-current to current liabilities.

Convertible Redeemable Preferred Shares

Our convertible redeemable preferred shares consist of our Series Angel preferred shares to Series Pre-B++ preferred shares. These preferred shares are redeemable upon the occurrence of specified events and will be automatically converted into ordinary shares of the Company upon the completion of the Listing. The convertible redeemable preferred shares were classified as current liabilities since the conversion options were not classified as equity and are exercisable at any time at the shareholders' options.

Our convertible redeemable preferred shares increased from US$145.2 million as of December 31, 2022 to US$629.0 million as of December 31, 2023, primarily due to the issuance of Series A and A+ preferred shares totaling US$307.0 million and a US$176.8 million increase in fair value. Our convertible redeemable preferred shares further increased to US$1,581.9 million as of December 31, 2024, mainly driven by the issuance of Series Pre-B and Pre-B+ shares totaling US$739.6 million and additional fair value adjustments of US$213.4 million, reflecting increased valuation of our Group. Our convertible redeemable preferred shares further increased to US$2,321.2 million as of September 30, 2025, primarily due to (i) additional issuances of Series Pre-B+ shares totaling US$35.8 million, (ii) issuances of Series Pre-B++ shares totaling US$390.5 million, and (iii) fair value adjustments of US$313.0 million, reflecting continued increases in our valuation.

See Note 24 to the Accountants' Report in Appendix I to this Prospectus for details of the fair value measurement of our convertible redeemable preferred shares, including the methods and key assumptions used in the measurement.

– 445 –

FINANCIAL INFORMATION

Non-Current Liabilities

Lease Liabilities

The non-current portion of our lease liabilities represent obligations under office lease agreements with payment terms extending beyond 12 months. Our non-current lease liabilities increased from US$91 thousand as of December 31, 2022 to US$1.9 million as of December 31, 2023, primarily due to the execution of long-term office leases to support headcount growth and operational expansion. Our non-current lease liabilities decreased to US$1.1 million as of December 31, 2024, primarily due to regular lease repayments and reclassification of a portion of the obligations into current liabilities as the maturity dates approached. Our non-current lease liabilities decreased to US$0.9 million as of September 30, 2025, mainly due to the approaching maturity of several long-term lease contracts within one year, leading to a reallocation of lease obligations from non-current to current liabilities.

Other Non-Current Liabilities

Our other non-current liabilities consist of government grants received in relation to our research and development and technology innovation activities. We did not record any such liabilities as of December 31, 2022. Our other non-current liabilities increased to US$1.2 million as of December 31, 2023, primarily due to the recognition of government grants received for ongoing research and development projects which were deferred as income to be recognised when the relevant R&D projects completed. Our other non-current liabilities remained stable at US$1.2 million and US$1.5 million as of both December 31, 2024 and September 30, 2025.

KEY FINANCIAL RATIOS

| | For the year ended December 31, | For the year ended December 31, | For the year ended December 31, | For the nine months ended September 30, | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | Revenue growth | N/A | N/A | 782.2% | 174.7% | | Gross margin | N/A | (24.7%) | 12.2% | 23.3% | | Net loss margin | N/A | (7,781.7%) | (1,524.2%) | (958.2%) | | Adjusted net loss margin (non-IFRS measure) | N/A | (2,574.4%) | (800.2%) | (348.6%) | | Research and development expenses growth rate | N/A | 562.9% | 170.0% | 30.0% | | Current ratio | 0.49 | 0.48 | 0.47 | 0.43 |

Note: 1. Current ratio is calculated based on total current assets divided by total current liabilities.

– 446 –

财务信息 流动性与资本资源 我们历史上主要通过融资活动所获得的现金来满足资金需求。全球发售完成后，我们拟通过业务经营产生的现金及全球发售所得款项净额为未来资本需求提供资金。我们目前预计，在不久的将来，用于为我们的运营提供资金的融资渠道不会发生任何变化。截至2022年、2023年、2024年12月31日及2025年9月30日，我们的现金及现金等价物分别为470万美元、2.063亿美元、2.889亿美元及3.626亿美元。

现金消耗 我们的现金消耗是指以下各项的合计金额：（i）经营活动所用现金净额；（ii）资本开支；及（iii）租赁付款。我们历史上的现金消耗在2022年、2023年、2024年及截至2025年9月30日止九个月分别为1,150万美元、6,590万美元、2.607亿美元及2.113亿美元。在整个往绩记录期内，我们的现金消耗持续增加，主要由于我们扩大研发活动规模，经营活动所用现金净额随之增加。

截至2025年9月30日，我们的现金结余为10.462亿美元，包括现金及现金等价物3.626亿美元、按公允价值计入损益的金融资产流动部分6.442亿美元，以及未动用银行融资额度3,940万美元，上述资产均代表可用于为我们运营提供资金的可用流动资金。按公允价值计入损益的金融资产流动部分代表我们根据现金管理政策购买的、将于一年内或更短期限内到期的财富管理产品。该等产品可随时变现，因此被视为我们现金结余的组成部分。未动用银行融资额度代表额外的已承诺流动资金来源，可在现有现金资源动用后提取，以为经营现金流出提供资金。假设未来平均每月现金消耗为2,810万美元（约为截至2024年12月31日止十二个月平均每月现金消耗的1.3倍，但该金额可能因我们的业务发展或模型训练活动投资等各种因素而有所变化），我们估计，在不计入首次公开招股所得款项的情况下，我们的现金结余足以支持我们运营约37个月，即持续至约2028年10月。在计入估计首次公开招股所得款项净额4.687亿美元的情况下（假设按每股151.0港元的发售价（即发售价区间的低端）发行25,389,220股新股，且发售规模调整权及超额配股权均未获行使，并扣除估计首次公开招股开支），我们的现金足以支持我们运营约54个月，即持续至约2030年3月。未来预期平均每月现金消耗2,810万美元约为截至2024年12月31日止十二个月平均每月现金消耗的1.3倍，主要由于经营活动所用现金净额预计将增加，尤其是与研发开支增加相关的现金流出，因为我们将继续扩大研发活动规模以维持模型智能方面的技术领导地位；预计这将部分被截至2025年12月31日止十二个月运营效率的预期改善及销售与分销开支的减少所抵消，主要得益于我们有机用户获取策略的成功。

－447－

FINANCIAL INFORMATION

We have demonstrated the initial evidence of improving profitability. As we continue to roll out top ranking AI-native products and optimize model inference cost, we have significantly improved our gross profit margin, from negative 24.7% in 2023 to 12.2% in 2024, and further to 23.3% in the nine months ended September 30, 2025. Unit gross profit of our AI-native products per MAU increased by 129.5% from the nine months ended September 30, 2024 to the same period in 2025. We have also improved our training efficiency and scalability of our infrastructure, as our R&D expenses as a percentage of revenue have decreased significantly, from over 2,000% in 2023 to 619.1% in 2024, and further decreased from 712.9% in the nine months ended September 30, 2024 to 337.4% in the nine months ended September 30, 2025. The aforementioned trends align with the key features of our recently released foundation models. Launched in June 2025, for example, MiniMax-M1 demonstrates improvements in cost-efficiency across both model training and inference phases. With respect to model training, MiniMax-M1 achieves its cost-efficiency through two key innovations. The primary innovation is CISPO (Clipped IS-weight Policy Optimization), a novel algorithm that materially improves the efficiency of reinforcement learning, a critical component for foundation model training. Comparative tests demonstrate CISPO's superiority over peer-developed reinforcement learning algorithms. In addition, MiniMax-M1's "hybrid-attention" design, a refined approach to processing training data, inherently facilitates the scaling of reinforcement learning. As a direct result of these advancements, the reinforcement learning training for MiniMax-M1 was completed within approximately three weeks, achieving significantly lower costs than most industry peers. Regarding model inference, MiniMax-M1 is powered by a hybrid MoE architecture combined with a "Lightning Attention". This variation of Linear Attention is posited to significantly reduce the computational resources required during the model inference stage. As a result, while achieving top-tier model performance among open-source foundation models, MiniMax-M1 concurrently offers the industry's best cost-effectiveness.

– 448 –

FINANCIAL INFORMATION

Cash Flow Analysis

The following table sets forth our cash flows for the periods indicated.

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | | | | | Net cash flows used in operating activities | (11,019) | (64,455) | (258,483) | (195,596) | (209,396) | | Net cash flows used in investing activities | (35,156) | (40,320) | (431,300) | (630,463) | (126,231) | | Net cash flows generated from financing activities | 49,786 | 306,243 | 771,092 | 718,827 | 407,913 | | Net increase/(decrease) in cash and cash equivalents | 3,611 | 201,468 | 81,309 | (107,232) | 72,286 | | Cash and cash equivalents at the beginning of the year/period | 994 | 4,691 | 206,295 | 206,295 | 288,912 | | Effect of foreign exchange differences, net | 86 | 136 | 1,308 | 500 | 1,449 | | Cash and cash equivalents at the end of the year/period | 4,691 | 206,295 | 288,912 | 99,563 | 362,647 |

– 449 –

FINANCIAL INFORMATION

Net Cash Flows Used in Operating Activities

Net cash flows used in operating activities in the nine months ended September 30, 2025 was US$209.4 million, which primarily consists of loss before tax of US$512.0 million, adjusted for certain non-cash and non-operating items. Adjustments for such non-cash and non-operating items primarily include (i) a fair value loss on financial liabilities of US$313.5 million, and (ii) Share-based payment expenses of US$8.6 million. The amount was further adjusted by changes in working capital, primarily including (i) an increase in trade and bills payables of US$19.0 million, and (ii) an increase in contract liabilities of US$3.1 million, partially offset by a decrease in other payables, accruals and other liabilities of US$22.9 million.

Net cash flows used in operating activities in 2024 was US$258.5 million, which primarily consists of loss before tax of US$465.2 million, adjusted for certain non-cash and non-operating items. Adjustments for such non-cash and non-operating items primarily include fair value loss on financial liabilities of US$214.2 million. The amount was further adjusted by changes in working capital, primarily including (i) an increase in trade and bills payables of US$34.0 million, and (ii) an increase in other payables, accruals and other liabilities of US$21.0 million, partially offset by an increase in restricted cash of US$27.3 million.

Net cash flows used in operating activities in 2023 was US$64.5 million, which consists primarily of loss before tax of US$269.2 million, adjusted for certain non-cash and non-operating items. Adjustments for such non-cash and non-operating items primarily include fair value loss on financial liabilities of US$176.8 million. The amount was further adjusted by changes in working capital, primarily including (i) an increase in trade and bills payables of US$14.8 million, and (ii) an increase in other payables, accruals and other liabilities of US$12.6 million, partially offset by an increase in prepayments, other receivables and other assets of US$4.2 million.

Net cash flows used in operating activities in 2022 was US$11.0 million, which consists primarily of loss before tax of US$73.7 million, adjusted for certain non-cash and non-operating items. Adjustments for such non-cash and non-operating items primarily include (i) fair value loss on financial liabilities of US$60.5 million. The amount was further adjusted by changes in working capital, primarily including (i) an increase in other payables, accruals and other liabilities of US$2.2 million, and (ii) an increase in trade and bills payables of US$2.4 million, partially offset by an increase in prepayments, other receivables and other assets of US$0.5 million.

– 450 –

FINANCIAL INFORMATION

We are still at a nascent stage in terms of monetization and commercialization as historically we have been largely focused on developing and training our AI foundation models, growing our customer base, and expanding our AI-native product suite, rather than seeking immediate financial return or profitability, thus incurring net operating cash outflows throughout the Track Record Period. It is industry norm that the initial research and training of foundation model will take approximately 2-3 years, where investment in research and development is needed (mainly the cloud and services expenses related to training) before the models and products could generate commercial value in scale. In 2023, 2024 and the nine months ended September 30, 2025, our cloud services expense related to training were US$47.2 million, US$140.6 million and US$142.4 million. As model intelligence continues to elevate and gradually unlock more application scenarios, we have been gradually monetizing our models and products. For example, we commenced revenue generation of Open Platform in May 2023, Talkie in June 2023, Hailuo AI in October 2024, and MiniMax in January 2025. As we scaled up operations, our revenue increased from US$3.5 million in 2023 to US$30.5 million in 2024, and for the nine months ended September 30, 2025, our revenue further increased to US$53.4 million, compared to US$19.5 million in the nine months ended September 30, 2024. We also significantly improved our gross profit margin, from negative 24.7% in 2023 to 12.2% in 2024, and further to 23.3% in the nine months ended September 30, 2025, primarily driven by advancement in the intelligence level of our models, improved model and system efficiency, optimization of infrastructure allocation, and increased scale of revenue relative to compute intensity, in line with our strategy to enhance efficiency of our AI infrastructure. As a result, we significantly narrowed our adjusted net loss (non-IFRS measure) as a percentage of revenue, from over 2,500% in 2023 to 800.2% in 2024 and further to 348.6% in the nine months ended September 30, 2025.

In the future, we aim to continue to enhance our profitability and improve our net operating cash outflows position through the following focus areas: (i) leveraging the rapid growth of the foundation model industry, (ii) continuing to enhance foundation model intelligence levels, (iii) enhancing the affordability of our AI technologies, (iv) broadening monetization of our AI-native product suite, and (v) optimizing organizational efficiency and scalability. Please refer to "Path to the Commercialization of our Specialist Technology Products" for our detailed strategies.

Net Cash Flows Used in Investing Activities

Net cash flows generated from investing activities in the nine months ended September 30, 2025 was US$126.2 million, which consists primarily of (i) purchase of financial assets at amortized cost of US$2,380.3 million, and (ii) purchases of financial assets at fair value through profit or loss of US$1,822.8 million, partially offset by (i) proceeds from disposal of financial assets at amortized cost of US$2,531.5 million, and (ii) proceeds from disposal of financial assets at fair value through profit or loss of US$1,519.4 million.

– 451 –

财务信息

2024年投资活动所用净现金流量为4.313亿美元，主要包括：(i) 购买以公允价值计量且其变动计入当期损益的金融资产22.104亿美元，及(ii) 购买以摊余成本计量的金融资产11.218亿美元，部分抵销于：(i) 出售以公允价值计量且其变动计入当期损益的金融资产所得款项18.513亿美元，及(ii) 出售以摊余成本计量的金融资产所得款项9.824亿美元。

2023年投资活动所用净现金流量为4,030万美元，主要包括：(i) 定期存款存入9,040万美元，及(ii) 购买以公允价值计量且其变动计入当期损益的金融资产8,530万美元，部分抵销于出售以公允价值计量且其变动计入当期损益的金融资产所得款项1.361亿美元。

2022年投资活动所用净现金流量为3,520万美元，主要包括购买以公允价值计量且其变动计入当期损益的金融资产4,600万美元，部分抵销于出售以公允价值计量且其变动计入当期损益的金融资产所得款项1,110万美元。

融资活动产生的净现金流量

截至2025年9月30日止九个月，融资活动产生的净现金流量为4.079亿美元，主要来自发行可转换可赎回优先股所得款项4.263亿美元。

2024年融资活动产生的净现金流量为7.711亿美元，主要来自发行可转换可赎回优先股所得款项7.396亿美元。

2023年融资活动产生的净现金流量为3.062亿美元，主要来自发行可转换可赎回优先股所得款项3.070亿美元。

2022年融资活动产生的净现金流量为4,980万美元，主要来自发行可转换可赎回优先股所得款项5,000万美元。

－452－

FINANCIAL INFORMATION CASH OPERATING COSTS

The following table sets forth key information relating to our cash operating costs for the periods indicated:

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | | | | | Workforce employment (1) | 4,406 | 15,861 | 38,270 | 30,793 | 43,397 | | R&D costs (2) | 2,697 | 37,517 | 114,622 | 93,120 | 131,872 | | Direct procurement costs (3) | – | 3,767 | 23,562 | 13,545 | 36,855 | | Marketing and promotion expenses (4) | 347 | 14,355 | 70,791 | 52,397 | 53,492 | | Administrative expenses (5) | 1,571 | 3,426 | 6,799 | 3,146 | 7,908 | | Non-income taxes and other charges | 5 | 27 | 43 | 31 | 87 | | **Total** | **9,026** | **74,953** | **254,087** | **193,032** | **273,611** |

Notes:

(1) Cash operating costs relating to workforce employment represent the sum of employee benefit expenses under R&D expenses, administrative expenses, cost of sales and selling and marketing expenses (excluding share-based compensation which is non-cash in nature), adjusted for changes in working capital relating to employee benefit expenses as of previous and current year end under the above operating expenses.

(2) R&D costs under cash operating costs represent R&D expenses (excluding employee benefit expenses and non-cash items under R&D expenses), adjusted for changes in working capital relating to R&D activities as of previous and current year end.

(3) Cash operating costs relating to direct procurement costs, including cloud services related to inference, as well as platform commission fees, adjusted for changes in working capital relating to cost of sales as of previous and current year end.

(4) Cash operating costs relating to marketing and promotion expenses represent selling and marketing expenses (excluding employee benefit expenses and non-cash items under selling and marketing expenses), adjusted for changes in working capital relating to sales and marketing activities as of previous and current year end.

(5) Cash operating costs relating to administrative activities represent administrative expenses (excluding employee benefit expenses and non-cash items under administrative expenses), adjusted for changes in working capital relating to administrative activities as of previous and current year end.

– 453 –

FINANCIAL INFORMATION

Net Current Assets/Liabilities

The following table sets forth our current assets and current liabilities as of the dates indicated.

| | As of December 31, | As of December 31, | As of December 31, | As of September 30, | As of November 30, | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 (unaudited) |

**Current Assets**

| | 2022 | 2023 | 2024 | As of September 30, 2025 | As of November 30, 2025 | |---|---|---|---|---|---| | Trade receivables | – | 1,338 | 6,982 | 8,063 | 9,283 | | Prepayments, other receivables and other assets | 569 | 4,378 | 13,470 | 11,811 | 15,038 | | Financial assets at amortized costs | – | – | 147,444 | – | 30,006 | | Financial assets at fair value through profit or loss | 65,791 | – | 295,220 | 644,154 | 731,855 | | Time deposits | – | 15,802 | 26,327 | – | 13,744 | | Restricted cash | 2,221 | 91,698 | 27,293 | 25,097 | 18,744 | | Cash and cash equivalents | 4,691 | 206,295 | 288,912 | 362,647 | 191,619 | | **Total current assets** | **73,272** | **319,511** | **805,648** | **1,051,772** | **1,010,289** |

**Current Liabilities**

| | 2022 | 2023 | 2024 | As of September 30, 2025 | As of November 30, 2025 | |---|---|---|---|---|---| | Interest-bearing bank borrowings | – | – | 19,455 | 19,102 | 14,130 | | Trade and bills payables | 2,394 | 17,242 | 51,212 | 70,219 | 62,121 | | Other payables, accruals and other liabilities | 2,326 | 14,741 | 51,512 | 17,322 | 21,950 | | Contract liabilities | – | 559 | 1,553 | 4,657 | 6,876 | | Lease liabilities | 349 | 1,248 | 1,964 | 1,694 | 1,688 | | Convertible redeemable preferred shares | 145,175 | 629,001 | 1,581,949 | 2,321,193 | 3,091,653 | | **Total current liabilities** | **150,244** | **662,791** | **1,707,645** | **2,434,187** | **3,198,418** |

| **Net current liabilities** | **(76,972)** | **(343,280)** | **(901,997)** | **(1,382,415)** | **(2,188,129)** |

– 454 –

FINANCIAL INFORMATION

Our net current liabilities increased from US$77.0 million as of December 31, 2022 to US$343.3 million as of December 31, 2023, primarily due to (i) an increase in convertible redeemable preferred shares from US$145.2 million as of December 31, 2022 to US$629.0 million as of December 31, 2023, reflecting new issuances of preferred shares and additional fair value changes, (ii) a decrease in financial assets at fair value through profit or loss from US$65.8 million as of December 31, 2022 to US$15.8 million as of December 31, 2023, mainly attributable to a partial redemption of previously held liquid financial products and reallocation of capital, (iii) an increase in trade and bills payables from US$2.4 million as of December 31, 2022 to US$17.2 million as of December 31, 2023, primarily due to higher procurement of cloud services, and (iv) an increase in other payables, accruals and other liabilities from US$2.3 million as of December 31, 2022 to US$14.7 million as of December 31, 2023 as we scaled up operations, partially offset by (i) an increase in cash and cash equivalents from US$4.7 million as of December 31, 2022 to US$206.3 million as of December 31, 2023, and (ii) the recognition of time deposits of US$91.7 million, which did not exist in 2022.

Our net current liabilities increased from US$343.3 million as of December 31, 2023 to US$902.0 million as of December 31, 2024, primarily due to (i) an increase in convertible redeemable preferred shares from US$629.0 million as of December 31, 2023 to US$1,581.9 million as of December 31, 2024 as we completed further rounds of equity financing in 2024 and recognised fair value changes, (ii) an increase in trade and bills payables from US$17.2 million as of December 31, 2023 to US$51.2 million as of December 31, 2024, primarily due to increased service received, and (iii) an increase in other payables, accruals and other liabilities from US$14.7 million as of December 31, 2023 to US$51.5 million as of December 31, 2024. This was partially offset by (i) an increase in financial assets at fair value through profit or loss from US$15.8 million as of December 31, 2023 to US$295.2 million as of December 31, 2024, mainly attributable to increased investment in structured wealth management products, and (ii) an increase in financial assets at amortized costs from nil as of December 31, 2023 to US$147.4 million as of December 31, 2024, as we began investing in fixed-income instruments.

Our net current liabilities increased from US$902.0 million as of December 31, 2024 to US$1,382.4 million as of September 30, 2025, primarily due to (i) an increase in convertible redeemable preferred shares from US$1,581.9 million as of December 31, 2024 to US$2,321.2 million as of September 30, 2025, mainly due to the increase in the fair value of our preferred shares, reflecting our enhanced operating performance and business prospects and the issuance of Series Pre-B++ shares, and (ii) a decrease in financial assets at amortized cost from US$147.4 million as of December 31, 2024 to nil as of September 30, 2025, primarily attributable to the maturity and subsequent redemption of certain short-term investments, and (iii) an increase in trade and bills payables from US$51.2 million as of December 31, 2024 to US$70.2 million as of September 30, 2025, primarily driven by the increased procurement of goods and services as well as increased spending on cloud services for model training to support our business expansion. This was partially offset by (i) an increase in financial assets at fair value through profit or loss from US$295.2 million as of December 31, 2024 to US$644.2 million as of September 30, 2025, primarily due to our increased investment amount in diversified financial

– 455 –

FINANCIAL INFORMATION

instruments, and (ii) a decrease in other payables, accruals and other liabilities from US$51.5 million as of December 31, 2024 to US$17.3 million as of September 30, 2025, mainly attributable to the repayment of convertible bonds and the reduction in marketing activities and procurement expenses.

Our net current liabilities increased from US$1,382.4 million as of September 30, 2025 to US$2,188.1 million as of November 30, 2025, primarily due to (i) an increase in convertible redeemable preferred shares from US$2,321.2 million as of September 30, 2025 to US$3,091.7 million as of November 30, 2025 mainly due to the increase in the fair value of our preferred share, reflecting our enhanced operating performance and business prospectus, (ii) a decrease in cash and cash equivalents from US$362.6 million as of September 30, 2025 to US$191.6 million as of November 30, 2025 mainly due to reallocation of capital to financial assets at fair value through profit or loss. This was partially offset by (i) an increase in financial assets at amortized costs from nil as of September 30, 2025 to US$30.0 million as of November 30, 2025 as we continued to deploy available funds into stable yield-generating instruments in accordance with our cash management strategy, and (ii) an increase in financial assets at fair value through profit or loss from US$644.2 million as of September 30, 2025 to US$731.9 million as of November 30, 2025.

As of November 30, 2025, we had net current liabilities of US$2,188.1 million, primarily because of convertible redeemable preferred shares totaling US$3,091.7 million, which are classified as current liabilities. Nevertheless, as these convertible redeemable preferred shares will be re-designated from financial liabilities to equity as a result of the automatic conversion into ordinary shares upon Listing, our net current liabilities position will turn into a net current assets position. Looking forward, we expect our net current liabilities position to improve through a combination of the following measures: (i) upon Listing, all of our outstanding Preferred Shares currently classified as current liabilities will be converted into ordinary shares and reclassified as equity, which is expected to result in a net current assets position; (ii) the estimated net proceeds from the Global Offering will further strengthen our cash and bank balances and may be applied, where appropriate, to fund our working capital needs, including the settlement of current operating liabilities as they fall due; (iii) as we continue to scale the commercialization of our foundation models and AI-native products, we expect increasing cash inflows from usage-based and contract-based revenue, which, together with disciplined cost management, are expected to improve our operating cash flows; and (iv) we will continue to actively manage our working capital by, among others, enhancing our billing and collection processes, encouraging upfront or periodic prepayments for certain enterprise solutions where commercially appropriate, and negotiating appropriate payment terms with major suppliers and service providers, so as to maintain a sufficient level of cash and near-cash financial assets (such as short-term fixed income products) to meet our current liabilities when due.

– 456 –

FINANCIAL INFORMATION INDEBTEDNESS The following table sets forth our indebtedness as of the dates indicated.

As of December 31,

Current Interest-bearing bank borrowings Lease liabilities Convertible bonds included in other payables, accruals and other liabilities Convertible redeemable preferred shares Non-Current Lease liabilities Total

As of September 30,

As of November 30,

2022

2023

2024

2025

2025

USD'000

USD'000

USD'000

USD'000

USD'000 (unaudited)

| | 2022 | 2023 | 2024 | As of September 30, 2025 | As of November 30, 2025 | |---|---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 (unaudited) | | **Current** | | | | | | | Interest-bearing bank borrowings | – | – | 19,455 | 19,102 | 14,130 | | Lease liabilities | 349 | 1,248 | 1,964 | 1,694 | 1,688 | | Convertible bonds included in other payables, accruals and other liabilities | – | – | 14,722 | – | – | | Convertible redeemable preferred shares | 145,175 | 629,001 | 1,581,949 | 2,321,193 | 3,091,653 | | **Non-Current** | | | | | | | Lease liabilities | 91 | 1,912 | 1,059 | 937 | 738 | | **Total** | **145,615** | **632,161** | **1,619,149** | **2,342,926** | **3,108,209** |

For details of our interest-bearing bank and other borrowings and lease liabilities during the Track Record Period, see "— Discussion of Certain Key Items from Our Consolidated Statements of Financial Position". As of November 30, 2025, our committed unutilized bank facilities amounted to US$44.9 million. During the Track Record Period and up to the date of this Prospectus, we did not have any contingent liabilities.

Our Directors confirm that as of the Latest Practicable Date, the agreements under our borrowings did not contain any covenants that would have a material and adverse effect on our ability to obtain additional borrowings or issue debt or equity securities in the future. Our Directors further confirm that we had no defaults in bank and other borrowings, nor did we breach any covenants (that were not waived) during the Track Record Period and up to the Latest Practicable Date. Additionally, our Directors confirm that, during the Track Record Period and up to the Latest Practicable Date, we did not experience any difficulties in obtaining credit facilities, nor any withdrawal of facilities or requests for early repayment.

Except as otherwise disclosed under the sections titled "— Indebtedness" and "— Contractual Obligations," as of November 30, 2025, the latest practicable date for determining our indebtedness, we did not have any material bank overdrafts, loans, or other similar indebtedness, liabilities under acceptances or acceptance credits, debentures, mortgages, charges, other recognised lease liabilities, guarantees, or other material contingent liabilities.

Our Directors confirm that there have been no material changes in our indebtedness since November 30, 2025 and up to the date of this Prospectus.

– 457 –

FINANCIAL INFORMATION

RESEARCH AND DEVELOPMENT EXPENDITURE AND TOTAL OPERATING EXPENDITURE

During the Track Record Period, we did not capitalize internal development costs as intangible assets. The following table sets forth our annual and total research and development expenditure for the periods indicated:

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | | | |

| | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | |---|---|---|---|---|---| | Research and development expenses | 10,560 | 70,002 | 188,979 | 138,684 | 180,312 | | Adjustments: | | | | | | | Add: intangible assets acquired from third parties and capitalized (1) | – | – | – | – | – | | Less: amortization expense of capitalized intangible assets included in research and development expenditure (1) | – | – | – | – | – | | Annual research and development expenditure | 10,560 | 70,002 | 188,979 | 138,684 | 180,312 |

Total research and development expenditure for the three financial years prior to the Global Offering: 269,541

– 458 –

FINANCIAL INFORMATION

The following table sets forth our annual and total operating expenditure for the periods indicated:

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | | | |

| | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | |---|---|---|---|---|---| | Research and development expenses | 10,560 | 70,002 | 188,979 | 138,684 | 180,312 | | Selling and distribution expenses | 587 | 22,827 | 86,995 | 53,389 | 39,325 | | Administrative expenses | 3,213 | 7,615 | 14,384 | 9,610 | 22,074 | | **Adjustments:** | | | | | | | Add: intangible assets acquired from third parties and capitalized | – | – | – | – | – | | Less: amortization expense of capitalized intangible assets included in research and development expenditure | – | – | – | – | – |

Total operating expenditure for the three financial years prior to the Global Offering: **405,162**

– 459 –

FINANCIAL INFORMATION

The following table sets forth our annual research and development expenditure ratio and total research and development expenditure ratio for the periods indicated:

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | Annual research and development expenditure ratio (1) | 73.5% | 69.7% | 65.1% | 68.8% | 74.6% | | Total research and development expenditure ratio (2) | | | | | 66.5% |

(1) Calculated by dividing annual research and development expenditure by annual total operating expenditure.

(2) Calculated by dividing total research and development expenditure for the three financial years prior to the Global Offering by total operating expenditure for the three financial years prior to the Global Offering.

CAPITAL EXPENDITURES

Our historical capital expenditures primarily consist of expenditures for plant and equipment, specifically leasehold improvements and office equipment. The following table sets forth our capital expenditures for the periods indicated.

| | For the year ended December 31, | | | For the nine months ended September 30, | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | (US$ in thousands) | | Property, Plant and Equipment | 256 | 697 | 759 | 496 | 479 | | Total | 256 | 697 | 759 | 496 | 479 |

We will continue to make capital expenditures to support the expected growth of our business and our expansion plans. We intend to fund these future capital expenditures with financial resources available to us, including our existing cash and bank balances, cash flows generated from our financing activities and net proceeds from the Global Offering.

– 460 –

FINANCIAL INFORMATION CONTRACTUAL OBLIGATIONS Capital Commitments We did not have significant capital commitments as of December 31, 2022, 2023, 2024 and September 30, 2024 and 2025. RELATED PARTY TRANSACTIONS We enter into transactions with our related parties from time to time. For details of our related party transactions, see Note 29 to the Accountants' Report included in Appendix I to this Prospectus. Our Directors are of the view that each of the related party transactions set out in Note 29 to the Accountants' Report included in Appendix I to this Prospectus was conducted in the ordinary course of business on an arm's length basis and with normal commercial terms between the relevant parties. Our Directors are also of the view that our related party transactions during the Track Record Period would not distort our track record results or cause our historical results to become non-reflective of our future performance. OFF-BALANCE SHEET COMMITMENTS AND ARRANGEMENTS As of the Latest Practicable Date, we had not entered into any financial guarantees or other commitments to guarantee the payment obligations of any third parties. We have not entered into any derivative contracts that are indexed to our shares and classified as Shareholder's equity or that are not reflected in our consolidated financial statements. Furthermore, we do not have any retained or contingent interest in assets transferred to an unconsolidated entity that serves as credit, liquidity or market risk support to such entity. We do not have any variable interest in any unconsolidated entity that provides financing, liquidity, market risk or credit support to us or engages in leasing, hedging or product development services with us. FINANCIAL RISKS DISCLOSURE We are exposed to a variety of market and other financial risks, including foreign currency risk, credit risk and liquidity risk. We manage and monitor these exposures to ensure appropriate measures are implemented in a timely and effective manner. Foreign currency risk Foreign currency risk is the risk of loss resulting from changes in foreign currency exchange rates. Such exposures arise from sales and purchases by operating units in currencies other than the units' functional currencies. We seek to limit our exposure to foreign currency risk by minimising its net foreign currency position.

– 461 –

FINANCIAL INFORMATION

信贷风险

我们仅与公认且信誉良好的第三方进行交易。我们的政策是，所有希望以信贷条款进行交易的客户均须接受信用核实程序。此外，应收款项余额受到持续监控，我们面临的坏账风险并不重大。

流动性风险

我们通过监控流动比率来评估流动性风险敞口，该比率通过将流动资产与流动负债进行比较来计算。我们的流动性主要取决于我们维持足够经营现金流入以在债务到期时履行债务义务的能力，以及我们获得外部融资以满足已承诺的未来资本支出的能力。

股息

自成立以来，我们及我们的任何附属公司均未派付或宣派任何股息。于记录期后至本招股说明书日期，我们未向股东宣派任何股息。

截至最后实际可行日期，我们并无正式的股息政策或固定的股息分配比率。任何股息的宣派及派付以及股息金额将受我们的组织章程细则及开曼群岛公司法的约束。我们目前没有任何股息政策指导我们的股息宣派或派付。我们的董事会有权酌情派付中期股息及建议股东派付末期股息，并将取决于多项因素，包括我们的盈利、资本需求、整体财务状况及合约限制。我们可通过普通决议决定以任何货币宣派股息，并授权从公司合法可动用的资金中派付股息，但须符合以下条件：(i) 股息不得超过我们董事会建议的金额；及 (ii) 股息仅可从公司已实现或未实现的利润、股份溢价账户中派付，或在法律允许的情况下以其他方式派付。根据我们的开曼群岛法律顾问的意见，依据开曼群岛公司法，开曼群岛公司可从利润及/或股份溢价账户中派付股息，但在任何情况下，若派付股息将导致公司在正常业务过程中无法于到期时偿还债务，则不得派付股息。鉴于本招股说明书所披露的累计亏损，我们在可预见的将来不太可能有资格从利润中派付股息。然而，我们可从股份溢价账户派付股息，除非此类股息的派付将导致我们的公司在正常业务过程中无法于到期时偿还债务。不保证任何年度均会宣派任何金额的股息予以分派。根据我们的开曼群岛法律顾问的意见，我们是一家依据开曼群岛法律注册成立的控股公司，依据该法律，净负债的财务状况并不妨碍我们向股东宣派及派付股息，因为股息仍可从股份溢价账户宣派及派付，尽管

— 462 —

FINANCIAL INFORMATION

our profitability, provided that our memorandum and articles of association do not prohibit such payment and our Company is able to pay its debts as they fall due in the ordinary course of business immediately after such payment.

If we pay dividends in the future, in order for us to distribute dividends to our Shareholders, we will rely to some extent on any dividends distributed by our PRC subsidiaries. Any dividend distributions from our PRC subsidiaries to us will be subject to PRC withholding tax. In addition, regulations in the PRC currently permit payment of dividends of a PRC company only out of accumulated distributable after-tax profits as determined in accordance with its articles of association and the accounting standards and regulations in China. See "Risk Factors — Risks Related to Doing Business in the Geographic Markets in Which We Operate" in this Prospectus.

WORKING CAPITAL SUFFICIENCY

Our Directors are of the opinion that, taking into account (i) the financial resources available to our Group, including the cash and cash equivalents of US$362.6 million as well as the current portion of financial assets at fair value through profit and loss US$644.2 million, as of September 30, 2025, (ii) the estimated net proceeds from the Global Offering and (iii) expected net cash used in operating activities and capital expenditures, we have sufficient working capital to cover at least 125% of our costs, including research and development costs and administrative expenses for the next 12 months from the date of this Prospectus.

DISTRIBUTABLE RESERVES

As of September 30, 2025, our Company did not have any distributable reserves.

LISTING EXPENSES

Our listing expenses mainly include (i) underwriting-related expenses, such as underwriting fees and commissions, and (ii) non-underwriting-related expenses, comprising professional fees paid to our legal advisors and reporting accountants for their services rendered in relation to the Listing and the Global Offering, and other fees and expenses. Assuming full payment of the discretionary incentive fee, the estimated total listing expenses (based on the mid-point of the Offer Price range and assuming that the Offer Size Adjustment Option and the Over-allotment Option are not exercised) for the Global Offering are approximately HK$193.2 million, accounting for approximately 4.8% of our gross proceeds. Among such estimated total listing expenses, we expect to pay underwriting-related expenses of HK$133.0 million, professional fees for our legal advisors and reporting accountants of HK$40.3 million and other fees and expenses of HK$19.9 million. During the Track Record Period, the listing expenses charged to our consolidated statements of profit or loss were US$3.7 million (HK$28.6 million) and the issuance costs which were recognized as prepayments and are expected to be deducted from equity upon the Listing, were US$0.4

– 463 –

FINANCIAL INFORMATION

million (HK$3.3 million). After the Track Record Period approximately HK$26.7 million is expected to be charged to our consolidated statements of profit or loss, and approximately HK$134.7 million is expected to be accounted for as a deduction from equity upon the Listing.

UNAUDITED PRO FORMA STATEMENT OF ADJUSTED CONSOLIDATED NET TANGIBLE ASSETS

The following unaudited pro forma adjusted consolidated net tangible assets of our Group, prepared in accordance with Rule 4.29 of the Listing Rules and with reference to Accounting Guideline 7 Preparation of Pro Forma Financial Information for inclusion in Investment Circulars issued by the Hong Kong Institute of Certified Public Accountants, is for illustration purposes only and is set out here to illustrate the effect of the Global Offering on the consolidated net tangible assets of our Group attributable to owners of the parent as of September 30, 2025, as if the Global Offering had taken place on September 30, 2025.

The unaudited pro forma statement of adjusted consolidated net tangible assets of our Group has been prepared for illustrative purposes only and, because of its hypothetical nature, it may not give a true picture of the consolidated net tangible assets of our Group to owners of the parent had the Global Offering been completed as of September 30, 2025 or at any future dates. The unaudited pro forma statement of adjusted consolidated net tangible liabilities does not form part of the Accountants' Report.

| | Unadjusted audited consolidated net tangible liabilities attributable to the owners of our Group as of September 30, 2025 | Estimated net proceeds from the Global Offering | Estimated impact related to the reclassification of convertible redeemable preferred shares upon Listing | Unaudited pro forma adjusted consolidated net tangible assets attributable to owners of our Group as of September 30, 2025 | Unaudited pro forma adjusted consolidated net tangible assets attributable to owners of our Group per Share as of September 30, 2025 | | |---|---|---|---|---|---|---| | | US$'000 | US$'000 | US$'000 | US$'000 | US$ | HK$ | | | Note (1) | Note (2) | Note (3) | Note (3) | Note (4) | | | Based on an Offer Price of HK$151.00 per Share | (1,303,499) | 472,382 | 2,321,193 | 1,490,076 | 4.88 | 37.96 | | Based on an Offer Price of HK$158.00 per Share | (1,303,499) | 494,423 | 2,321,193 | 1,512,117 | 4.95 | 38.52 | | Based on an Offer Price of HK$165.00 per Share | (1,303,499) | 516,464 | 2,321,193 | 1,534,158 | 5.02 | 39.08 |

– 464 –

财务信息

注释： (1)

截至2025年9月30日，本集团归属于本公司股东的综合有形净负债，乃根据本招股说明书附录I所载会计师报告中截至2025年9月30日归属于本公司股东的综合净负债1,303,499,000美元计算得出。

(2)

全球发售的估计净所得款项乃基于每股HK$151.00、HK$158.00及HK$165.00的估计低端、中间值及高端发售价，扣除由本公司支付的承销费用、佣金及其他相关开支后计算，且不计入因行使发售规模调整权及超额配股权而可能发行的任何股份。

(3)

就未经审核的备考财务信息而言，考虑到上市后可转换可赎回优先股重新分类的估计影响，归属于本公司股东的未经审核备考调整后有形净资产将增加2,321,193,000美元，即截至2025年9月30日可转换可赎回优先股的公允价值。于上市及完成全球发售后，所有可转换可赎回优先股将自动转换为股份。该等可转换可赎回优先股将由负债重新分类为权益。由负债重新分类为权益的金额将为全球发售当日优先股的公允价值。

(4)

归属于本公司股东的未经审核备考调整后综合有形净资产每股金额，乃经参考前述各段落（上述注释2及3）所述调整后得出，并基于假设全球发售及可转换可赎回优先股和普通股所产生的金融负债重新分类为权益已于2025年9月30日完成，届时已发行305,447,288股股份，且不计及发售规模调整权及超额配股权的行使。

(5)

就本未经审核备考调整后综合有形净资产而言，归属于本公司股东的未经审核备考调整后综合有形净资产每股金额（以美元计）按最后实际可行日期的现行汇率1.00美元兑7.7805港元换算为港元。本文件并不就上述港元金额已按该汇率或任何其他汇率或以任何方式换算或可换算为美元（或反之亦然）作出任何陈述。

(6)

未就本集团未经审核备考调整后综合有形净资产作出任何其他调整，以反映2025年9月30日后的任何交易业绩或所订立的其他交易。

请参阅"附录II — 未经审核备考财务信息"以了解进一步详情。

无重大不利变化

本公司董事已确认，直至本招股说明书日期，自2025年9月30日（即本招股说明书附录I所载会计师报告所涵盖期间的截止日期）起，本集团的财务、营运或贸易状况、债务、或有负债或前景均未发生任何重大不利变化，且自2025年9月30日起亦未发生任何可能对本招股说明书附录I所载会计师报告所示信息产生重大影响的事件。

根据上市规则第13.13至13.19条的披露

本公司董事确认，除本节所披露的应收关联方款项外，截至最后实际可行日期，并不存在任何须根据上市规则第13.13至13.19条作出披露的情况。

— 465 —

募集资金用途

我们估计，全球发售所得款项净额约为38.183亿港元，扣除承销佣金、费用及我们就全球发售应付的估计开支后，假设超额配股权调整选项或超额配售选项均未获行使，且发售价为每股发售股份158.00港元（即本招股说明书所述发售价指示区间的中间值）。

所得款项净额的约90%，即34.364亿港元，将于未来五年用于我们的研究与开发，包括我们基础模型及AI原生产品的开发。根据我们的战略，我们拟将所得款项净额用于以下用途，但须视乎我们不断变化的业务需求及市场状况而作出调整：

• 基础模型的开发。为巩固我们的技术领导地位，我们计划于未来五年将所得款项净额的约70.0%，即26.728亿港元，用于基础模型的研究与开发，包括对AI基础设施及研发人才的投入。我们的核心竞争优势在于我们在整个基础模型技术栈上的创新。提升基础模型的智能化程度、效率及可扩展性，对于增强我们在全球市场的竞争力和差异化至关重要。我们观察到，我们模型的竞争力与我们模型的市场定价及需求直接相关。

(i) 增强模型研发的AI基础设施。我们计划将所得款项净额的约50.0%，即19.091亿港元，用于增强支持我们基础模型开发的AI基础设施*。该款项将全部用于研发开支。由于我们所在行业依赖不断增长的算力来训练先进的AI模型，我们相信持续升级我们的AI基础设施至关重要。我们的研发活动，例如训练大型基础模型、试验模型设计、开展大规模评估以及开发早期原型，均在很大程度上依赖我们的AI基础设施。我们将继续与云基础设施及计算服务供应商合作，扩大并升级我们的AI基础设施，以支持日益

*　主要包括向第三方云服务提供商采购的计算服务，即我们从外部云平台租用而非自行建设和持有全部服务器的算力、存储及网络容量。在实践中，这主要包括我们用于训练、测试和运行大型语言模型，以及支撑我们开放平台和AI原生产品用户流量的高性能服务器、数据存储及高速网络。

— 466 —

FUTURE PLANS AND USE OF PROCEEDS

large and complex model architectures, enable faster training cycles, and improve model efficiency at scale. In addition, we will focus on integrating these models to enhance their overall intelligence across different types of information in a scalable manner.

Since we view the high costs of AI model training and usage as a key barrier to the widespread adoption of AI technologies, we plan to dedicate more internal resources to our in-house infrastructure team to upgrade our AI systems and make them more cost-efficient. We expect these upgrades and optimizations to continue lowering our costs for model training and inference, thereby improving our overall margin profile. Our ability to further reduce infrastructure costs while supporting increasingly complex models will remain pivotal for our long-term success. We will further focus on improving cost efficiency in model training and inference by refining our infrastructure and workflow, which we believe is critical to improving margins as we scale our operations.

Through these investments, we aim to enable the continuous advancement of our foundation models and maintain their competitiveness in the global market. We believe the strength of our foundation models is the most critical factor for our business success and financial performance, and we will continue advancing foundation model architectural breakthroughs such as dynamic multi-modal model integration. We believe sustained investment based on technological insights will allow us to push multi-modality integration, real-world applicability, and societal intelligence. The details of our foundation model development plans are set out below:

| Modality | For the year ending December 31 | | | | | |---|---|---|---|---|---| | | 2026 | 2027 | 2028 | 2029 | 2030 | | | (HK$ in million) | | | | | | For multi-modal integration | | | | | | | High performance server | 152.7 | 152.7 | 152.7 | 152.7 | 152.7 | | Data storage and network | 95.5 | 95.5 | 95.5 | 95.5 | 95.5 | | Others (data procurement, data labeling, and compliance) | 38.2 | 38.2 | 38.2 | 38.2 | 38.2 | | For text models | | | | | | | High performance server | 19.1 | 19.1 | 19.1 | 19.1 | 19.1 | | Data storage and network | 95.5 | 95.5 | 95.5 | 95.5 | 95.5 | | Others (data procurement, data labeling, and compliance) | 57.3 | 57.3 | 57.3 | 57.3 | 57.3 | | | 19.1 | 19.1 | 19.1 | 19.1 | 19.1 |

– 467 –

FUTURE PLANS AND USE OF PROCEEDS Modality

For the year ending December 31 2026

2027

2028

2029

2030

(HK$ in million)

For visual models High performance server Data storage and network Others (data procurement, data labeling, and compliance) For audio/music models High performance server Data storage and network Others (data procurement, data labeling, and compliance) Miscellaneous

76.4 38.2 19.1

76.4 38.2 19.1

76.4 38.2 19.1

76.4 38.2 19.1

76.4 38.2 19.1

19.1 38.2 11.5 19.1

19.1 38.2 11.5 19.1

19.1 38.2 11.5 19.1

19.1 38.2 11.5 19.1

19.1 38.2 11.5 19.1

7.6 19.1

7.6 19.1

7.6 19.1

7.6 19.1

7.6 19.1

Total

381.8

381.8

381.8

381.8

381.8

(ii) 培养模型开发的研发人才。我们计划在未来五年内拨出约20.0%（即约7.637亿港元）的募集资金净额，用于扩大和加强专注于提升我们基础模型智能性、效率和竞争力的研发团队。该款项将全部用于基础模型开发团队的员工成本。为维持并提升这一竞争优势，我们将持续在全球范围内招募顶尖AI研究人员、工程师和科学家，同时投资于内部人才的成长与发展。这些团队成员将专注于算法创建、探索先进模型架构、提升训练与推理效率，以及以可扩展的方式整合不同类型信息的多模态能力。

Cultivating R&D Talent for Model Development. We plan to allocate approximately 20.0%, or HK$763.7 million of the net proceeds to expand and strengthen our R&D teams focused on advancing the intelligence, efficiency and competitiveness of our foundation models over the next five years. This amount will be used entirely for staff costs for our foundation model development team. To sustain and advance this competitiveness, we will continue to recruit top-tier AI researchers, engineers and scientists globally, while also investing in the growth and development of our in-house talent. These team members will focus on creating algorithms, exploring advanced model architectures, improving training and inference efficiency, and integrating multi-modal capabilities across different types of information in a scalable manner.

– 468 –

FUTURE PLANS AND USE OF PROCEEDS

We will uphold rigorous hiring standards as we expand our R&D team. For experienced hires, we will seek candidates from top AI companies or academic institutions who have published in top peer-reviewed journals or have direct experience in developing foundation models. For campus recruits, we will focus our talent acquisition on top universities globally. The details of our recruitment plan are set forth as below:

| Position | For the year ending December 31 | | | | | |---|---|---|---|---|---| | | 2026 | 2027 | 2028 | 2029 | 2030 | | | (HK$ in million) | | | | | | Foundational model research | 50.0 | 50.0 | 50.0 | 50.0 | 50.0 | | Pre-training algorithms | 20.0 | 20.0 | 20.0 | 20.0 | 20.0 | | Post-training algorithms | 20.0 | 20.0 | 20.0 | 20.0 | 20.0 | | Training data | 10.0 | 10.0 | 10.0 | 10.0 | 10.0 | | AI infrastructure | 8.5 | 8.5 | 8.5 | 8.5 | 8.5 | | Training framework | 3.4 | 3.4 | 3.4 | 3.4 | 3.4 | | Inference optimization | 3.4 | 3.4 | 3.4 | 3.4 | 3.4 | | Load balancing | 1.7 | 1.7 | 1.7 | 1.7 | 1.7 | | Increased compensation for existing personnel | 94.2 | 94.2 | 94.2 | 94.2 | 94.2 | | **Total** | **152.7** | **152.7** | **152.7** | **152.7** | **152.7** |

In addition to our recruitment efforts, we plan to enhance compensation levels for our current and new R&D team members to stay ahead in the fierce AI talent race in the industry. We will offer competitive packages, including both cash compensation and share-based incentives, to attract and retain top-tier talent and to ensure their long-term interests align with the success of our company. Through these investments, we aim to cultivate a world-class R&D organization capable of driving continuous breakthroughs in model intelligence and sustaining our leadership position in the global AI industry. We plan to increase the compensation of our existing R&D personnel over the next five years, allocating approximately 12.3% of the net proceeds or HK$471.2 million in this regard.

– 469 –

FUTURE PLANS AND USE OF PROCEEDS

•

Development of Our AI-Native Products. To reinforce our product leadership and commercialization capabilities, we plan to allocate approximately 20.0%, or HK$763.7 million, of the net proceeds over the next five years to the development, refinement and global scaling of our AI-native products, including investments in product development and relevant talent. Delivering differentiated, highly interactive and scalable AI-native products is critical to driving user adoption, engagement and monetization across consumer, developer, and enterprise markets. The performance and capabilities of our products directly influences our competitiveness in the global market and our ability to generate sustainable revenue growth. Furthermore, positive product performance and user experience are key drivers of brand recognition, global market expansion and long-term commercial success.

(i)

Enhancing Resources for Product Development. We plan to allocate approximately 15.0%, or HK$572.7 million, of the net proceeds to refine our existing AI-native products and launch new ones powered by our advanced foundation models. This amount will be used entirely for R&D expenses. We aim to enhance user experience through stronger engagement, broader modality support, and improved model memory. We also intend to expand use cases and commercial applications across both consumer and business markets. We aim to accelerate the enhancement and expansion of our current AI-native products. For MiniMax, we will focus on enhancing our MiniMax Agent's ability to perform more complex tasks automatically. Hailuo AI will be upgraded to support faster, higher-quality image and longer video generation with more creative controls and agentic capabilities. For MiniMax Audio, we aim to introduce more interactive and immersive features. We will also scale our Talkie/Xingye platform by enriching the emotional intelligence of the themes and characters within such platforms and improving multi-modal interactions to boost user engagement globally. Meanwhile, our Open Platform will expand its industry-specific AI solutions and API capabilities. This will enable integration and adoption of our Open Platform by enterprise customers and developers worldwide, especially in emerging markets.

– 470 –

FUTURE PLANS AND USE OF PROCEEDS

In addition, we will keep developing new AI-native products that leverage continuous improvements in our foundation models. This includes designing new AI-native applications that offer enhanced user experiences, expanded multi-modal capabilities, and smooth integration across different platforms. We aim to discover new use cases and launch our products in new markets and verticals, thereby driving adoption and creating commercial value for both individual users, developers and enterprise customers. The details of our AI-native product development plans are set out below:

| Product | For the year ending December 31 | | | | | |---|---|---|---|---|---| | | 2026 | 2027 | 2028 | 2029 | 2030 | | | (HK$ in million) | | | | | | MiniMax | 28.6 | 28.6 | 28.6 | 28.6 | 28.6 | | Hailuo AI | 28.6 | 28.6 | 28.6 | 28.6 | 28.6 | | MiniMax Audio | 11.5 | 11.5 | 11.5 | 11.5 | 11.5 | | Talkie/Xingye | 22.9 | 22.9 | 22.9 | 22.9 | 22.9 | | Open Platform | 11.5 | 11.5 | 11.5 | 11.5 | 11.5 | | New products | 11.5 | 11.5 | 11.5 | 11.5 | 11.5 | | **Total** | **114.5** | **114.5** | **114.5** | **114.5** | **114.5** |

(ii) Cultivating Talent for Product Development. We plan to allocate approximately 5.0%, or HK$190.9 million of the net proceeds to expand and strengthen our product development and commercialization talent over the next five years. This amount will be used entirely for staff costs for our product development team. We will continue to recruit top-tier talent globally, while also investing in the growth and development of our in-house talent. These team members will focus on translating advancements in our foundation models into differentiated product capabilities, enhancing usability, interactivity and multi-modal functionality, and developing new AI-native products and commercial applications. In addition, they will collaborate closely with our marketing, sales and customer support teams to optimize user experience for global audiences, incorporate market insights into product innovation, and drive adoption across both consumer and enterprise segments.

We will also continue to expand our international sales and marketing team to support the global scaling of our AI-native products. We will continue our global expansion by building a sales and marketing team tailored for international markets. This team will be composed of professionals with global perspectives and experience in overseas operations. In parallel, we will further strengthen our customer service and support capabilities to enhance overall user satisfaction and long-term retention.

– 471 –

FUTURE PLANS AND USE OF PROCEEDS

We will uphold high hiring standards as we expand our global talent base for product development and commercialization. For experienced hires, we intend to recruit candidates with proven expertise in developing and scaling products, or in driving international commercialization and user growth. For campus hires, we will prioritize candidates from top universities worldwide with technical skills, creative problem-solving abilities and a demonstrated passion for product innovation. The details of our recruitment plan are set forth as below:

| Position | For the year ending December 31 | | | | | |---|---|---|---|---|---| | | 2026 | 2027 | 2028 | 2029 | 2030 | | | (HK$ in million) | | | | | | Product development and commercialization | | | | | | | Front-end development | 6.8 | 6.8 | 6.8 | 6.8 | 6.8 | | Back-end development | 6.8 | 6.8 | 6.8 | 6.8 | 6.8 | | Product manager | 3.4 | 3.4 | 3.4 | 3.4 | 3.4 | | International sales and marketing | | | | | | | Marketing and branding | 3.6 | 3.6 | 4.5 | 4.5 | 4.5 | | Key account | 0.9 | 0.9 | 1.8 | 1.8 | 1.8 | | Increased compensation for existing personnel | 2.7 | 2.7 | 2.7 | 2.7 | 2.7 | | Subtotal | 17.0 | 17.0 | 17.0 | 17.0 | 17.0 | | **Total** | **37.6** | **37.6** | **38.5** | **38.5** | **38.5** |

In addition to expanding our talent pipeline, we plan to enhance compensation for our current and new team members to remain competitive in attracting and retaining top talent. We will offer compensation packages, including both competitive salaries and share-based incentives, to align team members' long-term interests with the success of our company. We plan to increase the compensation of our existing product development and commercialization and international sales and marketing personnel over the next five years, allocating approximately 2.2% of the net proceeds or HK$85.2 million in this regard.

• Working Capital and General Corporate Purposes. The remaining approximately 10.0%, or HK$381.8 million, of the net proceeds will be allocated to working capital and general corporate purposes.

– 472 –

FUTURE PLANS AND USE OF PROCEEDS

If the Offer Price is fixed at the high-end or low-end of the Offer Price range (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised), the net proceeds will increase or decrease by approximately HK$171.5 million (after deducting underwriting fees and expenses related to the Global Offering). We intend to apply the additional or reduced net proceeds to the above uses on a pro rata basis.

If the Offer Size Adjustment and the Over-allotment Option are exercised in full, our Company will receive additional net proceeds of approximately HK$1,248.3 million for 8,188,020 Shares to be allotted and issued upon the full exercise of the Offer Size Adjustment and the Over-allotment Option based on the Offer Price of HK$158.00 per Offer Share, being the mid-point of the Offer Price range, and after deducting the underwriting fees and commissions payable by our Company. The additional amount raised will be applied to the above areas of use of proceeds on pro rata basis.

To the extent that the net proceeds from the Global Offering are not immediately required for the above purposes and to the extent permitted by the relevant law and regulations, we will only place the net proceeds from the Global Offering in short-term interest-bearing accounts at licensed commercial banks and/or other authorized financial institutions as defined under the Securities and Futures Ordinance or applicable laws in the relevant jurisdictions. We will make an appropriate announcement if there is any change to the above proposed use of proceeds.

– 473 –

UNDERWRITING

HONG KONG UNDERWRITERS

China International Capital Corporation Hong Kong Securities Limited UBS AG Hong Kong Branch Goldman Sachs (Asia) L.L.C. Morgan Stanley Asia Limited Futu Securities International (Hong Kong) Limited Tiger Brokers (HK) Global Limited

UNDERWRITING

This prospectus is published solely in connection with the Hong Kong Public Offering. The Hong Kong Public Offering is fully underwritten by the Hong Kong Underwriters on a conditional basis. The Company expects the International Offering to be fully underwritten by the International Underwriters. If, for any reason, the Offer Price is not agreed between the Overall Coordinators (for themselves and on behalf of the Underwriters) and the Company, the Global Offering will not proceed and will lapse.

The Global Offering comprises the Hong Kong Public Offering of initially 1,269,480 Hong Kong Offer Shares (subject to reallocation on the basis as set out in "Structure of the Global Offering" in this prospectus) and the International Offering of initially 24,119,740 International Offer Shares (subject to reallocation on the basis as described in "Structure of the Global Offering" in this prospectus as well as to the Offer Size Adjustment Option and the Over-allotment Option).

As the Company is likely to be deemed as a "covered foreign person" as described in the Final Rule, certain Underwriters have informed the Company that they may consider making notifications with the U.S. Department of the Treasury. None of the Underwriters has any obligation to inform the Company or any investor if they later decide that they will not file such notifications.

UNDERWRITING ARRANGEMENTS AND EXPENSES

Hong Kong Public Offering

Hong Kong Underwriting Agreement

The Hong Kong Underwriting Agreement was entered into on December 30, 2025. Pursuant to the Hong Kong Underwriting Agreement, the Company is offering the Hong Kong Offer Shares for subscription on the terms and conditions set out in this prospectus, and the Hong Kong Underwriting Agreement at the Offer Price.

– 474 –

UNDERWRITING

Subject to (a) the Stock Exchange granting approval for the listing of, and permission to deal in, the Class A Ordinary Shares in issue and to be issued pursuant to the Global Offering (including the Shares which may be issued pursuant to the exercise of the Offer Size Adjustment Option and the Over-allotment Option) on the Main Board of the Stock Exchange and such approval not having been withdrawn and (b) certain other conditions set out in the Hong Kong Underwriting Agreement, the Hong Kong Underwriters have agreed severally but not jointly to procure subscribers for, or themselves to subscribe for, their respective applicable proportions of the Hong Kong Offer Shares being offered which are not taken up under the Hong Kong Public Offering on the terms and conditions set out in this prospectus, and the Hong Kong Underwriting Agreement.

The Hong Kong Underwriting Agreement is conditional on, among other things, the International Underwriting Agreement having been executed and becoming unconditional and not having been terminated in accordance with its terms.

Grounds for Termination

If at any time prior to 8:00 a.m. on the day that trading in the Class A Ordinary Shares commences on the Stock Exchange:

(a) there develops, occurs, exists or comes into force:

(i) any new law or regulation or any change or development involving a prospective change or any event or series of events or circumstances likely to result in a change or a development involving a prospective change in existing laws or regulations, or the interpretation or application thereof by any court or any competent Authority in or affecting Hong Kong, the Cayman Islands, the PRC, the United States, the United Kingdom, Singapore and Philippines, or other jurisdictions relevant to the Group or the Global Offering (each a "Relevant Jurisdiction" and collectively, the "Relevant Jurisdictions"); or

(ii) 任何变化或涉及预期变化的发展，或任何可能导致任何本地、国家、地区或国际金融、政治、军事、工业、经济、财政、法律、监管、货币、信贷或市场条件或情绪、税收、股票证券或货币汇率或管制或任何货币或贸易结算系统，或外商投资法规发生变化或预期变化的事件或一系列事件或情形（包括但不限于港元、美元或人民币兑任何外币贬值，港元与美元挂钩制度或人民币与任何外币或数种外币挂钩制度发生变化），或其他金融市场（包括但不限于股票市场和债券市场、货币及外汇市场、银行同业市场及信贷市场的状况及情绪）在任何相关司法管辖区内或对任何相关司法管辖区产生影响，或影响对发售股份的投资；或

– 475 –

承销

(iii) 任何性质属不可抗力的事件或一系列事件或情形（包括但不限于政府行为、宣布区域性、全国性或国际性紧急状态或战争、灾难、危机、经济制裁、罢工、劳动争议、其他工业行动、停工、火灾、爆炸、洪水、海啸、地震、火山爆发、社会动乱、暴乱、叛乱、公共秩序混乱、政府运作瘫痪、战争行为、流行病、大流行病、疾病爆发或升级、变异或加重、意外或交通中断或延误、本地、全国、地区或国际敌对行动的爆发或升级（无论战争是否已经或曾经宣布）、天灾或恐怖主义行为（无论是否已有人声称负责））在任何相关司法管辖区内或对任何相关司法管辖区产生影响；或

(iv) 对以下交易施加或宣布任何暂停、中止或限制（包括但不限于施加或规定任何最低或最高价格限制或价格范围）：(i) 联交所、上海证券交易所、深圳证券交易所、东京证券交易所、新加坡证券交易所、纽约证券交易所、纳斯达克全球市场或伦敦证券交易所上一般股份或证券的交易；或 (ii) 本公司在任何证券交易所或场外交易市场上市或报价的任何证券的交易；或

(v) 对任何相关司法管辖区内的银行业务施加或宣布任何普遍暂停，或任何相关司法管辖区内或对任何相关司法管辖区产生影响的商业银行、外汇交易或证券结算或清算服务、程序或事项出现中断；或

(vi) 除经整体协调人事先书面同意外，本公司依据《公司（清盘及杂项条文）条例》或《上市规则》，或应联交所及╱或证监会的规定或要求，就发售股份的要约和出售而发出或被要求发出本招股章程或其他相关文件的补充或修订；或

(vii) 任何主管机关或其他监管机构或政治团体或组织对集团任何成员、集团任何成员的董事或高级管理人员展开任何公开行动或调查，或宣布有意采取任何该等行动；或

– 476 –

承销

(viii) 以任何形式直接或间接对集团任何成员或任何控股股东施加制裁或出口管制，或由任何相关司法管辖区施加或针对任何相关司法管辖区施加，或以任何形式直接或间接由任何相关司法管辖区撤销或针对任何相关司法管辖区撤销于香港包销协议签署之日存在的交易特权；或

(ix) 任何债权人就集团任何成员的债务提出有效的付款或还款要求，或就集团任何成员在到期日前所承担的责任提出有效的付款或还款要求；或

(x) 本招股章程（或用于拟议发售、配发、发行、认购或出售任何发售股份的任何其他文件）、中国证监会备案文件或全球发售的任何方面不符合《上市规则》或任何其他适用法律；或

(xi) 对集团任何成员、任何控股股东或本招股章程所列明的任何董事或高级管理人员提出或宣布任何诉讼、争议、法律行动或索偿，或对其展开任何监管或行政调查或行动；或

(xii) 集团任何成员或任何董事违反《上市规则》或适用法律；或

(xiii) 本招股章程"风险因素"一节所载任何风险出现变化或预期变化，或其中任何风险成真，

而在任何该等情况下，无论单独或合并计算，据联席保荐人及╱或整体协调人（代表其本身及代表香港包销商）的绝对酌情决定认为：

i. 已经、将会或可能对集团整体的盈利、亏损、经营业绩、资产、负债、整体事务、业务、管理、表现、前景、股东权益、地位或状况（财务、贸易或其他方面）产生重大不利影响或任何涉及预期重大不利影响的发展（"重大不利影响"）；

ii. 已经、将会或可能对全球发售的成功、香港公开发售的申请水平或国际发售的认购意向水平产生重大不利影响；或

iii.

UNDERWRITING

provisional liquidator, receiver or manager is appointed to any member of the Group or anything analogous thereto occurs in respect of any member of the Group.

**承销**

使香港包销协议、香港公开发售或全球发售的任何重要部分无法、不适宜、不便或不能按预期履行或实施，或使香港公开发售及/或全球发售无法进行，或无法向市场推介全球发售，或无法按照发售文件所述条款及方式交付或分发发售股份；或

iv.

已经或将会或可能导致香港包销协议的任何部分（包括包销）无法按照其条款履行，或妨碍处理根据全球发售或其包销而提出的申请及/或付款；或

(b) 联席保荐人及/或整体协调人（代表其本身及代表香港包销商）获悉：

(i) 任何发售文件（定义见香港包销协议）、操作文件（定义见香港包销协议）、中国证监会申报文件及/或任何由公司或代表公司就全球发售而发出或使用的通知、公告、广告、通讯或其他文件（包括任何补充或修订）（"全球发售文件"）所载的任何陈述，在发出时或其后，在任何重要方面属虚假、不正确、不准确或不完整，或具有误导性；或该等文件所载的任何估计、预测、意见、意向或预期，在发出时或其后，在任何方面属不公平或具有误导性，或基于虚假、不诚实或不合理的假设，或出于不良意图；或

(ii) 出现或发现任何事项，而该等事项倘若在本招股说明书日期之前出现或被发现，将构成任何全球发售文件的重大遗漏或失实陈述；或

(iii) 公司或控股股东在香港包销协议或国际包销协议中作出的任何陈述、保证及承诺遭到违反，或发生任何令该等陈述、保证及承诺在任何方面失实、不正确、不完整或具有误导性的事件或情况；或

(iv) 发生任何行为、作为或不作为，导致或可能导致任何赔偿方（定义见香港包销协议）须根据香港包销协议或国际包销协议（包括任何补充或修订）的相关赔偿条款承担任何法律责任；或

– 478 –

**承销**

(v) 公司、任何控股股东成员或任何基石投资者（视情况而定）违反香港包销协议、国际包销协议或基石投资协议所规定的任何义务或承诺；或

(vi) 发生或出现任何构成或具有重大不利影响的变化或发展（包括预期变化）；或

(vii) 公司董事会主席、任何董事或本招股说明书所载任何关键人员或高级管理人员寻求退休，或被免职或离任；或

(viii) 公司任何董事或本招股说明书所载任何关键人员或高级管理人员被控以可公诉罪行，或依法被禁止或被取消资格参与公司管理或出任公司董事；或

(ix) 公司撤回本招股说明书（及/或就全球发售认购或出售任何发售股份而使用的任何其他文件）或全球发售；或

(x) 上市委员会批准根据全球发售（包括根据行使超额配股权及发售规模调整权）发行及待发行的A类普通股上市及获准买卖的申请遭拒绝或未获批准（惯常条件除外），且该等拒绝或未获批准发生于上市日期当日或之前；或已获批准的，其后该批准被撤销、取消、附加条件（惯常条件除外）、撤回或暂缓；或

(xi) 任何人士（联席保荐人除外）撤回其对发行本招股说明书或任何发售文件的同意，而该等文件载有其报告、函件及/或法律意见（视情况而定），以及以相应形式和背景载有其名称的提述；或

(xii) 公司因任何原因被禁止根据全球发售条款提呈、配发、发行或出售任何发售股份；或

(xiii) 就集团任何成员提出清盘或清算命令或呈请，或集团任何成员与其债权人达成任何和解或安排，或订立债务重组安排，或通过任何关于集团任何成员清盘的决议，或向集团任何成员委任临时清盘人、接管人或管理人，或集团任何成员发生任何类似情况。

– 479 –

承销

临时清算人、接管人或管理人获委任管理集团任何成员的全部或部分资产或业务，或就集团任何成员发生任何类似情况；或

(xiv)（A）中国证监会就中国证监会备案所发出的接受通知，及/或于中国证监会网站公布的中国证监会备案结果遭拒绝、撤回、撤销或宣告无效；或（B）除经总协调人事先书面同意外，本公司根据中国证监会规则或应中国证监会的任何要求或请求，就中国证监会备案发出或须发出补充文件或修订文件；或（C）中国证监会备案不符合中国证监会规则或任何其他适用法律；或

(xv)（i）于簿记建档过程中所下达或确认的订单中，有重大部分；或（ii）任何基础投资者根据与其签订的基础投资协议所作出的任何投资承诺，已遭撤回、终止或取消，或相关订单及╱或投资承诺款项未能于规定时间及方式内收妥或结算，或存在其他情况，

则在上述每种情况下，联席保荐人及总协调人（为其本身及代表香港包销商）可在其绝对酌情决定下，向本公司发出书面通知，立即终止香港包销协议。

根据《上市规则》向联交所作出的承诺

（A）本公司的承诺

根据《上市规则》第10.08条，本公司已向联交所承诺，自上市日期起六（6）个月内（无论该等股份或证券的发行是否将于上市日期起六（6）个月内完成），本公司不会再发行任何股份或可转换为本公司股本证券的证券（不论是否属于已上市类别），亦不会就任何该等发行订立任何协议，惟根据全球发售（包括任何可能根据发售规模调整权及超额配股权发行的股份）或《上市规则》第10.08条所规定的任何其他情况而发行或拟发行的股份除外。

（B）控股股东的承诺

根据《上市规则》第10.07条及第18C.13条，各控股股东已向联交所及本公司承诺，除根据全球发售（包括发售规模调整权及超额配股权）外，在本招股说明书就其于本公司的持股权作出披露之日起至上市日期起计24个月届满之日止的期间内，其将不会，并将促使相关登记持有人不会在未获联交所事先书面同意，或除非符合《上市规则》适用规定的情况下，直接或间接出售或转让本招股说明书所示其实益拥有的本公司任何证券，亦不会就此订立任何出售或转让协议，或就上述证券创设任何期权、权利、权益或抵押。

– 480 –

承销

根据《上市规则》第18C.22条，希望在上市后不再被视为商业化前公司的商业化前公司须向联交所提出申请。根据《上市规则》第18C.23条，商业化前公司须向联交所提供已发布的经审计财务报告以支持根据第18C.22条提出的申请，证明：（1）就其最近一个经审计财务年度而言，已符合《上市规则》第18C.03(4)条所载的收入规定；或（2）就其整体业务营运而言，已符合《上市规则》第8.05条所载测试中的至少一项。若联交所通知本公司，本公司于上市后将不再被视为商业化前公司，则上述禁售期届满日期将为以下较晚者：（i）如本公司以商业公司身份申请上市，上述禁售期本应届满之日；及（2）根据《上市规则》第18C.24条就撤销商业化前公司指定作出公告后第30天。

根据《上市规则》第10.07(2)条附注3，各控股股东已向联交所及本公司承诺，在本招股说明书就其于本公司的持股权作出披露之日起至上市日期起计24个月届满之日止的期间内，其将，并将促使相关登记持有人：

(a) 在其根据《上市规则》第10.07条附注2，以其实益拥有的本公司任何证券向认可机构（定义见香港法例第155章《银行业条例》）作出质押或押记时，立即通知本公司有关质押或押记事宜及所质押或押记的证券数量；及

(b) 在收到任何本公司证券的质权人或押记人就任何已质押或押记证券将予出售的口头或书面迹象时，立即通知本公司该等迹象。

本公司在获控股股东通知上述第(i)及(ii)段所述事项后，将尽快通知联交所，并在符合届时适用的《上市规则》规定的情况下，以公告方式披露该等事项。

– 481 –

UNDERWRITING

(C) 主要人士及探路者战略性投资者的承诺

根据《上市规则》第18C.14条，公司主要人士颜博士及云女士、探路者战略性投资者（包括IDG战略性投资者及米哈游战略性投资者），以及彼等各自的紧密联系人（如本招股书"历史、重组及公司架构——禁售期"一节所述）（统称"承诺方"），已分别向联交所及本公司作出承诺，除根据全球发售（包括发售规模调整权及超额配股权）外，除非符合《上市规则》第18C.15条的规定，否则在本招股书披露其持股情况之日起至上市日期后24个月（就探路者战略性投资者而言为12个月）届满之日止的期间内，不得出售，亦不得订立任何协议出售或以其他方式就本招股书所示其实益拥有的任何公司股份创设任何期权、权利、权益或产权负担。

《上市规则》第18C.14条附注2规定，上述承诺不妨碍上述人士将其实益拥有的公司股份用作抵押（包括押记或质押），以向获授权机构（定义见《银行业条例》（香港法例第155章））申请真实商业贷款的担保。

此外，根据《上市规则》第18C.14条附注2，上述每名人士已向联交所及本公司承诺，在本招股书披露其持股情况之日起至上市日期后24个月（就探路者战略性投资者而言为12个月）届满之日止的期间内：

(a) 当其将实益拥有的任何股份质押或押记予获授权机构（定义见《银行业条例》（香港法例第155章））以申请真实商业贷款时，立即通知本公司及联交所有关质押或押记的详情及所质押或押记的股份数目；及

(b) 当其收到受押人或受押记人以口头或书面形式表示将出售任何已质押或已押记股份时，立即通知本公司及联交所有关迹象。

本公司在获悉上述人士就上述事宜（如有）作出通知后，将立即知会联交所，并在获悉后尽快披露相关事宜。

根据《上市规则》第18C.23条附注2，尽管有上述规定，倘本公司于上市后不再被视为商业化前公司，则上述相关股东须受《上市规则》第18C.14条规定约束的禁售期，将于以下较晚日期届满：(1) 倘本公司以商业公司身份申请上市，相关禁售期本应届满的日期；及(2) 根据《上市规则》第18C.24条规定就撤销商业化前公司称号作出公告后第30天。

根据香港包销协议作出的承诺

(A) 公司的承诺

根据香港包销协议，除(a) 公司根据全球发售（包括根据发售规模调整权及超额配股权）发行、发售或出售发售股份，及(b) 根据上市后股份激励计划授予奖励，以及为满足上市后股份激励计划所授予奖励而发行及交付A类普通股外，在香港包销协议签订之日起至上市日期后六个月届满之日（含当日）止的期间内（"首个六个月期间"），本公司已向各联席保荐人、保荐人整体协调人、整体协调人、联席全球协调人、联席账簿管理人、联席牵头经办人、资本市场中介机构及香港包销商承诺，未经联席保荐人及整体协调人（代表其本身及代表香港包销商）事先书面同意，且除非符合《上市规则》的规定，否则不会：

(i) 直接或间接、有条件或无条件地，分配、发行、出售、接受认购、要约分配、发行或出售、订立合约或协议分配、发行或出售、转让、抵押、押记、质押、抵押、借出、授予或出售任何期权、权证、合约或认购或购买权利、授予或购买任何期权、权证、合约或分配、发行或出售权利，或以其他方式转让或处置或创设产权负担，或协议转让或处置或创设产权负担，或回购公司股本、任何A类普通股或公司任何其他证券的任何法律或实益权益或上述任何权益（包括但不限于任何可转换为或可兑换或可行使换取或代表接收权利的证券，或购买公司任何股本、A类普通股或其他证券（视情况而定）的任何权证或其他权利），或将公司任何股本、A类普通股或其他证券（视情况而定）存入存管机构以发行存托凭证；或

– 482 –

UNDERWRITING

（二）承销

(ii) 订立任何互换或其他安排，将A类普通股或本公司任何其他证券的所有权（法律上或受益上）的经济利益（全部或部分）转让予他人，或将前述任何证券的任何权益（包括但不限于任何可转换为、可交换或可行使为股份或代表收取股份权利的证券，或任何购买股份的认股权证或其他权利）作出上述安排；或

(iii) 订立与上文第（i）或（ii）段所述的任何交易具有相同经济效果的任何交易；或

– 483 –

承销

(iv) 要约或同意进行上文第（i）、（ii）或（iii）段所述的任何事项，或宣布有意作出上述任何事项，

在任何情况下，无论上文第（i）、（ii）或（iii）段所述的任何交易是否以交付股本、A类普通股或其他证券、以现金或其他方式结算（无论此类股本或其他证券的发行是否将在首个六个月期间内完成）。

如在首个六个月期间届满之日起计的六个月期间（"第二个六个月期间"）内，本公司订立上文第（i）、（ii）或（iii）段所述的任何交易，或要约、同意或宣布有意进行任何该等交易，则本公司须采取一切合理步骤，确保其不会在A类普通股或本公司其他股权证券中造成混乱或虚假市场。

（B）控股股东的承诺

根据香港包销协议，各控股股东已向本公司、联席保荐人、保荐人整体协调人、整体协调人、联席全球协调人、CMI、联席账簿管理人、联席牵头经办人及香港包销商各方承诺，未经联席保荐人及整体协调人（为其自身及代表香港包销商）事先书面同意，且除非符合上市规则的规定，在香港包销协议签署之日起至上市日期后24个月届满之日（含当日）止的期间（"24个月期间"）内的任何时间：

(i) 除MiniMax Matrix根据借股协议借出A类普通股及根据全球发售出售A类普通股外，其/彼将不会，并将促使相关登记持有人、以信托方式为其持有股份的任何代名人或受托人以及其控制的公司不会，直接或间接、有条件或无条件地，要约、质押、抵押、出售、要约出售、订约或同意出售、按揭、押记、抵押、借出、授予或出售任何期权、认股权证、合约或购买权利、授予或购买任何期权、认股权证、合约或出售权利、授予或同意授予任何购买或认购期权、权利或认股权证、借出或以其他方式转让或处置或在任何A类普通股、B类普通股或本公司其他证券或前述任何证券的任何权益上设立产权负担，或同意转让或处置或设立产权负担（包括但不限于任何可转换为A类普通股或B类普通股、可交换或可行使为A类普通股或B类普通股或代表收取A类普通股或B类普通股权利的证券，或任何购买A类普通股或B类普通股的认股权证或其他权利，或将任何A类普通股或B类普通股存托

– 484 –

承销

或公司其他股权证券存入存托人以发行存托凭证）的任何控股股东锁定证券（"控股股东锁定证券"）；或

（ii）订立任何掉期或其他安排，将任何控股股东锁定证券的全部或部分经济利益（法定或实益）转让予他人；或

（iii）订立与上文第（i）或（ii）段所述任何交易具有相同经济效果的任何交易，或要约或同意或宣布任何意图进行上文第（i）或（ii）段所述任何交易，在各种情况下，无论上述任何交易是否以交付A类普通股、B类普通股或公司其他证券或以现金或其他方式结算，以及无论该等交易是否在24个月期间内完成。

在24个月期间届满之前，如任何控股股东订立上述任何交易或要约、同意、订约或公开宣布任何意图进行任何该等交易，其将采取一切合理措施确保该等处置不会在本公司证券中造成混乱或虚假市场。

上述限制不妨碍控股股东（i）按照上市规则购买额外的A类普通股或公司其他证券，并处置该等额外的A类普通股或公司证券，但任何该等购买或处置不得违反上述与控股股东有关的锁定安排或本公司遵守最低公众持股量规定；（ii）在上市规则第18C.15条所规定的情况下处置任何控股股东锁定证券的权益；以及（iii）根据《上市规则》第10.07(2)条附注（2），将其实益拥有的A类普通股或公司其他证券或其中任何权益作为抵押（包括押记或质押）以支持向认可机构（定义见《银行业条例》（香港法例第155章））申请的真实商业贷款，但须符合以下条件：在24个月期间内，（a）相关控股股东须于其或相关注册持有人质押或押记其实益拥有的任何A类普通股或公司其他证券时，立即以书面形式通知本公司及总协调人有关质押或押记情况及所质押或押记的A类普通股或公司其他证券数目；及（b）当相关控股股东从受质人或受押人收到任何股份将被处置的口头或书面迹象时，其须立即通知本公司及总协调人。

– 485 –

UNDERWRITING

In the event that upon the notification by the Stock Exchange that the Company will no longer be regarded as a Pre-Commercial Company after the Listing, the lock-up period set out above will expire on the later of: (i) the date on which such lock-up periods would have ended if the Company had applied for listing as a Commercial Company; and (2) the date falling on the 30th day after the announcement on the removal of designation as a Pre-Commercial Company as required under Rule 18C.24 of the Listing Rules.

Undertaking by the other existing shareholders

In addition to the respective undertakings by the Controlling Shareholders and the Underwriting Providers as disclosed above in this section, each of the Pre-IPO Investors and MiniMax Gene has agreed to provide a lock-up undertaking (the "Lock-up Undertakings") in favour of the Company, the Joint Sponsors and the Overall Coordinators (for themselves and on behalf of the Underwriters).

Pursuant to the Lock-up Undertakings, (a) existing shareholders holding an aggregate of 48.92% of the total number of issued shares of the Company as of the Latest Practicable Date have agreed to a lock-up in respect of the Shares they held prior to the Listing for a period commencing from the date of their respective Lock-up Undertakings and ending on the date which is six months from the Listing Date, subject to customary exceptions, and (b) existing shareholders holding an aggregate of 20.34% of the total number of issued shares of the Company as of the Latest Practicable Date have agreed to a lock-up in respect of the Shares they held prior to the Listing for a period commencing from the date of their respective Lock-up Undertakings or the date of this Prospectus and ending on a date which is the earlier of (i) the twentieth (20th) trading day starting from the date on which the Class A Ordinary Shares are included as an eligible stocks of Stock Connect and can be traded via Stock Connect of the Stock Exchange, or (ii) nine months from the date on which trading in the Class A Ordinary Shares commences on the Stock Exchange, subject to customary exceptions.

Hong Kong Underwriters' Interests in the Company

Save for their respective obligations under the Hong Kong Underwriting Agreement and otherwise as disclosed in the prospectus, as of the Latest Practicable Date, none of the Hong Kong Underwriters was interested, legally or beneficially, directly or indirectly, in any Shares or any securities of any member of the Group or had any right or option (whether legally enforceable or not) to subscribe for or purchase, or to nominate persons to subscribe for or purchase, any Shares or any securities of any member of the Group.

Following the completion of the Global Offering, the Hong Kong Underwriters and their affiliated companies may hold a certain portion of the Shares as a result of fulfilling their respective obligations under the Hong Kong Underwriting Agreement.

– 486 –

UNDERWRITING International Offering International Underwriting Agreement In connection with the International Offering, the Company expects to enter into the International Underwriting Agreement with the International Underwriters on or around the Price Determination Date. Under the International Underwriting Agreement and subject to the Offer Size Adjustment Option and the Over-allotment Option, the International Underwriters would, subject to certain conditions set out therein, agree severally but not jointly to procure subscribers for, or themselves to subscribe for, their respective applicable proportions of the International Offer Shares initially being offered pursuant to the International Offering. It is expected that the International Underwriting Agreement may be terminated on similar grounds as the Hong Kong Underwriting Agreement. Potential investors should note that in the event that the International Underwriting Agreement is not entered into, the Global Offering will not proceed. See "Structure of the Global Offering — The International Offering." Over-allotment Option The Company is expected to grant to the International Underwriters the Over-allotment Option, exercisable by the Overall Coordinators (on behalf of the International Underwriters) at any time from the Listing Date until 30 days after the last day for lodging applications under the Hong Kong Public Offering, pursuant to which the Company may be required to issue up to an aggregate of 3,808,380 Class A Ordinary Shares (representing not more than 15% of the number of Offer Shares initially available under the Global Offering assuming the Offer Size Adjustment Option is not exercised at all) or up to an aggregate of 4,379,640 Class A Ordinary Shares (representing approximately 15% of the Offer Shares initially available under the Global Offering assuming the Offer Size Adjustment Option is exercised in full), at the Offer Price, to, among other things, cover over-allocations in the International Offering, if any. See "Structure of the Global Offering — Over-allotment Option." Offer Size Adjustment Option The Company is expected to grant to the Overall Coordinators the Offer Size Adjustment Option, exercisable by the Overall Coordinators (for themselves and on behalf of the International Underwriters) on or before the second Business Day prior to the Listing Date and will lapse immediately thereafter, whichever is earlier, to require our Company to allot and issue up to an aggregate of 3,808,380 additional Offer Shares, representing approximately 15.0% of the Offer Shares initially being offered under the Global Offering at the Offer Price to cover any excess demand in the International Offering. The Offer Size Adjustment Option provides flexibility for the Overall Coordinators to increase the number of Offer Shares available for purchase under the International Offering to cover additional market demand. Further details are set out in the section headed "Structure of the Global Offering — International Offering — Offer Size Adjustment Option" in this Prospectus.

– 487 –

UNDERWRITING Commissions and Expenses The Capital Market Intermediaries will receive an underwriting commission of 2% of the aggregate Offer Price of all the Offer Shares (including any Offer Shares to be issued pursuant to the exercise of the Offer Size Adjustment Option and the Over-allotment Option), out of which they will pay any sub-underwriting commissions and other fees (if any). The Capital Market Intermediaries may receive a discretionary incentive fee of up to 1.5% of the aggregate Offer Price of all the Offer Shares to be issued by the Company under the Global Offering (including any Offer Shares to be issued pursuant to the exercise of the Offer Size Adjustment Option and the Over-allotment Option). Assuming an indicative Offer Price of HK$158.00 per Offer Share (which is the mid-point of the Offer Price range), the exercise of the Offer Size Adjustment Option and the Over-allotment Option in full and the full payment of the discretionary fees, the fixed fees and the discretionary fees payable to the Capital Market Intermediaries represent approximately 56.84% and 43.16%, respectively, of the aggregate fees payable to the Capital Market Intermediaries in total in connection with the Global Offering. For any unsubscribed Hong Kong Offer Shares reallocated to the International Offering, the underwriting commission will not be paid to the Hong Kong Underwriters but will instead be paid, at the rate applicable to the International Offering, to the relevant International Underwriters. The aggregate underwriting commissions payable to the Capital Market Intermediaries in relation to the Global Offering (assuming an indicative Offer Price of HK$158.00 per Offer Share (which is the mid-point of the Offer Price range), the full payment of the discretionary incentive fee and the exercise of the Offer Size Adjustment Option and the Over-allotment Option in full) will be approximately HK$177.9 million representing approximately 3.4% of the estimated gross proceeds from the Global Offering. The aggregate underwriting commissions and fees together with the Stock Exchange listing fees, the SFC transaction levy, the AFRC transaction levy and the Stock Exchange trading fee, legal and other professional fees and printing and all other expenses relating to the Global Offering are estimated to be approximately HK$238.6 million (assuming an indicative Offer Price of HK$158.00 per Offer Share (which is the mid-point of the Offer Price range), the full payment of the discretionary incentive fee and the exercise of the Offer Size Adjustment Option and the Over-allotment Option in full) and will be paid by the Company.

– 488 –

UNDERWRITING 承销

赔偿 公司及控股股东各自同意就香港承销商可能遭受或承担的若干损失向其作出赔偿，包括因履行香港承销协议项下义务而产生的损失，以及公司或控股股东违反香港承销协议而产生的损失。

承销团成员的活动 香港公开发售及国际发售的承销商（统称"承销团成员"）及其联属公司各自可能单独从事多种活动（详见下文），该等活动并非承销或价格稳定过程的组成部分。

承销团成员及其联属公司为多元化金融机构，在世界各地均建立有业务关系。该等实体为自身账户及他人账户从事范围广泛的商业及投资银行业务、经纪业务、基金管理、交易、对冲、投资及其他活动。在各项日常业务活动中，承销团成员及其各自联属公司可能购买、出售或持有各类投资，并为自身账户及客户账户积极买卖证券、衍生工具、贷款、大宗商品、货币、信用违约掉期及其他金融工具。该等投资及交易活动可能涉及或关乎公司的资产、证券及╱或工具，及╱或与公司存在关系的人士及实体，亦可能包括为对冲目的而就本集团的贷款及其他债务订立的掉期及其他金融工具。

就股份而言，承销团成员及其联属公司的活动可能包括：担任股份买卖双方的代理人；以主事人身份与该等买卖双方订立交易，包括在全球发售中向股份初始认购人提供贷款（该等融资可能以股份作为抵押）；自营买卖股份；以及订立场外或上市衍生工具交易或上市或非上市证券交易（包括发行在证券交易所上市的证券，如衍生认股权证），以股份等资产作为相关资产。该等交易可与选定交易对手以双边协议或交易方式进行。上述活动可能需要该等实体进行对冲活动，直接或间接涉及买卖股份，可能对股份的交易价格产生负面影响。所有上述活动均可在香港及世界其他地方进行，且可能导致承销团成员及其联属公司持有股份的好仓及╱或淡仓、持有包含股份的一篮子证券或指数、持有可能购买股份的基金单位，或持有与上述任何资产相关的衍生工具。

– 489 –

UNDERWRITING 承销

就承销团成员或其联属公司在联交所或任何其他证券交易所发行以股份为相关证券的任何上市证券而言，该证券交易所的规则可能要求该等证券的发行人（或其联属公司或代理人之一）担任该证券的庄家或流动性提供者，且在大多数情况下，此亦将导致对股份进行对冲活动。

上述活动可能影响股份的市场价格或价值、股份的流动性或交易量以及股份价格的波动性，且此类情况每日发生的程度无法估计。

应注意，承销团成员在从事上述任何活动时，须遵守若干限制，包括以下各项：

• 承销团成员（稳定价格负责人或代其行事的任何人士除外）不得就要约股份的分销，在公开市场或以其他方式进行任何交易（包括发行或订立与要约股份相关的任何期权或其他衍生工具交易），以将任何要约股份的市场价格稳定或维持在公开市场上原本可能存在的水平以外的水平；及

• 承销团成员必须遵守所有适用法律法规，包括《证券及期货条例》中有关市场失当行为的条文，包括禁止内幕交易、虚假交易、操控价格及操控股票市场的条文。

承销团成员或其各自联属公司过去曾不时向公司及其各联属公司提供投资银行及其他服务，预期未来亦将继续提供，该等承销团成员或其各自联属公司已就此收取或将收取惯常费用及佣金。

此外，承销团成员或其各自联属公司可能向投资者提供融资，以资助其认购全球发售中的要约股份。

– 490 –

全球发售的结构 全球发售

本招股章程的刊发与全球发售的组成部分香港公开发售有关。

股份于联交所主板上市由联席保荐人保荐。联席保荐人已代表公司向联交所申请，要求批准本招股章程所述将予发行的A类普通股上市及买卖。

全球发售初步提呈发售合共25,389,220股要约股份，包括：

(a) 香港公开发售：初步提呈1,269,480股要约股份（可予重新分配）在香港公开发售，详情见下文"— 香港公开发售"；及

(b)

STRUCTURE OF THE GLOBAL OFFERING

THE HONG KONG PUBLIC OFFERING

Number of Offer Shares initially offered

The Company is initially offering 1,269,480 Offer Shares for subscription by the public in Hong Kong at the Offer Price, representing approximately 5% of the total number of Offer Shares initially available under the Global Offering. The number of Offer Shares initially offered under the Hong Kong Public Offering, subject to any reallocation of Offer Shares between the International Offering and the Hong Kong Public Offering, will represent approximately 0.4% of the enlarged issued share capital of the Company immediately following the completion of the Global Offering (assuming the Offer Size Adjustment Option and the Over-allotment Option are not exercised).

The Hong Kong Public Offering is open to members of the public in Hong Kong as well as to institutional and professional investors. Professional investors generally include brokers, dealers, companies (including fund managers) whose ordinary business involves dealing in shares and other securities and corporate entities that regularly invest in shares and other securities.

Completion of the Hong Kong Public Offering is subject to the conditions set out in "— Conditions of the Global Offering" below.

Allocation

Allocation of Offer Shares to investors under the Hong Kong Public Offering will be based solely on the level of valid applications received under the Hong Kong Public Offering. The basis of allocation may vary, depending on the number of Hong Kong Offer Shares validly applied for by applicants. Such allocation could, where appropriate, consist of balloting, which could mean that some applicants may receive a higher allocation than others who have applied for the same number of Hong Kong Offer Shares, and those applicants who are not successful in the ballot may not receive any Hong Kong Offer Shares.

For allocation purposes only, the total number of Hong Kong Offer Shares available under the Hong Kong Public Offering (after taking into account any reallocation referred to below) will be divided equally into two pools: pool A and pool B with any odd lots being allocated to pool A. The Hong Kong Offer Shares in pool A will be allocated on an equitable basis to valid applicants who have applied for Hong Kong Offer Shares with an aggregate subscription price of HK$5 million (excluding the brokerage, the SFC transaction levy, the Stock Exchange trading fee and the AFRC transaction levy payable) or less. The Hong Kong Offer Shares in pool B will be allocated on an equitable basis to valid applicants who have applied for Hong Kong Offer Shares with an aggregate subscription price of more than HK$5 million (excluding the brokerage, the SFC transaction levy, the Stock Exchange trading fee and the AFRC transaction levy payable) and up to the total value in pool B.

– 492 –

STRUCTURE OF THE GLOBAL OFFERING

Investors should be aware that applications in pool A and applications in pool B may receive different allocation ratios. If any Hong Kong Offer Shares in one (but not both) of the pools are unsubscribed, such unsubscribed Hong Kong Offer Shares will be transferred to the other pool to satisfy demand in that other pool and be allocated accordingly. For the purpose of the immediately preceding paragraph only, the "price" for Hong Kong Offer Shares means the price payable on application therefor (without regard to the Offer Price as finally determined). Applicants can only receive an allocation of Hong Kong Offer Shares from either pool A or pool B and not from both pools. Multiple or suspected multiple applications under the Hong Kong Public Offering and any application for more than 634,740 Hong Kong Offer Shares (being 50% of the Hong Kong Offer Shares initially available under the Hong Kong Public Offering) is liable to be rejected.

回拨机制 Reallocation

The allocation of Offer Shares between the Hong Kong Public Offering and the International Offering is subject to reallocation under the Listing Rules. Paragraph 4.2 of Practice Note 18 of the Listing Rules (as modified by Rule 18C.09 of the Listing Rules) requires a clawback mechanism to be put in place, which would have the effect of increasing the number of Hong Kong Offer Shares to certain percentages of the total number of Offer Shares to be offered in the Global Offering if certain prescribed total demand levels in the Hong Kong Public Offering are reached. 1,269,480 Offer Shares are initially available in the Hong Kong Public Offering, representing approximately 5% of the Offer Shares initially available for subscription under the Global Offering; and in the event of full subscription or oversubscription of the International Offer Shares, the Overall Coordinators shall apply a clawback mechanism following the closing of the application lists on the following basis, subject to the allocation basis as stated in Chapter 4.14 of the Guide for New Listing Applicants:

(a) If the Hong Kong Public Offering is not fully subscribed for, the Overall Coordinators has the authority to reallocate all or any unsubscribed Hong Kong Offer Shares to the International Offering, in such proportions as the Overall Coordinators deems appropriate, and the Allocation Cap as defined in and stated under Chapter 4.14 of the Guide for New Listing Applicants will not be triggered;

(b) If the number of Offer Shares validly applied for under the Hong Kong Public Offering represents 10 times or more but less than 50 times of the number of the Offer Shares initially available for subscription under the Hong Kong Public Offering, then Offer Shares will be reallocated to the Hong Kong Public Offering from the International Offering, so that the total number of Offer Shares available under the Hong Kong Public Offering will be 2,538,940 Offer Shares, representing approximately 10% of the Offer Shares initially available under the Global Offering (before any exercise of the Offer Size Adjustment Option or the Over-allotment Option); and

– 493 –

STRUCTURE OF THE GLOBAL OFFERING

(c) If the number of Offer Shares validly applied for under the Hong Kong Public Offering represents 50 times or more of the number of the Offer Shares initially available for subscription under the Hong Kong Public Offering, then Offer Shares will be reallocated to the Hong Kong Public Offering from the International Offering, so that the total number of Offer Shares available under the Hong Kong Public Offering will be 5,077,860 Offer Shares, representing approximately 20% of the Offer Shares initially available under the Global Offering (before any exercise of the Offer Size Adjustment Option or the Over-allotment Option).

STRUCTURE OF THE GLOBAL OFFERING

在每种情况下，重新分配至香港公开发售的额外发售股份将在甲组和乙组之间进行分配，而分配至国际发售的发售股份数目将相应减少，减少方式由总协调人酌情决定。

香港公开发售及国际发售中拟发售的发售股份可由总协调人（代表其本身及代表承销商）按照联交所刊发的《新上市申请人指引》第4.14章及《上市规则》实务守则第18条第4.2段，酌情在上述两项发售之间重新分配。在不违反上述段落的前提下，总协调人可酌情将发售股份由国际发售重新分配至香港公开发售，以满足香港公开发售项下的有效申请。

根据《新上市申请人指引》第4.14章，倘(i)国际发售未获足额认购，而香港公开发售已获足额认购或超额认购（无论超额认购倍数为何）；或(ii)国际发售已获足额认购或超额认购，而香港公开发售亦已获足额认购或超额认购，但香港公开发售项下有效申请认购的发售股份数目少于香港公开发售初始可供认购股份数目的10倍，总协调人有权将原本纳入国际发售的若干国际发售股份重新分配至香港公开发售，重新分配数目由总协调人酌情决定，惟重新分配后香港公开发售项下可供认购的发售股份总数不得超过2,538,960股发售股份（相当于香港公开发售项下初始可供认购的发售股份总数的两倍（于行使发售规模调整权或超额配股权前）），且最终发售价将定于本招股说明书所述发售价指示性区间的低端（即每股发售股份151.00港元）。

香港公开发售与国际发售之间任何重新分配发售股份的详情，将于全球发售结果公告中披露，该公告预计于2026年1月8日（星期四）刊发。

– 494 –

全球发售架构

若国际发售股份未获足额认购，且香港发售股份亦未获足额认购，则除非承销商愿意认购或寻找认购人认购其各自应占比例的全球发售项下未获认购的发售股份（认购条款及条件以本招股说明书及承销协议为准），否则全球发售将不会进行。

申请

香港公开发售的每位申请人须在其提交的申请中作出承诺及确认，申请人本人及其为之提出申请的任何人士均未曾申请或认购，或表示有意认购，且不会申请或认购，或表示有意认购国际发售项下的任何国际发售股份。若上述承诺及╱或确认遭到违反及╱或失实（视情况而定），该申请人于国际发售项下的申请可能会被拒绝。

香港公开发售的申请人在提出申请时，可能须缴付最高发售价（视乎申请渠道而定），以及就每股发售股份应付的经纪佣金、证监会交易征费、联交所交易费及会计及财务汇报局交易征费，合计每手20股发售股份须缴付港币3,333.28元。倘最终以下文"— 定价及分配"所述方式厘定的发售价低于最高发售价，将向成功申请人退还差额（包括就超额申请款项应付的经纪佣金、证监会交易征费、联交所交易费及会计及财务汇报局交易征费），不计利息（视乎申请渠道而定）。详情载于"如何申请香港发售股份"。

国际发售

初步发售的发售股份数目

国际发售将包括由本公司发售初步合共24,119,740股发售股份（须视乎重新分配、发售规模调整权及超额配股权而定），占全球发售初步可供认购发售股份总数约95%。国际发售项下初步发售的发售股份数目（须视乎国际发售与香港公开发售之间的任何重新分配而定）将占全球发售完成后本公司扩大后已发行股本的约7.9%（假设发售规模调整权及超额配股权均未获行使）。

– 495 –

the total number of Offer Shares initially being offered under the Global Offering; and

(b) the additional Offer Shares that may be issued pursuant to the Offer Size Adjustment Option;

(ii) the level of applications under the Hong Kong Public Offering;

(iii) the overall market conditions and whether they are conducive to a successful Listing; and

(iv) such other factors as the Overall Coordinators consider relevant.

– 497 –

全球发售架构

发售股份总数，即全球发售项下初始可供认购的发售股份与行使发售规模调整权后额外发售股份的总和；以及

(b) 超额配股权项下相应的股份数目；

(ii) 潜在专业及机构投资者在簿记建档过程中表示愿意购买发售股份的价格； (iii) 投资者素质，以期为本公司及其全体股东的利益建立稳固的专业机构投资者及股东基础；以及 (iv) 整体市场情况。

发售规模调整权的摊薄效应（假设超额配股权未获行使）如下：

| 全球发售项下于行使发售规模调整权前已发行股份数目（"原始认购人"） | 原始认购人于行使发售规模调整权前持有的占已发行总股本约百分比 | 全球发售项下于全面行使发售规模调整权后已发行股份数目 | 原始认购人于全面行使发售规模调整权后持有的占已发行总股本约百分比 | |---|---|---|---| | 25,389,220 | 8.3% | 29,197,600 | 8.2% |

发售规模调整权不会用于价格稳定目的，且不受香港法律第571W章《证券及期货（价格稳定）规则》的条文所规限。发售规模调整权将在超额配股权之外另行设立。

若发售规模调整权获全面行使，根据配发及发行额外股份所得的额外净收益将按比例依照本招股章程"未来计划及所得款项用途"一节所披露的分配方式进行分配。

– 498 –

全球发售架构

倘发售规模调整权已获行使，本公司将在其配发结果公告中披露行使情况及行使程度；若发售规模调整权于价格厘定日前尚未行使，本公司将确认该权利将告失效，且不得于日后任何日期行使。

价格稳定

价格稳定是承销商在部分市场用以便利证券分销的惯常做法。为进行价格稳定，承销商可在指定期间内在二级市场竞价或购入证券，以遏制（并于可能的情况下防止）证券在首次公开市场的价格跌破发售价。上述交易可在所有允许进行此类交易的司法管辖区内进行，且须遵守各相关地区的所有适用法律及监管规定，包括香港的相关规定。在香港，进行价格稳定的价格不得超过发售价。

就全球发售而言，稳定价格代理人（或其任何代理人）可代表承销商进行超额配发，或进行交易以在上市日期后有限期间内将股份市价稳定或维持于较其他情况下可能出现的水平为高的水平。然而，稳定价格代理人（或其任何代理人）并无责任采取任何上述稳定行动。如采取上述稳定行动，(a) 将由稳定价格代理人（或其任何代理人）全权酌情决定，并以稳定价格代理人（或其任何代理人）合理认为符合本公司最佳利益的方式进行，(b) 可随时终止，及 (c) 须于香港公开发售截止申请日期后30日内结束。

根据《证券及期货条例》的《证券及期货（价格稳定）规则》，在香港获准进行的价格稳定行动包括：(a) 超额配发以防止或尽量减少股份市价下跌；(b) 出售或同意出售股份，从而建立股份空仓以防止或尽量减少股份市价下跌；(c) 根据超额配股权购入或同意购入股份，以平仓上述第(a)或(b)段所建立的任何仓位；(d) 纯粹为防止或尽量减少股份市价下跌而购入或同意购入任何股份；(e) 出售或同意出售任何股份以了结因上述购入而建立的任何仓位；以及(f) 提出或尝试进行上述第(b)、(c)、(d)或(e)段所述的任何事宜。

具体而言，发售股份的潜在申请人及投资者应注意：

• 稳定价格代理人（或其任何代理人）可就稳定行动而持有股份好仓；

– 499 –

全球发售架构

• 稳定价格代理人（或其任何代理人）将维持该好仓的程度、时间或期限均存在不确定性；

• 稳定价格代理人（或其任何代理人）就任何该等好仓进行平仓及在公开市场出售股份，可能对股份市价产生不利影响；

•

STRUCTURE OF THE GLOBAL OFFERING

no stabilizing action can be taken to support the price of the Shares for longer than the stabilization period, which will begin on the Listing Date, and is expected to expire on Thursday, February 5, 2026, being the 30th day after the last day for lodging applications under the Hong Kong Public Offering. After this date, when no further stabilizing action may be taken, demand for the Shares, and therefore the price of the Shares, could fall;

•

the price of the Shares cannot be assured to stay at or above the Offer Price by the taking of any stabilizing action; and

•

stabilizing bids or transactions effected in the course of the stabilizing action may be made at any price at or below the Offer Price and can, therefore, be done at a price below the price paid by applicants for, or investors in, the Offer Shares.

The Company will ensure or procure that an announcement in compliance with the Securities and Futures (Price Stabilizing) Rules of the SFO will be made within seven days of the expiration of the stabilization period.

Over-allocation

Following any over-allocation of Shares in connection with the Global Offering, the Stabilizing Manager (or any person acting for it) may cover such over-allocations by, among others, exercising the Over-allotment Option in full or in part, using Shares purchased by the Stabilizing Manager (or any person acting for it) in the secondary market at prices that do not exceed the Offer Price or a combination of these means.

STOCK BORROWING AGREEMENT

In order to facilitate the settlement of over-allocations, if any, in connection with the Global Offering, the Stabilizing Manager (or any person acting for it) may choose to borrow up to 4,379,640 Offer Shares (representing approximately 15% of the Offer Shares offered under the Global Offering assuming the Offer Size Adjustment Option is exercised in full) or up to 3,808,380 Offer Shares (representing approximately 15% of the Offer Shares initially being offered under the Global Offering assuming the Offer Size Adjustment Option is not exercised) from MiniMax Matrix pursuant to the Stock Borrowing Agreement, which is expected to be entered into between the Stabilizing Manager (or any person acting for it) and MiniMax Matrix on or about the Price Determination Date.

– 500 –

STRUCTURE OF THE GLOBAL OFFERING

Such stock borrowing arrangement under the Stock Borrowing Agreements, if entered into, will not be subject to the restrictions of Rule 10.07(1)(a) of the Listing Rules provided that the requirements set out in Rule 10.07(3) of the Listing Rules are complied with.

Such stock borrowing arrangement is fully described in this prospectus and must be for the sole purpose of covering any short position prior to the exercise of the Over-allotment Option.

The same number of Shares so borrowed must be returned to MiniMax Matrix or its nominees, as the case may be, on or before the third business day following the earlier of (a) the last day on which the Over-allotment Option may be exercised and (b) the day on which the Over-allotment Option is exercised in full.

The stock borrowing arrangement described above will be effected in compliance with all applicable laws, rules and regulatory requirements. No payment will be made to MiniMax Matrix by the Stabilizing Manager (or any person acting for it) in relation to such Shares borrowing arrangement.

PRICING AND ALLOCATION

Determining the Pricing of the Offer Shares

Pricing for the Offer Shares for the purpose of the various offerings under the Global Offering will be determined on the Price Determination Date, which is expected to be on or before Wednesday, January 7, 2026 and, in any event, no later than 12:00 noon on Wednesday, January 7, 2026 by agreement between the Overall Coordinators (for themselves and on behalf of the Underwriters) and the Company, and the number of Offer Shares to be allocated under the various offerings will be determined shortly thereafter.

The Offer Price will not be more than HK$165.00 per Offer Share and is expected to be not less than HK$151.00 per Offer Share, unless otherwise announced, as further explained below. Applicants under the Hong Kong Public Offering may be required to pay, on application (subject to application channels), the maximum Offer Price plus brokerage of 1.0%, SFC transaction levy of 0.0027%, Stock Exchange trading fee of 0.00565% and AFRC transaction levy of 0.00015%, amounting to a total of HK$3,333.28 for one board lot of 20 Offer Shares.

Prospective investors should be aware that the Offer Price to be determined on the Price Determination Date may be, but is not expected to be, lower than the indicative Offer Price range stated in this prospectus.

The International Underwriters will be soliciting from prospective investors' indications of interest in acquiring Offer Shares in the International Offering. Prospective professional and institutional investors will be required to specify the number of Offer Shares under the International Offering they would be prepared to acquire either at different prices or at a particular price. This process, known as "book-building," is expected to continue up to, and to cease on or about, the last day for lodging applications under the Hong Kong Public Offering.

– 501 –

STRUCTURE OF THE GLOBAL OFFERING

The Overall Coordinators (for themselves and on behalf of the Underwriters) may, where it deems appropriate, based on the level of interest expressed by prospective investors during the book-building process in respect of the International Offering, and with the consent of the Company, reduce the number of Offer Shares offered below and/or the Offer Price range as stated in this prospectus at any time on or prior to the morning of the last day for lodging applications under the Hong Kong Public Offering. In such a case, the Company will, as soon as practicable following the decision to make such reduction, and in any event not later than the morning of the last day for lodging applications under the Hong Kong Public Offering, cause to be published on the websites of the Company and the Stock Exchange at https://www.minimaxi.com and www.hkexnews.hk, respectively, notices of the reduction, and the cancellation of the Global Offering and relaunch of the offer at the revised number of Offer Shares and/or the revised indicative Offer Price range.

The Company will also, as soon as practicable following the decision to make such change, issue a supplemental or new prospectus updating investors of the change in the number of Offer Shares and/or the indicative Offer Price range, and giving investors at least three business days to consider the new information. The supplemental or new prospectus should include at least the following: (i) updated Offer Price range and market capitalization; (ii) updated listing timetable and underwriting obligations; (iii) updated price/earnings multiple, unaudited pro forma and adjusted net tangible assets; and (iv) updated use of proceeds and confirmation of the working capital adequacy based on the revised estimated proceeds.

Before submitting applications for the Hong Kong Offer Shares, applicants should have regard to the possibility that any announcement of a reduction in the number of Offer Shares and/or the indicative Offer Price range may not be made until the last day for lodging applications under the Hong Kong Public Offering. In the absence of any such notice or announcement and any such supplemental or new prospectus so published, the number of Offer Shares and the indicative Offer Price range will not be reduced and/or the Offer Price, if agreed upon by the Overall Coordinators (for themselves and on behalf of the other Underwriters) and the Company, will under no circumstances be set outside the Offer Price range as stated in this prospectus.

If there is any change to the offer size due to change in the number of Offer Shares offered in the Global Offering (other than pursuant to the reallocation mechanism as disclosed in this prospectus), or change to the Offer Price falling outside the indicative Offer Price range as stated in this prospectus, or if the Company becomes aware that there has been a significant change affecting any matter contained in this prospectus or a significant new matter has arisen, the inclusion of information in respect of which would have been required to be in this prospectus if it had arisen before this prospectus was issued, after the issue of this prospectus and before the commencement of dealings in our Shares as prescribed under Rule 11.13 of the Listing Rules, we are required to cancel the Global Offering and relaunch the offer with a supplemental prospectus or a new prospectus in FINI.

– 502 –

STRUCTURE OF THE GLOBAL OFFERING

Announcement of Final Pricing of the Offer Shares

The final Offer Price, the level of indications of interest in the International Offering, the level of applications in the Hong Kong Public Offering, the basis of allocations of the Hong Kong Offer Shares and the results of allocations in the Hong Kong Public Offering are expected to be made available through a variety of channels in the manner described in "How to Apply for Hong Kong Offer Shares — B. Publication of Results."

UNDERWRITING

The Hong Kong Public Offering is fully underwritten by the Hong Kong Underwriters under the terms and conditions of the Hong Kong Underwriting Agreement and is subject to, among other things, the Overall Coordinators (for themselves and on behalf of the Underwriters) and the Company agreeing on the Offer Price.

The Company expects to enter into the International Underwriting Agreement relating to the International Offering on the Price Determination Date.

These underwriting arrangements, including the Underwriting Agreements, are summarized in "Underwriting."

CONDITIONS OF THE GLOBAL OFFERING

Acceptance of all applications for Offer Shares will be conditional on:

• the Stock Exchange granting approval for the listing of, and permission to deal in, the Shares to be issued pursuant to the Global Offering on the Main Board of the Stock Exchange and such approval not subsequently having been withdrawn or revoked prior to the Listing Date;

• the Offer Price having been agreed between the Overall Coordinators (for themselves and on behalf of the Underwriters) and the Company;

• the execution and delivery of the International Underwriting Agreement on or around the Price Determination Date; and

• the obligations of the Hong Kong Underwriters under the Hong Kong Underwriting Agreement and the obligations of the International Underwriters under the International Underwriting Agreement becoming and remaining unconditional and not having been terminated in accordance with the terms of the respective Underwriting Agreements,

– 503 –

STRUCTURE OF THE GLOBAL OFFERING

在各情況下須於各承銷協議所列明的日期及時間或之前達成（除非及在該等條件在該等日期及時間或之前獲有效豁免的範圍內），且無論如何不得遲於本招股說明書日期後30天。

若因任何原因，總協調人（代表其自身及承銷商）與公司未能於2026年1月7日（星期三）中午12:00前就發售價達成協議，全球發售將不會進行並將失效。

香港公開發售及國際發售各自的完成，均以（其中包括）另一發售成為無條件並未根據其條款被終止為條件。

若上述條件未能於指定日期及時間前獲達成或豁免，全球發售將失效，聯交所將即時獲得通知。香港公開發售失效的通知將由公司於失效後翌日在公司網站https://www.minimaxi.com及聯交所網站www.hkexnews.hk分別刊登。在此情況下，所有申請股款將予退還，不計利息，退款條款載於「如何申請香港發售股份 — D. 股票寄發╱領取及申請股款退還」。與此同時，所有申請股款將存放於接受銀行或其他根據香港法例第155章《銀行業條例》在香港獲發牌的銀行的獨立銀行賬戶。

發售股份的股票僅於2026年1月9日（星期五）上午8:00起成為有效的所有權憑證，前提是全球發售須於該時間或之前在所有方面成為無條件。

股份交易

假設香港公開發售於2026年1月9日（星期五）香港時間上午8:00或之前成為無條件，預期A類普通股將於2026年1月9日（星期五）上午9:00起在聯交所開始買賣。

A類普通股將以每手20股A類普通股進行買賣，A類普通股的股份代號為0100。

— 504 —

STRUCTURE OF THE GLOBAL OFFERING

根據證監會操守準則第21段向資本市場中介機構及準投資者發出的通知

向資本市場中介機構（包括私人銀行）發出的重要通知

本通知向資本市場中介機構（包括私人銀行）概述《持牌人或註冊人操守準則》（「操守準則」）對資本市場中介機構施加的若干義務，該等義務需要其他資本市場中介機構（包括私人銀行）的關注及配合。某些資本市場中介機構亦可能擔任本次發售的總協調人，並須遵守操守準則下的額外規定。

操守準則第21.3.3(c)段規定，資本市場中介機構應採取一切合理步驟，識別投資者是否可能與公司存在任何關聯，並向總協調人提供足夠資料，以便其評估該等投資者所下達的訂單是否可能對價格發現過程產生負面影響。

根據操守準則，屬於公司、資本市場中介機構或其集團公司的董事、僱員或主要股東的準投資者，將被視為與公司、該資本市場中介機構或相關集團公司（視情況而定）存在「關聯」（「關聯關係」）。資本市場中介機構在提交發售股份訂單時，應特別披露其投資者客戶是否存在任何關聯關係。此外，私人銀行應採取一切合理步驟，識別其投資者客戶是否可能與公司或任何資本市場中介機構（包括其集團公司）存在任何關聯關係，並相應通知承銷商。

根據上市規則及聯交所不時發出的其他監管規定或指引（「聯交所規定」），發售股份的分配將受到限制或須事先獲得聯交所同意的準投資者（例如上市發行人的關連人士）將被視為「受限投資者」。發售股份只可按照適用的聯交所規定分配予受限投資者。資本市場中介機構在提交發售股份訂單時，應特別披露其投資者客戶是否為受限投資者。

資本市場中介機構獲悉，本次發售的市場推廣及投資者目標策略包括機構投資者、純多頭投資者、主權財富基金、養老基金、對沖基金，各情況均須遵守適用的聯交所規定（就聯交所上市發行人而言）及本招股說明書其他部分所載的銷售限制。資本市場中介機構應確保所下達的訂單屬真實訂單，並無虛報，且不構成重複訂單（即通過兩個或以上資本市場中介機構下達兩個或以上相應或相同訂單）。資本市場中介機構應就任何看似不尋常或不規則的訂單向其投資者客戶查詢。資本市場中介機構在提交發售股份訂單時，應披露所有投資者的身份（綜合訂單除外，其中相關投資者資料

— 505 —

STRUCTURE OF THE GLOBAL OFFERING

should be provided to the Overall Coordinators when submitting orders). Failure to provide underlying investor information for omnibus orders, where required to do so, may result in that order being rejected. CMIs should not place "X-orders" into the order book.

CMIs should segregate and clearly identify their own proprietary orders (and those of their group companies, including private banks as the case may be) in the order book and book messages.

CMIs (including private banks) should not offer any rebates to prospective investors or pass on any rebates provided by the Company. In addition, CMIs (including private banks) should not enter into arrangements which may result in prospective investors paying different prices for the Offer Shares.

The Code requires that a CMI disclose complete and accurate information in a timely manner on the status of the order book and other relevant information it receives to targeted investors for them to make an informed decision. In order to do this, those Underwriters in control of the order book should consider disclosing order book updates to all CMIs.

When placing an order for the Offer Shares, private banks should disclose, at the same time, if such order is placed other than on a "principal" basis (whereby it is deploying its own balance sheet for onward selling to investors). Private banks who do not provide such disclosure are hereby deemed to be placing their order on such a "principal" basis. Private banks who disclose that they are placing their order other than on a "principal" basis (i.e. they are acting as an agent) should note that such order may be considered to be an omnibus order pursuant to the Code. Private banks should be aware that if any of their group companies is a CMI of this offering, placing an order on a "principal" basis may require the Underwriters to apply the "proprietary orders" of the Code to such order and will require the Underwriters to apply the "rebates" requirements of the Code to such order.

In relation to omnibus orders, when submitting such orders, CMIs (including private banks) are requested to provide the underlying investor information, preferably in Excel Workbook format, in respect of each order constituting the relevant omnibus order (failure to provide such information may result in that order being rejected). To the extent information being disclosed by CMIs and investors is personal and/or confidential in nature, CMIs (including private banks) agree and warrant: (A) to take appropriate steps to safeguard the transmission of such information to the Overall Coordinators; (B) that they have obtained the necessary consents from the underlying investors to disclose such information to the Overall Coordinators. By submitting an order and providing such information to the Overall Coordinators, each CMI (including private banks) further warrants that they and the underlying investors have understood and consented to the collection, disclosure, use and transfer of such information by the Overall Coordinators and/or any other third parties as may be required by the Code, including to the Company, relevant regulators and/or any other third parties as may be required by the Code, for the purpose of complying with the Code, during the book-building process for this offering. CMIs that receive such underlying investor information are reminded that such information should be used only for submitting orders in this offering. The

– 506 –

STRUCTURE OF THE GLOBAL OFFERING

承销商可能被要求证明其遵守《守则》项下义务的情况，并可要求其他资本市场中介机构（包括私人银行）提供证据，证明其遵守上述义务（特别是已取得必要的同意）。在此情况下，其他资本市场中介机构（包括私人银行）须在所要求的时限内向相关承销商提供上述证据。

致潜在投资者的重要告示

潜在投资者须注意，本次发售要约股份过程中的某些中介机构，包括某些承销商，均为《守则》第21段所规范的资本市场中介机构。本告示为《守则》对该等资本市场中介机构所施加的若干义务的摘要，需要潜在投资者的注意与配合。某些资本市场中介机构亦可能担任本次发售的整体协调人，并须遵守《守则》项下的额外规定。

凡属本公司、某资本市场中介机构或其集团公司的董事、雇员或主要股东的潜在投资者，根据《守则》将被视为与本公司、该资本市场中介机构或相关集团公司（视情况而定）存在"关联关系"。与本公司或任何资本市场中介机构（包括其集团公司）存在关联关系的潜在投资者，在认购要约股份时应特别披露此关联关系，并同时披露其认购指令是否可能对本次发售的价格发现过程产生负面影响。未披露其关联关系的潜在投资者，在此被视为不存在该等关联关系。凡潜在投资者披露其关联关系但未披露其认购指令可能对本次发售的价格发现过程产生负面影响的，该等认购指令在此被视为不会对本次发售的价格发现过程产生负面影响。

根据联交所规定，其要约股份的配发须受到限制或须事先获得联交所批准的潜在投资者（例如上市发行人的关联人士），将被视为"受限投资者"。要约股份仅可按照联交所适用规定向受限投资者配发。属于受限投资者的潜在投资者，在认购要约股份时应特别披露其受限投资者身份。未披露受限投资者身份的潜在投资者，在此被视为并非受限投资者。

潜在投资者应确保其所提交的认购指令均为真实、未经虚报且不构成重复认购指令（即通过两个或以上资本市场中介机构提交两个或以上相同或相似的认购指令），且通过提交认购指令即被视为确认上述事项。若某潜在投资者为任何承销商的关联资产管理部门，该潜在投资者在提交认购指令时应注明其认购是否针对该承销商或其集团公司持有逾50%权益的基金或投资组合，若属此情况，该指令将被归类为"自营指令"，并由资本市场中介机构按照《守则》进行适当处理，同时应披露该等"自营指令"是否可能对本次发售的价格发现过程产生负面影响。在提交认购指令时未注明上述信息的潜在投资者，在此被视为确认其认购指令并非该等"自营指令"。若某潜在投资者因其他原因与任何承销商存在关联，使其认购指令可能被视为"自营指令"（依据《守则》），该潜在投资者在提交认购指令时应向相关承销商注明，且该等指令须遵守《守则》的适用规定。在提交认购指令时未注明上述信息的潜在投资者，在此被视为确认其认购指令并非该等"自营指令"。凡潜在投资者披露上述信息但未披露该等"自营指令"可能对本次发售的价格发现过程产生负面影响的，该等"自营指令"在此被视为不会对本次发售的价格发现过程产生负面影响。

潜在投资者须注意，资本市场中介机构（包括私人银行）可能披露属于潜在投资者个人性质及／或保密性质的若干信息。通过提交认购指令，潜在投资者被视为已理解并同意承销商及／或任何其他第三方按照《守则》的要求对上述信息进行收集、披露、使用及转让，包括向本公司、整体协调人、相关监管机构及／或《守则》可能要求的任何其他第三方披露，并理解及同意该等信息仅用于在本次发售簿记建档过程中遵守《守则》之目的。未能提供上述信息可能导致相关认购指令被拒绝。

— 508 —

如何申请香港发售股份

致香港发售股份投资者的重要告示

全电子化申请程序

本公司已就香港公开发售采用全电子化申请程序，以下为申请程序的相关说明。

本招股章程载于联交所网站www.hkexnews.hk的"披露易 > 新上市 > 新上市资料"栏目，以及本公司网站https://www.minimaxi.com。

本招股章程的内容与根据《公司（清盘及杂项条文）条例》第342C条在香港公司注册处登记的招股章程内容完全相同。

A.

APPLICATION FOR HONG KONG OFFER SHARES

1.

Who Can Apply

You can apply for Hong Kong Offer Shares if you or the person(s) for whose benefit you are applying for: •

are 18 years of age or older; and

•

have a Hong Kong address (for the HK eIPO White Form service only); and

•

are outside the United States (within the meaning of Regulation S) or are a person described in paragraph (h)(3) of Rule 902 of Regulation S.

Unless permitted by the Listing Rules or a waiver and/or consent has been granted by the Stock Exchange to the Company, you cannot apply for any Hong Kong Offer Shares if you or the person(s) for whose benefit you are applying for: •

are an existing Shareholder;

•

are a Director or chief executive of the Company and/or a director or chief executive of any of its subsidiaries;

•

are a close associate (as defined in the Listing Rules) of any of the above persons;

•

are a connected person (as defined in the Listing Rules) of the Company or will become a connected person of the Company immediately upon the completion of the Global Offering; or

•

have been allocated or have applied for or indicated an interest in any International Offer Shares or otherwise participate in the International Offering. – 509 –

HOW TO APPLY FOR HONG KONG OFFER SHARES 2.

Application Channels

The Hong Kong Public Offering period will begin at 9:00 a.m. on Wednesday, December 31, 2025 and end at 12:00 noon on Tuesday, January 6, 2026 (Hong Kong time). To apply for Hong Kong Offer Shares, you may use one of the following application channels:

| Application Channel | Platform | Target Investors | Application Time | |---|---|---|---| | HK eIPO White Form service | www.hkeipo.hk | Applicants who would like to receive a physical Share certificate. Hong Kong Offer Shares successfully applied for will be allotted and issued in your own name. | From 9:00 a.m. on Wednesday, December 31, 2025 until 11:30 a.m. on Tuesday, January 6, 2026, Hong Kong time. | | HKSCC EIPO channel | Your broker or custodian who is a HKSCC Participant will submit an EIPO application on your behalf through HKSCC's FINI system in accordance with your instruction | Applicants who would not like to receive a physical Share certificate. Hong Kong Offer Shares successfully applied for will be allotted and issued in the name of HKSCC Nominees, deposited directly into CCASS and credited to your designated HKSCC Participant's stock account. | Contact your broker or custodian for the earliest and latest time for giving such instructions, as this may vary by broker or custodian. The latest time for completing full payment of application monies in respect of such applications will be 12:00 noon on Tuesday, January 6, 2026, Hong Kong time. |

The HK eIPO White Form service and the HKSCC EIPO channel are facilities subject to capacity limitations and potential service interruptions and you are advised not to wait until the last day of the application period to apply for Hong Kong Offer Shares.

– 510 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

For those applying through the HK eIPO White Form service, once you complete payment in respect of any application instructions given by you or for your benefit through the HK eIPO White Form service to make an application for Hong Kong Offer Shares, an actual application shall be deemed to have been made. If you are a person for whose benefit the electronic application instructions are given, you shall be deemed to have declared that only one set of electronic application instructions has been given for your benefit. If you are an agent for another person, you shall be deemed to have declared that you have only given one set of electronic application instructions for the benefit of the person for whom you are an agent and that you are duly authorized to give those instructions as an agent.

For the avoidance of doubt, giving an application instruction under the HK eIPO White Form service more than once and obtaining different application reference numbers without effecting full payment in respect of a particular reference number will not constitute an actual application.

If you apply through the HK eIPO White Form service, you are deemed to have authorized the HK eIPO White Form service to apply on the terms and conditions in this prospectus, as supplemented and amended by the terms and conditions of the HK eIPO White Form service.

By instructing your broker or custodian to apply for the Hong Kong Offer Shares on your behalf through the HKSCC EIPO Channel, you (and, if you are joint applicants, each of you jointly and severally) are deemed to have instructed and authorized HKSCC to cause HKSCC Nominees (acting as nominee for the relevant HKSCC Participants) to apply for Hong Kong Offer Shares on your behalf and to do on your behalf all the things stated in this prospectus and any supplement to it.

For those applying through the HKSCC EIPO channel, an actual application will be deemed to have been made for any application instructions given by you or for your benefit to HKSCC (in which case an application will be made by HKSCC Nominees on your behalf) provided such application instruction has not been withdrawn or otherwise invalidated before the closing time of the Hong Kong Public Offering.

HKSCC Nominees will only be acting as a nominee for you and neither HKSCC nor HKSCC Nominees shall be liable to you or any other person in respect of any actions taken by HKSCC or HKSCC Nominees on your behalf to apply for Hong Kong Offer Shares or for any breach of the terms and conditions of this prospectus.

Section 40 of the Companies (Winding Up and Miscellaneous Provisions) Ordinance

For the avoidance of doubt, the Company and all other parties involved in the preparation of this prospectus acknowledge that each applicant who gives or causes to give electronic application instructions is a person who may be entitled to compensation under section 40 of the Companies (Winding Up and Miscellaneous Provisions) Ordinance.

– 511 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

3. Information Required to Apply

You must provide the following information with your application:

| For Individual/Joint Applicants | For Corporate Applicants | |---|---| | • Full name(s)² as shown on your identity document | • Full name(s)² as shown on your identity document | | • Identity document's issuing country or jurisdiction | • Identity document's issuing country or jurisdiction | | • Identity document type, with order of priority: | • Identity document type, with order of priority: | | i. HKID card; or | i. LEI registration document; or | | ii. National identification document; or | ii. Certificate of incorporation; or | | iii. Passport; and | iii. Business registration certificate; or | | | iv. Other equivalent document; and | | • Identity document number | • Identity document number |

Notes:

1. If you are applying through the HK eIPO White Form service, you are required to provide a valid e-mail address, a contact telephone number and a Hong Kong Address. You are also required to declare that the identity information provided by you follows the requirements as described in Note 2 below. In particular, where you cannot provide a HKID number, you must confirm that you do not hold a HKID card. The number of joint applicants may not exceed four. If you are a firm, the applicant must be in the individual members' names.

2. The applicant's full name as shown on their identity document must be used and the surname, given name, middle and other names (if any) must be input in the same order as shown on the identity document. If an applicant's identity document contains both an English and Chinese name, both English and Chinese names must be used. Otherwise, either English or Chinese names will be accepted. The order of priority of the applicant's identity document type must be strictly followed and where an individual applicant has a valid HKID card (including both Hong Kong Residents and Hong Kong Permanent Residents), the HKID number must be used when making an application to subscribe for Hong Kong Offer Shares in the Hong Kong Public Offering. Similarly for corporate applicants, a LEI number must be used if an entity has a LEI certificate.

3. If the applicant is a trustee, the client identification data ("CID") of the trustee, as set out above, will be required. If the applicant is an investment fund (i.e. a collective investment scheme, or CIS), the CID of the asset management company or the individual fund, as appropriate, which has opened a trading account with the broker will be required, as above.

4. The maximum number of joint applicants on FINI is capped at 4 in accordance with market practice.

– 512 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

5. If you are applying as a nominee, you must provide: (i) the full name (as shown on the identity document), the identity document's issuing country or jurisdiction, the identity document type; and (ii) the identity document number, for each of the beneficial owners or, in the case(s) of joint beneficial owners, for each joint beneficial owner. If you do not include this information, the application will be treated as being made for your benefit.

6. If you are applying as an unlisted company and (i) the principal business of that company is dealing in securities; and (ii) you exercise statutory control over that company, then the application will be treated as being for your benefit and you should provide the required information in your application as stated above.

"Unlisted company" means a company with no equity securities listed on the Stock Exchange or any other stock exchange.

"Statutory control" means you:

• control the composition of the board of directors of the company;

• control more than half of the voting power of the company; or

• hold more than half of the issued share capital of the company (not counting any part of it which carries no right to participate beyond a specified amount in a distribution of either profits or capital).

For those applying through the HKSCC EIPO channel, and making an application under a power of attorney, the Company and the Overall Coordinators, as the Company's agents, have discretion to consider whether to accept it on any conditions the Company thinks fit, including evidence of the attorney's authority.

Failing to provide any required information may result in your application being rejected.

4. Permitted Number of Hong Kong Offer Shares for Application

| Board lot size | : | 20 Shares for one board lot | |---|---|---| | Permitted number of Hong Kong Offer Shares for application and amount payable on application/successful allotment | : | Hong Kong Offer Shares are available for application in specified board lot sizes only. Please refer to the amount payable associated with each specified board lot size in the table below. The maximum Offer Price is HK$165.00 per Share. If you are applying through the HKSCC EIPO channel, your broker or custodian may require you to pre-fund your application, in such amount as determined by the broker or custodian, based on the applicable laws and regulations in Hong Kong. You are responsible for complying with any such pre-funding requirement imposed by your broker or custodian with respect to the Hong Kong Offer Shares you applied for. |

– 513 –

HOW TO APPLY FOR HONG KONG OFFER SHARES By instructing your broker or custodian to apply for the Hong Kong Offer Shares on your behalf through the HKSCC EIPO Channel, you (and, if you are joint applicants, each of you jointly and severally) are deemed to have instructed and authorized HKSCC to cause HKSCC Nominees (acting as nominee for the relevant HKSCC Participants) to arrange payment of the final Offer Price, brokerage, SFC transaction levy, the Stock Exchange trading fee and the AFRC transaction levy by debiting the relevant nominee bank account at the designated bank for your broker or custodian. If you are applying through the HK eIPO White Form service, you may refer to the table below for the amount payable for the number of Hong Kong Offer Shares you have selected. You must pay the respective maximum amount payable on application in full upon application for Hong Kong Offer Shares.

| No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | No. of Hong Kong Offer Shares applied for | Maximum Amount payable(2) on application/ successful allotment HK$ | |---|---|---|---|---|---|---|---| | 20 | 3,333.28 | 400 | 66,665.61 | 6,000 | 999,984.16 | 80,000 | 13,333,122.00 | | 40 | 6,666.56 | 500 | 83,332.01 | 7,000 | 1,166,648.18 | 90,000 | 14,999,762.26 | | 60 | 9,999.84 | 600 | 99,998.41 | 8,000 | 1,333,312.20 | 100,000 | 16,666,402.50 | | 80 | 13,333.13 | 700 | 116,664.82 | 9,000 | 1,499,976.23 | 200,000 | 33,332,805.00 | | 100 | 16,666.40 | 800 | 133,331.22 | 10,000 | 1,666,640.26 | 300,000 | 49,999,207.50 | | 120 | 19,999.68 | 900 | 149,997.62 | 20,000 | 3,333,280.50 | 400,000 | 66,665,610.00 | | 140 | 23,332.96 | 1,000 | 166,664.03 | 30,000 | 4,999,920.76 | 500,000 | 83,332,012.50 | | 160 | 26,666.24 | 2,000 | 333,328.06 | 40,000 | 6,666,561.00 | 634,740(1) | 105,788,323.23 | | 180 | 29,999.52 | 3,000 | 499,992.08 | 50,000 | 8,333,201.26 | | | | 200 | 33,332.80 | 4,000 | 666,656.10 | 60,000 | 9,999,841.50 | | | | 300 | 49,999.21 | 5,000 | 833,320.13 | 70,000 | 11,666,481.76 | | |

(1) Maximum number of Hong Kong Offer Shares you may apply for and this is 50% of the Hong Kong Offer Shares initially offered.

(2) The amount payable is inclusive of brokerage, SFC transaction levy, the Stock Exchange trading fee and AFRC transaction levy. If your application is successful, brokerage will be paid to the Exchange Participants (as defined in the Listing Rules) or to the HK eIPO White Form Service Provider (for applications made through the application channel of the HK eIPO White Form service) while the SFC transaction levy, the Stock Exchange trading fee and the AFRC transaction levy will be paid to the SFC, the Stock Exchange and the AFRC, respectively.

– 514 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

5. Multiple Applications Prohibited

You or your joint applicant(s) shall not make more than one application for your own benefit, except where you are a nominee and provide the information of the underlying investor in your application as required under the paragraph headed "— A. Application for Hong Kong Offer Shares — 3. Information Required to Apply" in this section. If you are suspected of submitting or cause to submit more than one application, all of your applications will be rejected.

Multiple applications made either through (i) the HK eIPO White Form service, (ii) HKSCC EIPO channel, or (iii) both channels concurrently are prohibited and will be rejected. If you have made an application through the HK eIPO White Form service or HKSCC EIPO channel, you or the person(s) for whose benefit you have made the application shall not apply for any International Offer Shares.

The Hong Kong Share Registrar would record all applications into its system and identify suspected multiple applications with identical names and identification document numbers according to the Best Practice Note on Treatment of Multiple/Suspected Multiple Applications ("Best Practice Note") issued by the Federation of Share Registrars Limited.

Since applications are subject to personal information collection statements, identification document numbers displayed are redacted.

6. Terms and Conditions of An Application

By applying for Hong Kong Offer Shares through the HK eIPO White Form service or HKSCC EIPO channel, you (or as the case may be, HKSCC Nominees will do the following things on your behalf):

(i) undertake to execute all relevant documents and instruct and authorize the Company and/or the Overall Coordinators, as the Company's agents, to execute any documents for you and to do on your behalf all things necessary to register any Hong Kong Offer Shares allocated to you in your name or in the name of HKSCC Nominees as required by the Articles of Association, and (if you are applying through the HKSCC EIPO channel) to deposit the allotted Hong Kong Offer Shares directly into CCASS for the credit of your designated HKSCC Participant's stock account on your behalf;

(ii) confirm that you have read and understand the terms and conditions and application procedures set out in this prospectus, the designated website of the HK eIPO White Form service (or as the case may be, the agreement you entered into with your broker or custodian), and agree to be bound by them;

(iii) (if you are applying through the HKSCC EIPO channel) agree to the arrangements, undertakings and warranties under the participant agreement between your broker or custodian and HKSCC and observe the General Rules of HKSCC and the HKSCC Operational Procedures for giving application instructions to apply for Hong Kong Offer Shares;

– 515 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

(iv) confirm that you are aware of the restrictions on offers and sales of Shares set out in this prospectus and they do not apply to you, or the person(s) for whose benefit you have made the application;

(v)

HOW TO APPLY FOR HONG KONG OFFER SHARES

confirm that you have read this prospectus and any supplement to it and have relied only on the information and representations contained therein in making your application (or as the case may be, causing your application to be made) and will not rely on any other information or representations;

(vi) agree that the Company, the Joint Sponsors, the Overall Coordinators, the Joint Global Coordinators, the Joint Bookrunners, the Joint Lead Managers, the Capital Market Intermediaries, the Underwriters, any of their respective directors, officers, employees, partners, agents, advisers and any other parties involved in the Global Offering (the "Relevant Persons"), the Hong Kong Share Registrar and HKSCC will not be liable for any information and representations not in this prospectus and any supplement to it;

(vii) agree to disclose the details of your application and your personal data and any other personal data which may be required about you and the person(s) for whose benefit you have made the application to the Company, the Relevant Persons, the Hong Kong Share Registrar, HKSCC, HKSCC Nominees, the Stock Exchange, the SFC and any other statutory regulatory or governmental bodies or otherwise as required by laws, rules or regulations, for the purposes under the paragraph headed "— G. Personal Data — 3. Purposes and 4. Transfer of personal data" in this section;

(viii) agree (without prejudice to any other rights which you may have once your application (or as the case may be, HKSCC Nominees' application) has been accepted) that you will not rescind it because of an innocent misrepresentation;

(ix) agree that subject to Section 44A(6) of the Companies (Winding Up and Miscellaneous Provisions) Ordinance, any application made by you or HKSCC Nominees on your behalf cannot be revoked once it is accepted, which will be evidenced by the notification of the result of the ballot by the Hong Kong Share Registrar by way of publication of the results at the time and in the manner as specified in "— B. Publication of Results" in this section;

(x) confirm that you are aware of the situations specified in the "— C. Circumstances in which You Will Not Be Allocated Hong Kong Offer Shares" in this section;

(xi) agree that your application or HKSCC Nominees' application, any acceptance of it and the resulting contract will be governed by and construed in accordance with the laws of Hong Kong;

– 516 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

(xii) agree to comply with the Companies Ordinance, the Companies (Winding Up and Miscellaneous Provisions) Ordinance, the Articles of Association and laws of any place outside Hong Kong that apply to your application and that neither the Company nor the Relevant Persons will breach any law inside and/or outside Hong Kong as a result of the acceptance of your offer to purchase, or any action arising from your rights and obligations under the terms and conditions contained in this prospectus;

(xiii) confirm that (a) your application or HKSCC Nominees' application on your behalf is not financed directly or indirectly by the Company, any of the directors, chief executives, substantial shareholder(s) or existing shareholder(s) of the Company or any of its subsidiaries or any of their respective close associates; and (b) you are not accustomed or will not be accustomed to taking instructions from the Company, any of the directors, chief executives, substantial shareholder(s) or existing shareholder(s) of the Company or any of its subsidiaries or any of their respective close associates in relation to the acquisition, disposal, voting or other disposition of the Shares registered in your name or otherwise held by you;

(xiv) warrant that the information you have provided is true and accurate;

(xv) confirm that you understand that the Company and the Overall Coordinators will rely on your declarations and representations in deciding whether or not to allocate any Hong Kong Offer Shares to you and that you may be prosecuted for making a false declaration;

(xvi) agree to accept Hong Kong Offer Shares applied for or any lesser number allocated to you under the application;

(xvii) declare and represent that this is the only application made and the only application intended by you to be made to benefit you or the person for whose benefit you are applying;

(xviii) (if the application is made for your own benefit) warrant that no other application has been or will be made for your benefit by giving electronic application instructions to HKSCC directly or indirectly or through the application channel of the HK eIPO White Form service or by any one as your agent or by any other person; and

(xix) (if you are making the application as an agent for the benefit of another person) warrant that (1) no other application has been or will be made by you as agent for or for the benefit of that person or by that person or by any other person as agent for that person by giving electronic application instructions to HKSCC or the HK eIPO White Form Service Provider and (2) you have due authority to give electronic application instructions on behalf of that other person as its agent.

– 517 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

B. PUBLICATION OF RESULTS

**Results of Allocation**

You can check whether you are successfully allocated any Hong Kong Offer Shares through:

| Platform | Date/Time | |---|---| | **Applying through HK eIPO White Form service or HKSCC EIPO channel:** | | | Website | From the "Allotment Results" page on the designated results of allocations website at www.tricor.com.hk/ipo/result or www.hkeipo.hk/IPOResult | 24 hours, from 11:00 p.m. on Thursday, January 8, 2026 to 12:00 midnight on Wednesday, January 14, 2026 (Hong Kong time) |

HOW TO APPLY FOR HONG KONG OFFER SHARES

(i) 全部或部分成功申請人（使用香港eIPO白表服務及香港結算EIPO渠道）的完整名單，以及（ii）有條件配發予彼等的香港發售股份數目等資料，將於www.hkeipo.hk/IPOResult或www.tricor.com.hk/ipo/result上顯示。

電話 ў ў ў ў ў ў ў

聯交所網站www.hkexnews.hk及本公司網站https://www.minimaxi.com將提供連結至上述香港股份過戶登記處網站的連結。

不遲於2026年1月8日（星期四）晚上11時正（香港時間）

+852 3691 8488 — 由香港股份過戶登記處提供的配發結果電話查詢熱線

2026年1月9日（星期五）至2026年1月14日（星期三）（香港時間）上午9時至下午6時（星期六、日及香港公眾假期除外）

– 518 –

如何申請香港發售股份

就通過香港結算EIPO渠道申請者而言，閣下亦可於2026年1月7日（星期三）下午6時起（香港時間）向閣下的經紀或保管人查詢。香港結算參與者可於2026年1月7日（星期三）下午6時起（香港時間）以全天24小時方式登入FINI並查閱配發結果，如發現配發有任何差異，應盡快向香港結算反映。

**配發公告**

本公司預期於2026年1月8日（星期四）晚上11時正（香港時間）前，在聯交所網站www.hkexnews.hk及本公司網站https://www.minimaxi.com公布最終發售價結果、國際發售的意向指示水平、香港公開發售的申請水平及香港發售股份的配發基準。

---

C. 不獲配發香港發售股份的情況

請注意以下情況，在此等情況下，香港發售股份將不會配發予閣下或閣下代其申請之人士：

**1. 若閣下的申請被撤銷：**

閣下的申請或香港結算（代理人）有限公司代表閣下所作的申請，可根據《公司（清盤及雜項條文）條例》第44A(6)條予以撤銷。

**2. 若本公司或其代理人行使酌情權拒絕閣下的申請：**

本公司、總協調人、香港股份過戶登記處及彼等各自的代理人及代名人有絕對酌情權拒絕或接納任何申請，或僅接納任何申請的一部分，而毋須給予任何理由。

**3. 若香港發售股份的配發無效：**

倘聯交所未批准股份上市，香港發售股份的配發將告無效，無效情形如下：

- 於申請名單截止日期後三週內；或 - 若聯交所於申請名單截止日期後三週內通知本公司該較長期限，則於最長六週的較長期限內。

– 519 –

如何申請香港發售股份

**4. 若出現以下情況：**

- 閣下提出多項申請或疑似多項申請。就何謂多項申請，閣下可參閱本節「— A. 申請香港發售股份 — 5. 禁止多項申請」； - 閣下的申請指示不完整； - 閣下的付款（或視情況而定的資金確認）未能正確完成； - 包銷協議未能成為無條件或遭終止； - 本公司或總協調人認為，接納閣下的申請將使其或本公司違反適用的證券或其他法律、規則或法規。

**5. 若已配發發售股份出現資金交收失敗：**

根據香港結算參與者與香港結算之間的安排，香港結算參與者須於抽籤前在其指定銀行存入足夠的申請資金。香港發售股份抽籤後，收款銀行將從各香港結算參與者的指定銀行收取結算各香港結算參與者實際所獲配發香港發售股份所需的相應資金。

資金交收存在失敗的風險。在香港結算參與者（或其指定銀行）代表閣下結算已配發發售股份款項時，若出現資金交收失敗的極端情況，香港結算將聯絡違約的香港結算參與者及其指定銀行，以確定失敗原因，並要求該違約香港結算參與者糾正或促使糾正該失敗情況。

然而，若確定無法履行相關交收責任，受影響的香港發售股份將重新分配至國際發售。閣下透過經紀或保管人申請的香港發售股份，可能在資金交收失敗的範圍內受到影響。在極端情況下，由於該香港結算參與者發生資金交收失敗，閣下將不獲配發任何香港發售股份。如因資金交收失敗而導致香港發售股份未能配發予閣下，本公司、相關人士、香港股份過戶登記處及香港結算均不承擔或將不承擔任何責任。

– 520 –

如何申請香港發售股份

D. 股票過戶收據及申請股款退款的發送╱領取

HOW TO APPLY FOR HONG KONG OFFER SHARES

You will receive one Share certificate for all Hong Kong Offer Shares allotted to you under the Hong Kong Public Offering (except pursuant to applications made through the HKSCC EIPO channel where the Share certificates will be deposited into CCASS as described below).

No temporary document of title will be issued in respect of the Offer Shares. No receipt will be issued for sums paid on application.

Share certificates will only become valid evidence of title at 8:00 a.m. on Friday, January 9, 2026 (Hong Kong time), provided that the Global Offering has become unconditional and the right of termination described in the section headed "Underwriting" has not been exercised. Investors who trade Shares prior to the receipt of Share certificates or the Share certificates becoming valid do so entirely at their own risk.

The right is reserved to retain any Share certificate(s) and (if applicable) any surplus application monies pending clearance of application monies.

The following sets out the relevant procedures and time:

| | HK eIPO White Form service | HKSCC EIPO channel | |---|---|---| | **Despatch/collection of Share certificate(1)** | | | | For application of 500,000 Hong Kong Offer Shares or more | Collection in person from the Hong Kong Share Registrar, Tricor Investor Services Limited, at 17/F, Far East Finance Centre, 16 Harcourt Road, Hong Kong. Time: from 9:00 a.m. to 1:00 p.m. on Friday, January 9, 2026 (Hong Kong time), or any other place or date notified by the Company. If you are an individual, you must not authorize any other person to collect for you. | Share certificate(s) will be issued in the name of HKSCC Nominees, deposited into CCASS and credited to your designated HKSCC Participant's stock account. No action by you is required. |

– 521 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

| | HK eIPO White Form service | HKSCC EIPO channel | |---|---|---| | | If you are a corporate applicant, your authorized representative must bear a letter of authorization from your corporation stamped with your corporation's chop. Both individuals and authorized representatives must produce, at the time of collection, evidence of identity acceptable to the Hong Kong Share Registrar. Note: If you do not collect your Share certificate(s) personally within the time above, it/they will be sent to the address specified in your application instructions by ordinary post at your own risk. | | | For application of less than 500,000 Hong Kong Offer Shares | Your Share certificate(s) will be sent to the address specified in your application instructions by ordinary post at your own risk. Date: Thursday, January 8, 2026 | | | **Refund mechanism for surplus application monies paid by you** | | | | Date | Friday, January 9, 2026 | Subject to the arrangement between you and your broker or custodian | | Responsible party | Hong Kong Share Registrar | Your broker or custodian | | Application monies paid through single bank account | HK eIPO White Form e-Auto Refund payment instructions to your designated bank account | Your broker or custodian will arrange refund to your designated bank account subject to the arrangement between you and it. | | Application monies paid through multiple bank accounts | Refund check(s) will be despatched to the address as specified in your application instructions by ordinary post at your own risk. | |

– 522 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

Note: (1) Except in the event of a tropical cyclone warning signal number 8 or above, a black rainstorm warning and/or Extreme Conditions in force in Hong Kong in the morning on Thursday, January 8, 2026, rendering it impossible for the relevant Share certificates to be dispatched to HKSCC in a timely manner, in which case the Company shall procure the Hong Kong Share Registrar to arrange for delivery of the supporting documents and Share certificates in accordance with the contingency arrangements as agreed between them. You may refer to "— E. Severe Weather Arrangements" in this section.

E. SEVERE WEATHER ARRANGEMENTS

The Opening and Closing of the Application Lists

The application lists will not open or close on Tuesday, January 6, 2026 if, there is:

• a tropical cyclone warning signal number 8 or above;

• a black rainstorm warning; and/or

• an "extreme conditions" announcement issued after a super typhoon ("Extreme Conditions"),

(collectively, "Severe Weather Signals"),

in force in Hong Kong at any time between 9:00 a.m. and 12:00 noon on Tuesday, January 6, 2026.

Instead they will open between 11:45 a.m. and 12:00 noon and/or close at 12:00 noon on the next business day which does not have Severe Weather Signals in force at any time between 9:00 a.m. and 12:00 noon.

Prospective investors should be aware that a postponement of the opening/closing of the application lists may result in a delay in the Listing Date. Should there be any changes to the dates mentioned in the section headed "Expected Timetable" in this prospectus, an announcement will be made and published on the Stock Exchange's website at www.hkexnews.hk and the Company's website at https://www.minimaxi.com of the revised timetable.

If a Severe Weather Signal is hoisted on Thursday, January 8, 2026, the Hong Kong Share Registrar will make appropriate arrangements for the delivery of the Share certificates to the CCASS Depository's service counter so that they would be available for trading on Friday, January 9, 2026.

If a Severe Weather Signal is hoisted on Thursday, January 8, 2026, for application of less than 500,000 Hong Kong Offer Shares, the despatch of physical Share certificate(s) will be made by ordinary post when the post office re-opens after the Severe Weather Signal is lowered or cancelled (e.g. in the afternoon of Thursday, January 8, 2026 or on Friday, January 9, 2026).

– 523 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

If a Severe Weather Signal is hoisted on Friday, January 9, 2026, for application of 500,000 Hong Kong Offer Shares or more, the physical Share certificate(s) will be available for collection in person at the Hong Kong Share Registrar's office after the Severe Weather Signal is lowered or cancelled (e.g. in the afternoon of Friday, January 9, 2026 or on Monday, January 12, 2026).

Prospective investors should be aware that if they choose to receive physical Share certificates issued in their own name, there may be a delay in receiving the Share certificates.

F.

ADMISSION OF THE CLASS A ORDINARY SHARES INTO CCASS

If the Stock Exchange grants the listing of, and permission to deal in, the Class A Ordinary Shares on the Stock Exchange and the Company complies with the stock admission requirements of HKSCC, the Class A Ordinary Shares will be accepted as eligible securities by HKSCC for deposit, clearance and settlement in CCASS with effect from the date of commencement of dealings in the Class A Ordinary Shares on the Stock Exchange or any other date HKSCC chooses. Settlement of transactions between Exchange Participants (as defined in the Listing Rules) is required to take place in CCASS on the second settlement day after any trading day.

All activities under CCASS are subject to the General Rules of HKSCC and HKSCC Operational Procedures in effect from time to time.

All necessary arrangements have been made enabling the Class A Ordinary Shares to be admitted into CCASS.

You should seek the advice of your broker or other professional advisor for details of the settlement arrangement as such arrangements may affect your rights and interests.

G.

PERSONAL DATA

The following Personal Information Collection Statement applies to any personal data collected and held by the Company, the Hong Kong Share Registrar, the receiving bank and the Relevant Persons about you in the same way as it applies to personal data about applicants other than HKSCC Nominees. This personal data may include client identifier(s) and your identification information. By giving application instructions to HKSCC, you acknowledge that you have read, understood and agree to all of the terms of the Personal Information Collection Statement below.

– 524 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

1.

Personal Information Collection Statement

This Personal Information Collection Statement informs the applicant for, and holder of, Hong Kong Offer Shares, of the policies and practices of the Company and the Hong Kong Share Registrar in relation to personal data and the Personal Data (Privacy) Ordinance (Chapter 486 of the Laws of Hong Kong).

2.

Reasons for the collection of your personal data

It is necessary for applicants and registered holders of Hong Kong Offer Shares to ensure that personal data supplied to the Company or its agents and the Hong Kong Share Registrar is accurate and up-to-date when applying for Hong Kong Offer Shares or transferring Hong Kong Offer Shares into or out of their names or in procuring the services of the Hong Kong Share Registrar.

Failure to supply the requested data or supplying inaccurate data may result in your application for Hong Kong Offer Shares being rejected, or in the delay or the inability of the Company or the Hong Kong Share Registrar to effect transfers or otherwise render their services. It may also prevent or delay registration or transfers of Hong Kong Offer Shares which you have successfully applied for and/or the despatch of Share certificate(s) to which you are entitled.

It is important that applicants for and holders of Hong Kong Offer Shares inform the Company and the Hong Kong Share Registrar immediately of any inaccuracies in the personal data supplied.

3.

Purposes

Your personal data may be used, held, processed, and/or stored (by whatever means) for the following purposes:

• processing your application and refund check and HK eIPO White Form e-Auto Refund payment instruction(s), where applicable, verification of compliance with the terms and application procedures set out in this prospectus and announcing results of allocation of Hong Kong Offer Shares;

• compliance with applicable laws and regulations in Hong Kong and elsewhere;

• registering new issues or transfers into or out of the names of the holders of the Class A Ordinary Shares including, where applicable, HKSCC Nominees;

• maintaining or updating the register of members of the Company;

• verifying identities of applicants for and holders of the Class A Ordinary Shares and identifying any duplicate applications for the Class A Ordinary Shares;

– 525 –

HOW TO APPLY FOR HONG KONG OFFER SHARES

• facilitating Hong Kong Offer Shares balloting;

• establishing benefit entitlements of holders of the Class A Ordinary Shares, such as dividends, rights issues, bonus issues, etc.;

• distributing communications from the Company and its subsidiaries;

• compiling statistical information and profiles of the holders of the Class A Ordinary Shares;

• disclosing relevant information to facilitate claims on entitlements; and

• any other incidental or associated purposes relating to the above and/or to enable the Company and the Hong Kong Share Registrar to discharge their obligations to applicants and holders of the Class A Ordinary Shares and/or regulators and/or any other purposes to which applicants and holders of the Class A Ordinary Shares may from time to time agree.

4.

Transfer of personal data

Personal data held by the Company and the Hong Kong Share Registrar relating to the applicants for and holders of Hong Kong Offer Shares will be kept confidential but the Company and the Hong Kong Share Registrar may, to the extent necessary for achieving any of the above purposes, disclose, obtain or transfer (whether within or outside Hong Kong) the personal data to, from or with any of the following:

•

the Company's appointed agents such as financial advisers, receiving bank and overseas principal share registrar;

•

HKSCC or HKSCC Nominees, who will use the personal data and may transfer the personal data to the Hong Kong Share Registrar for the purposes of providing its services or facilities or performing its functions in accordance with its rules or procedures and operating FINI and CCASS (including where applicants for the Hong Kong Offer Shares request a deposit into CCASS);

•

any agents, contractors or third-party service providers who offer administrative, telecommunications, computer, payment or other services to the Company or the Hong Kong Share Registrar in connection with their respective business operation;

•

the Stock Exchange, the SFC and any other statutory regulatory or governmental bodies or otherwise as required by laws, rules or regulations, including for the purpose of the Stock Exchange's administration of the Listing Rules and the SFC's performance of its statutory functions; and

•

any persons or institutions with which the holders of Hong Kong Offer Shares have or propose to have dealings, such as their bankers, solicitors, accountants or brokers etc. – 526 –

HOW TO APPLY FOR HONG KONG OFFER SHARES 5.

Retention of personal data

The Company and the Hong Kong Share Registrar will keep the personal data of the applicants and holders of Hong Kong Offer Shares for as long as necessary to fulfil the purposes for which the personal data were collected. Personal data which is no longer required will be destroyed or dealt with in accordance with the Personal Data (Privacy) Ordinance (Chapter 486 of the Laws of Hong Kong). 6.

Access to and correction of personal data

Applicants for and holders of Hong Kong Offer Shares have the right to ascertain whether the Company or the Hong Kong Share Registrar hold their personal data, to obtain a copy of that data, and to correct any data that is inaccurate. The Company and the Hong Kong Share Registrar have the right to charge a reasonable fee for the processing of such requests. All requests for access to data or correction of data should be addressed to the Company and the Hong Kong Share Registrar, at their registered address disclosed in the section headed "Corporate information" in this prospectus or as notified from time to time, for the attention of the company secretary, or the Hong Kong Share Registrar for the attention of the privacy compliance officer.

– 527 –

APPENDIX I

ACCOUNTANT'S REPORT

Ernst & Young 27/F, One Taikoo Place 979 King's Road Quarry Bay, Hong Kong

Ernst & Young (安永会计师事务所) 979 King's Road, Quarry Bay, Hong Kong 27th Floor

Tel: +852 2846 9888 Fax: +852 2868 4432 ey.com

ACCOUNTANTS' REPORT ON HISTORICAL FINANCIAL INFORMATION TO THE DIRECTORS OF MINIMAX GROUP INC. AND CHINA INTERNATIONAL CAPITAL CORPORATION HONG KONG SECURITIES LIMITED AND UBS SECURITIES HONG KONG LIMITED

引言

我们就MINIMAX GROUP INC.（"本公司"）及其附属公司（合称"本集团"）载于第I-4至I-67页的历史财务资料报告，该等资料包括本集团截至2022年、2023年及2024年12月31日止各年度及截至2025年9月30日止九个月（"相关期间"）的综合损益表、综合收益表、综合权益变动表及综合现金流量表，以及本集团及本公司于2022年、2023年及2024年12月31日及2025年9月30日的综合财务状况表，连同重要会计政策资料及其他说明资料（合称"历史财务资料"）。载于第I-4至I-67页的历史财务资料构成本报告的组成部分，本报告已编制以纳入本公司日期为2025年12月31日的招股说明书（"招股章程"），有关本公司股份在香港联合交易所有限公司（"联交所"）主板首次上市事宜。

董事对历史财务资料的责任

本公司董事负责按历史财务资料附注2.1及2.2所载呈列基准及编制基准编制可呈现真实公正状况的历史财务资料，并负责建立董事认为必要的内部控制，以确保编制的历史财务资料不含有因欺诈或错误而导致的重大错误陈述。

申报会计师的责任

我们的责任是就历史财务资料发表意见，并向您报告我们的意见。我们根据香港会计师公会（"香港会计师公会"）颁布的《香港投资通函报告业务准则第200号——投资通函中历史财务资料的会计师报告》开展工作。该准则要求我们遵守道德标准，并规划和执行我们的工作，以合理确保历史财务资料不含重大错误陈述。

– I-1 –

附录一

会计师报告

APPENDIX I

ACCOUNTANT'S REPORT

关于香港联合交易所证券上市规则及《公司（清盘及杂项条文）条例》所列事项的报告

调整

在编制历史财务资料时，并未对第I-4页所界定的基础财务报表作出任何调整。

股息

我们参考历史财务资料附注11，该附注指出公司在相关期间内并未派付任何股息。

安永 执业会计师 香港 2025年12月31日

— I-3 —

附录一

会计师报告

---

我们的工作包括执行相关程序，以获取有关历史财务资料中金额及披露事项的证据。所选择的程序取决于申报会计师的判断，包括对历史财务资料（无论因舞弊或错误所致）存在重大错误陈述风险的评估。在作出该等风险评估时，申报会计师考虑与实体编制历史财务资料相关的内部控制，该等历史财务资料须按照历史财务资料附注2.1及2.2分别载列的呈列基准及编制基准，呈现真实公平的财务状况，以设计在当时情况下适当的程序，但并非为就实体内部控制的有效性发表意见。我们的工作亦包括评估所采用会计政策的恰当性及董事所作会计估计的合理性，以及评估历史财务资料的整体呈列。

我们相信，我们所获得的证据已充分及适当，足以为我们的意见提供基础。

意见

我们认为，历史财务资料就会计师报告的目的而言，依据历史财务资料附注2.1及2.2分别载列的呈列基准及编制基准，真实公平地反映了本集团及本公司于2022年、2023年及2024年12月31日及2025年9月30日的财务状况，以及本集团在各相关期间的财务表现及现金流量。

中期财务资料审阅

我们已审阅本集团的中期财务资料，该资料包括截至2024年9月30日止九个月的综合损益及其他全面收益表、权益变动表及现金流量表以及其他说明性资料（「中期财务资料」）。本公司董事负责按照历史财务资料附注2.1及2.2分别载列的呈列基准及编制基准编制及呈列中期财务资料。我们的责任是根据我们的审阅就中期财务资料发表结论。我们按照香港会计师公会颁布的《香港审阅委聘准则第2410号——由实体独立核数师执行的中期财务资料审阅》进行审阅。审阅工作包括主要向负责财务及会计事务的人员进行询问，以及采用分析性及其他审阅程序。审阅工作的范围远较按照香港审计准则进行的核数工作为窄，因此不能令我们取得保证，使我们能察觉到在核数工作中可能被识别的所有重大事项。因此，我们不发表核数意见。根据我们的审阅，没有任何事项引起我们注意，使我们认为中期财务资料就会计师报告的目的而言，未能在所有重大方面按照历史财务资料附注2.1及2.2分别载列的呈列基准及编制基准编制。

— I-2 —

附录一

会计师报告

关于香港联合交易所证券上市规则及《公司（清盘及杂项条文）条例》所列事项的报告

调整

在编制历史财务资料时，并未对第I-4页所界定的基础财务报表作出任何调整。

股息

我们参考历史财务资料附注11，该附注指出公司在相关期间内并未派付任何股息。

安永 执业会计师 香港 2025年12月31日

— I-3 —

附录一 I

会计师报告

HISTORICAL FINANCIAL INFORMATION

Preparation of Historical Financial Information Set out below is the Historical Financial Information which forms an integral part of this accountants' report. The financial statements of the Group for the Relevant Periods, on which the Historical Financial Information is based, were audited by Ernst & Young in accordance with Hong Kong Standards on Auditing issued by the HKICPA (the "Underlying Financial Statements"). The Historical Financial Information is presented in United States dollars ("USD") and all values are rounded to the nearest thousand (USD'000) except when otherwise indicated.

– I-4 –

APPENDIX I

ACCOUNTANT'S REPORT

CONSOLIDATED STATEMENTS OF PROFIT OR LOSS

| | | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---|---| | | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | REVENUE | 5 | – | 3,460 | 30,523 | 19,454 | 53,437 | | Cost of sales | 5 | – | (4,314) | (26,785) | (18,944) | (40,961) | | Gross (loss)/profit | | – | (854) | 3,738 | 510 | 12,476 | | Other income and gains, net | | 1,155 | 8,942 | 36,151 | 25,278 | 31,232 | | Selling and distribution expenses | | (587) | (22,827) | (86,995) | (53,389) | (39,325) | | Administrative expenses | | (3,213) | (7,615) | (14,384) | (9,610) | (22,074) | | Research and development expenses | | (10,560) | (70,002) | (188,979) | (138,684) | (180,312) | | Fair value loss on financial liabilities | | (60,509) | (176,826) | (214,172) | (128,063) | (313,477) | | Finance costs | 6 | (14) | (61) | (509) | (316) | (511) | | Impairment losses on financial assets, net | | (3) | (88) | (68) | (22) | (73) | | LOSS BEFORE TAX | 7 | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | Income tax expense | 10 | – | – | – | – | – | | LOSS FOR THE YEAR/PERIOD | | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) |

Attributable to: | Owners of the parent | | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | Non-controlling interests | | – | – | – | – | – | | | | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) |

LOSS PER SHARE ATTRIBUTABLE TO ORDINARY EQUITY HOLDERS OF THE PARENT Basic and diluted — For loss for the year/period (USD)

| | Notes | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | |---|---|---|---|---|---|---| | Basic and diluted | 12 | (0.74) | (2.56) | (4.28) | (2.80) | (4.71) |

– I-5 –

APPENDIX I

ACCOUNTANT'S REPORT

CONSOLIDATED STATEMENTS OF COMPREHENSIVE INCOME

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | LOSS FOR THE YEAR/PERIOD | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) |

OTHER COMPREHENSIVE INCOME/(LOSS) Other comprehensive income/(loss) to be reclassified to profit or loss in subsequent periods:

| Exchange differences on translation of foreign operations | 99 | 360 | 347 | (86) | (1,255) | |---|---|---|---|---|---| | Net other comprehensive income/(loss) to be reclassified to profit or loss in subsequent periods | 99 | 360 | 347 | (86) | (1,255) |

Other comprehensive income not to be reclassified to profit or loss in subsequent periods:

| Changes in fair value of equity investments designated at fair value through other comprehensive income | – | – | 662 | (839) | 1,604 | |---|---|---|---|---|---| | Net other comprehensive income not to be reclassified to profit or loss in subsequent periods | – | – | 662 | (839) | 1,604 |

| TOTAL COMPREHENSIVE LOSS FOR THE YEAR/PERIOD | (73,629) | (268,886) | (464,229) | (305,267) | (511,664) | |---|---|---|---|---|---|

Attributable to: | Owners of the parent | (73,629) | (268,886) | (464,229) | (305,267) | (511,664) | | Non-controlling interests | – | – | – | – | – | | | (73,629) | (268,886) | (464,229) | (305,267) | (511,664) |

– I-6 –

APPENDIX I

ACCOUNTANT'S REPORT

CONSOLIDATED STATEMENTS OF FINANCIAL POSITION

| | | As at 31 December | | | As at 30 September | |---|---|---|---|---|---| | | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | NON-CURRENT ASSETS | | | | | | | Property, plant and equipment | 13 | 231 | 709 | 1,093 | 1,134 | | Right-of-use assets | 14(a) | 458 | 3,313 | 3,077 | 2,746 | | Prepayments, other receivables and other assets | 16 | – | 435 | 561 | 731 | | Financial assets at fair value through profit or loss | 17 | – | – | 95,331 | 70,228 | | Financial assets at fair value through other comprehensive income | 17 | – | – | 4,836 | 6,440 | | Restricted cash | 18 | – | 39 | 38 | 41 | | Total non-current assets | | 689 | 4,496 | 104,936 | 81,320 | | CURRENT ASSETS | | | | | | | Trade receivables | 15 | – | 1,338 | 6,982 | 8,063 | | Prepayments, other receivables and other assets | 16 | 569 | 4,378 | 13,470 | 11,811 | | Financial assets at amortised cost | 17 | – | – | 147,444 | – | | Financial assets at fair value through profit or loss | 17 | 65,791 | 15,802 | 295,220 | 644,154 | | Restricted cash | 18 | 2,221 | – | 27,293 | 25,097 | | Time deposits | 18 | – | 91,698 | 26,327 | – | | Cash and cash equivalents | 18 | 4,691 | 206,295 | 288,912 | 362,647 | | Total current assets | | 73,272 | 319,511 | 805,648 | 1,051,772 | | CURRENT LIABILITIES | | | | | | | Interest-bearing bank borrowings | 19 | – | – | 19,455 | 19,102 | | Trade and bills payables | 20 | 2,394 | 17,242 | 51,212 | 70,219 | | Other payables, accruals and other liabilities | 21 | 2,326 | 14,741 | 51,512 | 17,322 | | Contract liabilities | 22 | – | 559 | 1,553 | 4,657 | | Lease liabilities | 14(b) | 349 | 1,248 | 1,964 | 1,694 | | Convertible redeemable preferred shares | 24 | 145,175 | 629,001 | 1,581,949 | 2,321,193 | | Total current liabilities | | 150,244 | 662,791 | 1,707,645 | 2,434,187 |

– I-7 –

APPENDIX I

ACCOUNTANT'S REPORT

| | | As at 31 December | | | As at 30 September | |---|---|---|---|---|---| | | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | NET CURRENT LIABILITIES | | (76,972) | (343,280) | (901,997) | (1,382,415) |

APPENDIX I

ACCOUNTANT'S REPORT

CONSOLIDATED STATEMENTS OF CHANGES IN DEFICITS Attributable to owners of the parent | | Share capital | Share option reserve* | Exchange fluctuation reserve* | Accumulated losses* | Total | |---|---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 |

At 31 December 2021 (unaudited) | – | – | – | (3,814) | (3,814) Loss for the year | – | – | – | (73,728) | (73,728) Other comprehensive income for the year: Exchange differences on translation of foreign operations | – | – | 99 | – | 99 Total comprehensive loss for the year | – | – | 99 | (73,728) | (73,629) Recognition of share-based payment expenses | – | 1,069 | – | – | 1,069 At 31 December 2022 | – | 1,069 | 99 | (77,542) | (76,374)

Attributable to owners of the parent

| | Share capital | Share option reserve* | Exchange fluctuation reserve* | Accumulated losses* | Total | |---|---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 |

At 31 December 2022 | – | 1,069 | 99 | (77,542) | (76,374) Loss for the year | – | – | – | (269,246) | (269,246) Other comprehensive income for the year: Exchange differences on translation of foreign operations | – | – | 360 | – | 360 Total comprehensive loss for the year | – | – | 360 | (269,246) | (268,886) Recognition of share-based payment expenses | – | 3,346 | – | – | 3,346 At 31 December 2023 | – | 4,415 | 459 | (346,788) | (341,914)

– I-9 –

APPENDIX I

ACCOUNTANT'S REPORT

Attributable to owners of the parent

| | Share capital | Share option reserve* | Fair value reserve of financial assets at fair value through other comprehensive income* | Exchange fluctuation reserve* | Accumulated losses* | Total | |---|---|---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 |

At 31 December 2023 | – | 4,415 | – | 459 | (346,788) | (341,914) Loss for the year | – | – | – | – | (465,238) | (465,238) Other comprehensive income for the year: Change in fair value of equity investments at fair value through other comprehensive income, net of tax | – | – | 662 | – | – | 662 Exchange differences on translation of foreign operations | – | – | – | 347 | – | 347 Total comprehensive loss for the year | – | – | 662 | 347 | (465,238) | (464,229) Recognition of share-based payment expenses | – | 6,823 | – | – | – | 6,823 At 31 December 2024 | – | 11,238 | 662 | 806 | (812,026) | (799,320)

– I-10 –

APPENDIX I

ACCOUNTANT'S REPORT

Attributable to owners of the parent

| | Share capital | Share option reserve | Fair value reserve of financial assets at fair value through other comprehensive income | Exchange fluctuation reserve | Accumulated losses | Total | |---|---|---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 |

At 31 December 2023 | – | 4,415 | – | 459 | (346,788) | (341,914) Loss for the period (unaudited) | – | – | – | – | (304,342) | (304,342) Other comprehensive income for the period: Change in fair value of equity investments at fair value through other comprehensive income, net of tax (unaudited) | – | – | (839) | – | – | (839) Exchange differences on translation of foreign operations (unaudited) | – | – | – | (86) | – | (86) Total comprehensive loss for the period (unaudited) | – | – | (839) | (86) | (304,342) | (305,267) Recognition of share-based payment expenses (unaudited) | – | 6,100 | – | – | – | 6,100 At 30 September 2024 (unaudited) | – | 10,515 | (839) | 373 | (651,130) | (641,081)

– I-11 –

APPENDIX I

ACCOUNTANT'S REPORT

Attributable to owners of the parent

| | Share capital | Share option reserve* | Fair value reserve of financial assets at fair value through other comprehensive income* | Exchange fluctuation reserve* | Accumulated losses* | Total | |---|---|---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 |

At 31 December 2024 | – | 11,238 | 662 | 806 | (812,026) | (799,320) Loss for the period | – | – | – | – | (512,013) | (512,013) Other comprehensive income for the period: Change in fair value of equity investments at fair value through other comprehensive income, net of tax | – | – | 1,604 | – | – | 1,604 Exchange differences on translation of foreign operations | – | – | – | (1,255) | – | (1,255) Total comprehensive loss for the period | – | – | 1,604 | (1,255) | (512,013) | (511,664) Recognition of share-based payment expenses | – | 8,581 | – | – | – | 8,581 Deemed distribution | – | – | – | – | (1,096) | (1,096) At 30 September 2025 | – | 19,819 | 2,266 | (449) | (1,325,135) | (1,303,499)

* These deficits accounts comprise the consolidated deficits of USD76,374,000, USD341,914,000, USD799,320,000 and USD1,303,499,000 in the consolidated statements of financial position as at 31 December 2022, 2023 and 2024 and 30 September 2025, respectively.

– I-12 –

APPENDIX I

ACCOUNTANT'S REPORT

CONSOLIDATED STATEMENTS OF CASH FLOWS

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | Notes | | | | |

CASH FLOWS FROM OPERATING ACTIVITIES Loss before tax Adjustments for: Finance costs Interest income Fair value gain on financial assets at fair value through profit or loss Fair value loss on financial liabilities (Gains)/losses on disposal of right-of-use assets Depreciation of property, plant and equipment Depreciation of right-of-use assets Share-based payment expense Provision for impairment on financial assets

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | 6 | 14 | 61 | 509 | 316 | 511 | | 5 | (39) | (7,785) | (20,448) | (17,199) | (7,876) | | 5 | (941) | (788) | (15,710) | (6,682) | (20,414) | | 7 | 60,509 | 176,826 | 214,172 | 128,063 | 313,477 | | | – | (70) | 1 | – | (175) | | 13 | 25 | 180 | 451 | 325 | 582 | | 14 | 182 | 631 | 1,450 | 1,072 | 1,478 | | 26 | 1,069 | 3,346 | 6,823 | 6,100 | 8,581 | | 15 | – | 3 | 88 | 68 | 22 | | | (12,909) | (96,842) | (277,902) | (192,279) | (215,827) | | | – | (1,341) | (5,732) | (4,230) | (1,103) | | | (513) | (4,190) | (9,272) | (11,925) | 1,846 | | | 2,394 | 14,848 | 33,970 | 37,949 | 19,007 | | | 2,191 | 12,624 | 21,048 | 4,557 | (22,865) |

Increase in trade receivables (Increase)/decrease in prepayments, other receivables and other assets Increase in trade and bills payables Increase/(decrease) in other payables, accruals and other liabilities Increase in other non-current liabilities Increase in contract liabilities (Increase)/decrease in restricted cash

| | – | 1,218 | – | – | 267 | | | – | 559 | 994 | 481 | 3,104 | | | (2,221) | 2,182 | (27,292) | (35,347) | 2,193 |

Cash flows used in operating activities Interest received

| | (11,058) | (70,942) | (264,186) | (200,794) | (213,378) | | | 39 | 6,487 | 5,703 | 5,198 | 3,982 |

Net cash flows used in operating activities

| | (11,019) | (64,455) | (258,483) | (195,596) | (209,396) |

– I-13 –

APPENDIX I

ACCOUNTANT'S REPORT

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 |

CASH FLOWS FROM INVESTING ACTIVITIES Purchases of items of property, plant and equipment Placement of time deposits Maturity of time deposits Proceeds from disposal of financial assets at amortised cost Purchase of financial assets at amortised cost Purchases of financial assets at fair value through other comprehensive income Proceeds from disposal of financial assets at fair value through profit or loss Purchases of financial assets at fair value through profit or loss Net cash flows used in investing activities

| | (256) | (697) | (759) | (496) | (479) | | | – | (90,400) | (199,100) | (195,200) | – | | | – | – | 271,201 | 267,036 | 26,513 | | | – | – | 982,359 | 862,084 | 2,531,476 | | | – | – | – | – | (4,174) | | | (4,174) | – | 11,050 | 136,076 | 1,851,346 | | | 1,056,303 | 1,519,366 | | | | | | (45,950) | (85,299) | (2,210,385) | (1,582,273) | (1,822,783) | | | (35,156) | (40,320) | (431,300) | (630,463) | (126,231) |

CASH FLOWS FROM FINANCING ACTIVITIES Proceeds from issuance of convertible bonds Proceeds from issuance of convertible redeemable preferred shares New bank and other borrowings Repayment of bank and other borrowings Repayment of convertible bonds Interest paid for bank borrowings Principal portion of lease payments Interest paid for leases Payment of Listing expenses Others Net cash flows from financing activities

| | – | – | 13,910 | 13,910 | – | | | 50,000 | 307,000 | 739,588 | 686,372 | 426,262 | | | – | – | 19,455 | 19,455 | 44,565 | | | – | – | – | – | (44,918) | | | – | – | – | – | (14,668) | | | – | – | (355) | (199) | (404) | | | (200) | (696) | (1,352) | (1,097) | (1,364) | | | (14) | (61) | (154) | (117) | (107) | | | – | – | – | – | (357) | | | – | – | – | 503 | (1,096) | | | 49,786 | 306,243 | 771,092 | 718,827 | 407,913 |

– I-14 –

APPENDIX I

ACCOUNTANT'S REPORT

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 |

NET INCREASE/(DECREASE) IN CASH AND CASH EQUIVALENTS Cash and cash equivalents at beginning of year/period Effect of foreign exchange rate changes, net CASH AND CASH EQUIVALENTS AT END OF YEAR/PERIOD

| 18 | 3,611 | 201,468 | 81,309 | (107,232) | 72,286 | | | 994 | 4,691 | 206,295 | 206,295 | 288,912 | | | 86 | 136 | 1,308 | 500 | 1,449 | | | 4,691 | 206,295 | 288,912 | 99,563 | 362,647 |

– I-15 –

APPENDIX I

ACCOUNTANT'S REPORT

STATEMENTS OF FINANCIAL POSITION OF THE COMPANY

| | As at 31 December | | | As at 30 September | | |---|---|---|---|---| | Notes | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 |

NON-CURRENT ASSETS Investments in subsidiaries Financial assets at fair value through profit or loss

| 17 | 1,069 | 4,415 | 11,238 | 19,819 | | 17 | – | – | 95,331 | 70,228 | | | 1,069 | 4,415 | 106,569 | 90,047 | | 16 | 14,100 | 103,895 | 360,091 | 663,642 | | 17 | – | – | 147,444 | – | | 17 | 65,791 | 10,152 | 295,220 | 639,899 | | 18 | – | – | 11,802 | – | | 18 | – | 91,598 | 26,327 | – | | 18 | 1,784 | 191,634 | 235,209 | 250,712 | | | 81,675 | 397,279 | 1,076,093 | 1,554,253 | | | 145,175 | 629,001 | 1,581,949 | 2,321,193 | | | – | 330 | 71 | 1,319 |

Total current liabilities

| | 145,175 | 629,331 | 1,582,020 | 2,322,512 |

NET CURRENT LIABILITIES

| | (63,500) | (232,052) | (505,927) | (768,259) |

TOTAL ASSETS LESS CURRENT LIABILITIES

| | (62,431) | (227,637) | (399,358) | (678,212) |

NON-CURRENT LIABILITIES Total non-current liabilities

| | – | – | – | – |

Net liabilities ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

(62,431)

(227,637)

(399,358)

(678,212)

DEFICITS Share capital ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Deficits ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

– (62,431)

– (227,637)

– (399,358)

– (678,212)

(62,431)

(227,637)

(399,358)

(678,212)

Total non-current assets ў ў ў ў ў ў CURRENT ASSETS Prepayments, other receivables and other assets ў ў ў ў ў ў ў ў ў ў ў Financial assets at amortised cost ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Financial assets at fair value through profit or loss ў ў ў ў ў ў ў Restricted cash ў ў ў ў ў ў ў ў ў ў ў ў ў ў Time deposits ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Cash and cash equivalents ў ў ў ў ў Total current assets ў ў ў ў ў ў ў ў ў ў CURRENT LIABILITIES Convertible redeemable preferred shares ў ў ў ў ў ў ў ў ў ў ў Other payables, accruals and other liabilities ў ў ў ў ў ў ў ў ў ў ў ў

Total deficits ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

24

25

– I-16 –

APPENDIX I

ACCOUNTANT'S REPORT

II

NOTES TO THE HISTORICAL FINANCIAL INFORMATION

1.

CORPORATE AND GROUP INFORMATION

MINIMAX GROUP INC.（以下简称"本公司"）于2021年6月在开曼群岛注册成立为有限责任公司。本公司注册办事处地址为：Maples Corporate Services Limited, PO Box 309, Ugland House, Grand Cayman, KY1-1104, Cayman Islands。

于相关期间内，本公司及其附属公司（合称"本集团"）主要从事人工智能（"AI"）基础模型的研究与开发，以及基于开放应用程序编程接口（"API"）平台提供相关服务、其他人工智能（"AI"）相关服务及AI原生产品。

附属公司资料

于相关期间末及本招股说明书日期，本公司直接及间接持有其附属公司的权益，所有附属公司均为私人有限责任公司，主要附属公司的详情如下：

| 名称 | 注册成立／登记地点及日期、经营地点 | 已发行普通股／注册股本 | 本公司直接持有的股权百分比 | 本公司间接持有的股权百分比 | 主要业务 | |---|---|---|---|---|---| | SUBSUP PTE. LTD. (a) | 新加坡，2022年9月14日 | 新加坡元50,000 | – | 100 | 运营AI原生产品 | | 北京稀宇极智科技有限公司* ("Beijing Jizhi")（"北京稀宇極智科技有限公司"）(b) | 中国大陆，2021年11月18日 | 人民币139,995,700元 | – | 100 | 研发AI基础模型 | | 上海稀宇极智科技有限公司* ("Shanghai Jizhi")（"上海稀宇極智科技有限公司"）(c) | 中国大陆，2021年11月3日 | 人民币1,000,000,000元 | – | 100 | 研发AI基础模型 | | 上海稀宇科技有限公司* ("Shanghai MiniMax")（"上海稀宇科技有限公司"）(d)(e) | 中国大陆，2023年1月28日 | 人民币2,030,303元 | – | 100 | 运营开放平台及AI原生产品 | | NanoNoble PTE. LTD. (a) | 新加坡，2024年3月19日 | 新加坡元50,000 | – | 100 | 运营开放平台及AI原生产品 | | MiniMax HONGKONG Limited (f) | 香港，2021年7月23日 | 港币1元 | 100 | – | 投资控股 |

*

上述中国大陆公司的英文名称系管理层就相关公司中文名称进行翻译所作的最佳努力，该等公司并未注册英文名称。

(a)

上述实体就截至2022年、2023年及2024年12月31日止年度均未编制经审计财务报表，原因在于根据其注册司法管辖区的相关规则及法规，上述实体并不受任何法定审计规定的约束。

(b)

北京极智（Beijing Jizhi）依据中国法律注册为有限责任公司。截至2022年12月31日止年度按中国公认会计原则（"中国会计准则"）编制的法定财务报表由在中国注册的注册会计师事务所北京东财会计师事务所（普通合伙）审计。截至2023年及2024年12月31日止年度按中国会计准则编制的法定财务报表由在中国注册的注册会计师事务所上海旭日会计师事务所（普通合伙）审计。

— I-17 —

附录一

会计师报告

(c)

上海极智（Shanghai Jizhi）依据中国法律注册为有限责任公司。截至2022年、2023年及2024年12月31日止年度按中国会计准则编制的法定财务报表由在中国注册的注册会计师事务所上海旭日会计师事务所（普通合伙）审计。

(d)

上海MiniMax（Shanghai MiniMax）依据中国法律注册为有限责任公司。截至2023年及2024年12月31日止年度按中国会计准则编制的法定财务报表由在中国注册的注册会计师事务所上海旭日会计师事务所（普通合伙）审计。

(e)

本集团于相关期间内通过合约安排将上海MiniMax列为附属公司入账。2023年，上海极智（Shanghai Jizhi）与上海MiniMax订立了一系列合约安排，据此，本集团对上海MiniMax的财务及经营事项拥有实际控制权，并有权享有上海MiniMax产生的全部经济利益，因此，上海MiniMax已作为可变利益实体并入本集团。2025年6月，本公司终止了上述与上海MiniMax的合约安排。上海MiniMax通过收购100%股权成为本集团的全资附属公司。

(f)

MiniMax HONGKONG Limited依据香港法律注册为有限责任公司。截至2024年止年度按香港财务报告准则（私人主体）编制的法定财务报表由在香港注册的注册会计师Raymond Li & Co.审计。

2.

会计政策

2.1

编制基础

历史财务资料已在综合基础上编制。所有集团内部交易及结余于综合时已予以抵销。

2.2

编制依据

Effective for annual periods beginning on or after 1 January 2016 but the effective date is deferred indefinitely

– I-19 –

APPENDIX I

ACCOUNTANT'S REPORT

Apart from the above, the directors of the Company are of the opinion that the adoption of the above new and revised IFRS Accounting Standards will not result in significant changes to the Group's existing accounting policies and are therefore unlikely to have a significant impact on the Group's results and financial position. 2.4

Summary of Material Accounting Policies

Business combinations and goodwill

Business combinations are accounted for using the acquisition method. The consideration transferred is measured at the acquisition date fair value which is the sum of the acquisition date fair values of assets transferred by the Group, liabilities assumed by the Group to the former owners of the acquiree and the equity interests issued by the Group in exchange for control of the acquiree. For each business combination, the Group elects whether to measure the non-controlling interests in the acquiree that are present ownership interests and entitle their holders to a proportionate share of net assets in the event of liquidation either at fair value or at the proportionate share of the acquiree's identifiable net assets. All other components of non-controlling interests are measured at their acquisition date fair value. Acquisition-related costs are expensed as incurred.

When the Group acquires a business, it assesses the financial assets and liabilities assumed for appropriate classification and designation in accordance with the contractual terms, economic circumstances and pertinent conditions as at the acquisition date. This includes the separation of embedded derivatives in host contracts of the acquiree.

If the initial accounting for a business combination is incomplete by the end of the reporting period in which the combination occurs, the Group reports provisional amounts for the items for which the accounting is incomplete. Those provisional amounts are adjusted during the measurement period (which cannot exceed one year from the acquisition date), or additional assets or liabilities are recognised, to reflect new information obtained about facts and circumstances that existed at the acquisition date that, if known, would have affected the amounts recognised at that date.

Goodwill is initially measured at cost, being the excess of the aggregate of the consideration transferred, the amount recognised for non-controlling interests and any fair value of the Group's previously held equity interests in the acquiree over the identifiable net assets acquired and liabilities assumed. If the sum of this consideration and other items is lower than the fair value of the net assets acquired, the difference is recognised as a gain on bargain purchase in profit or loss.

After initial recognition, goodwill is measured at cost less any accumulated impairment losses. Goodwill is tested for impairment annually or more frequently if events or changes in circumstances indicate that the carrying value may be impaired. The Group performs its annual impairment test of goodwill as at 31 December. For the purpose of impairment testing, goodwill acquired in a business combination is, from the acquisition date, allocated to each of the Group's cash-generating units that are expected to benefit from the combination, irrespective of whether other assets or liabilities of the Group are assigned to those units.

Impairment is determined by assessing the recoverable amount of the cash-generating unit (group of cash-generating units) to which the goodwill relates. Where the recoverable amount of the cash-generating unit (group of cash-generating units) is less than the carrying amount, an impairment loss is recognised. An impairment loss recognised for goodwill is not reversed in a subsequent period.

Where goodwill has been allocated to a cash-generating unit and part of the operation within that unit is disposed of, the goodwill associated with the disposed operation is included in the carrying amount of the operation when determining the gain or loss on the disposal. Goodwill disposed in these circumstances is measured based on the relative values of the disposed operation and the portion of the cash-generating unit retained.

No mandatory effective date yet determined but available for adoption

The Group is in the process of making an assessment of the impact of these new and amended standards upon initial application. IFRS 18 introduces new requirements for presentation within the statement of profit or loss, including specified totals and subtotals. Entities are required to classify all income and expenses within the statement of profit or loss into one of the five categories: operating, investing, financing, income taxes and discontinued operations and to present two new defined subtotals. It also requires disclosure of management-defined performance measures in a note and introduces new requirements for aggregation and disaggregation of financial information. The

– I-19 –

APPENDIX I

ACCOUNTANT'S REPORT

new requirements are expected to impact the Group's presentation of the statement of profit or loss and disclosures of the Group's financial performance. Except for IFRS 18, the directors of the Company anticipate that the application of these new and revised IFRS Accounting Standards will have no material impact on the Group's financial performance and financial position in the foreseeable future.

2.4

Material Accounting Policies

Fair value measurement

The Group measures its financial assets at fair value through profit or loss, financial assets at fair value through other comprehensive income, convertible redeemable preferred shares and convertible bonds at the end of each of the Relevant Periods. Fair value is the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date. The fair value measurement is based on the presumption that the transaction to sell the asset or transfer the liability takes place either in the principal market for the asset or liability, or in the absence of a principal market, in the most advantageous market for the asset or liability. The principal or the most advantageous market must be accessible by the Group. The fair value of an asset or a liability is measured using the assumptions that market participants would use when pricing the asset or liability, assuming that market participants act in their economic best interest.

A fair value measurement of a non-financial asset takes into account a market participant's ability to generate economic benefits by using the asset in its highest and best use or by selling it to another market participant that would use the asset in its highest and best use.

The Group uses valuation techniques that are appropriate in the circumstances and for which sufficient data are available to measure fair value, maximising the use of relevant observable inputs and minimising the use of unobservable inputs.

All assets and liabilities for which fair value is measured or disclosed in the Historical Financial Information are categorised within the fair value hierarchy, described as follows, based on the lowest level input that is significant to the fair value measurement as a whole:

Level 1 – based on quoted prices (unadjusted) in active markets for identical assets or liabilities

Level 2 – based on valuation techniques for which the lowest level input that is significant to the fair value measurement is observable, either directly or indirectly

Level 3 – based on valuation techniques for which the lowest level input that is significant to the fair value measurement is unobservable

For assets and liabilities that are recognised in the Historical Financial Information on a recurring basis, the Group determines whether transfers have occurred between levels in the hierarchy by reassessing categorisation (based on the lowest level input that is significant to the fair value measurement as a whole) at the end of each of the Relevant Periods.

Impairment of non-financial assets

Where an indication of impairment exists, or when annual impairment testing for an asset is required, the asset's recoverable amount is estimated. An asset's recoverable amount is the higher of the asset's or cash-generating unit's value in use and its fair value less costs of disposal, and is determined for an individual asset, unless the asset does not generate cash inflows that are largely independent of those from other assets or groups of assets, in which case the recoverable amount is determined for the cash-generating unit to which the asset belongs.

An impairment loss is recognised only if the carrying amount of an asset exceeds its recoverable amount. In assessing value in use, the estimated future cash flows are discounted to their present value using a pre-tax discount rate that reflects current market assessments of the time value of money and the risks specific to the asset. An impairment loss is charged to the statement of profit or loss in the period in which it arises in those expense categories consistent with the function of the impaired asset.

– I-20 –

APPENDIX I

ACCOUNTANT'S REPORT

Right-of-use assets are recognised at the commencement date of the lease (i.e., the date the underlying asset is available for use). Right-of-use assets are measured at cost, less any accumulated depreciation and impairment losses, and adjusted for any remeasurement of lease liabilities. The cost of right-of-use assets includes the amount of lease liabilities recognised, initial direct costs incurred, and lease payments made at or before the commencement date less any lease incentives received. Right-of-use assets are depreciated on a straight-line basis over the shorter of the lease term and the estimated useful lives of the assets.

If ownership of the leased asset transfers to the Group by the end of the lease term or the cost reflects the exercise of a purchase option, depreciation is calculated using the estimated useful life of the asset.

(b) Lease liabilities

Lease liabilities are recognised at the commencement date of the lease at the present value of lease payments to be made over the lease term. The lease payments include fixed payments (including in-substance fixed payments) less any lease incentives receivable, variable lease payments that depend on an index or a rate, and amounts expected to be paid under residual value guarantees. The lease payments also include the exercise price of a purchase option reasonably certain to be exercised by the Group and payments of penalties for terminating a lease, if the lease term reflects the Group exercising the option to terminate. Variable lease payments that do not depend on an index or a rate are recognised as an expense in the period in which the event or condition that triggers the payment occurs.

In calculating the present value of lease payments, the Group uses the incremental borrowing rate at the lease commencement date if the interest rate implicit in the lease is not readily determinable. After the commencement date, the amount of lease liabilities is increased to reflect the accretion of interest and reduced for the lease payments made. In addition, the carrying amount of lease liabilities is remeasured if there is a modification, a change in the lease term, a change in the in-substance fixed lease payments or a change in the assessment to purchase the underlying asset.

(c) Short-term leases

The Group applies the short-term lease recognition exemption to its short-term leases of machinery and equipment (i.e., those leases that have a lease term of 12 months or less from the commencement date and do not contain a purchase option). Lease payments on short-term leases are recognised as an expense on a straight-line basis over the lease term.

— I-21 —

APPENDIX I

ACCOUNTANT'S REPORT

**Investments and other financial assets**

**Initial recognition and measurement**

Financial assets are classified, at initial recognition, as subsequently measured at amortised cost, fair value through other comprehensive income, and fair value through profit or loss.

The classification of financial assets at initial recognition depends on the financial asset's contractual cash flow characteristics and the Group's business model for managing them. With the exception of trade receivables that do not contain a significant financing component or for which the Group has applied the practical expedient of not adjusting the effect of a significant financing component, the Group initially measures a financial asset at its fair value plus, in the case of a financial asset not at fair value through profit or loss, transaction costs. Trade receivables that do not contain a significant financing component or for which the Group has applied the practical expedient are measured at the transaction price determined under HKFRS 15 on initial recognition.

In order for a financial asset to be classified and measured at amortised cost or fair value through other comprehensive income, it needs to give rise to cash flows that are solely payments of principal and interest (SPPI) on the principal amount outstanding. This assessment is referred to as the SPPI test and is performed at an instrument level.

The Group's business model for managing financial assets refers to how it manages its financial assets in order to generate cash flows. The business model determines whether cash flows will result from collecting contractual cash flows, selling the financial assets, or both.

Purchases or sales of financial assets that require delivery of assets within a time frame established by regulation or convention in the market place (regular way trades) are recognised on the trade date, i.e., the date that the Group commits to purchase or sell the asset.

**Subsequent measurement**

For purposes of subsequent measurement, financial assets are classified in four categories:

- Financial assets at amortised cost (debt instruments) - Financial assets at fair value through other comprehensive income with recycling of cumulative gains and losses (debt instruments) - Financial assets designated at fair value through other comprehensive income with no recycling of cumulative gains and losses upon derecognition (equity instruments) - Financial assets at fair value through profit or loss

**Financial assets at amortised cost (debt instruments)**

The Group measures financial assets at amortised cost if both of the following conditions are met:

- The financial asset is held within a business model with the objective to hold financial assets in order to collect contractual cash flows; and - The contractual terms of the financial asset give rise on specified dates to cash flows that are solely payments of principal and interest on the principal amount outstanding.

Financial assets at amortised cost are subsequently measured using the effective interest method and are subject to impairment. Gains and losses are recognised in profit or loss when the asset is derecognised, modified or impaired.

APPENDIX I

ACCOUNTANT'S REPORT

使用权资产于租赁开始日（即相关资产可供使用之日）予以确认。使用权资产按成本减累计折旧及任何减值损失计量，并就租赁负债的任何重新计量进行调整。使用权资产的成本包括已确认租赁负债的金额、已发生的初始直接费用，以及于开始日当日或之前支付的租赁款项减去已收取的任何租赁激励。

使用权资产按以下租赁期与资产估计使用寿命两者中较短者采用直线法计提折旧：

建筑物 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

2至3年

若于租赁期届满时租赁资产的所有权转移至本集团，或成本反映了购买选择权的行使，则折旧按资产的估计使用寿命计算。

– I-22 –

附录一 (b)

会计师报告

租赁负债

租赁负债于租赁开始日按租赁期内应付租赁款项的现值确认。租赁款项包括固定款项（包括实质上的固定款项）减去任何应收租赁激励、取决于指数或利率的可变租赁款项，以及预期根据残值担保支付的金额。租赁款项亦包括本集团合理确定将行使的购买选择权的行权价格，以及终止租赁的罚款款项（如租赁期反映本集团行使终止租赁的选择权）。不取决于指数或利率的可变租赁款项，于触发该款项的事件或条件发生的期间确认为费用。

在计算租赁款项的现值时，由于租赁隐含利率不易确定，本集团采用租赁开始日的增量借款利率。开始日后，租赁负债的金额随利息的累积而增加，并随已支付的租赁款项而减少。此外，若发生变更、租赁期变更、租赁款项变更（例如，因指数或利率变化导致未来租赁款项变化）或对购买相关资产选择权的评估发生变化，则租赁负债的账面金额须重新计量。

(c)

短期租赁及低价值资产租赁

本集团对机器及设备的短期租赁（即自开始日起租赁期为12个月或以下且不含购买选择权的租赁）采用短期租赁确认豁免。短期租赁的租赁款项在租赁期内按直线法确认为费用。

投资及其他金融资产

初始确认及计量

金融资产于初始确认时分类为其后按摊销成本计量、按公允价值计量且其变动计入其他综合收益，以及按公允价值计量且其变动计入损益。

金融资产于初始确认时的分类取决于金融资产的合同现金流量特征及本集团管理金融资产的业务模式。除不含重大融资成分的应收账款，或本集团已采用不调整重大融资成分影响的实际权宜方法的应收账款外，本集团最初按公允价值计量金融资产，对于非以公允价值计量且其变动计入损益的金融资产，则加上交易费用。不含重大融资成分的应收账款，或本集团已采用实际权宜方法的应收账款，按照下文"收入确认"政策所述，依据《国际财务报告准则第15号》确定的交易价格计量。

金融资产须产生仅为对未偿还本金金额支付本金和利息（"SPPI"）的现金流量，方可分类及计量为摊销成本或以公允价值计量且其变动计入其他综合收益。现金流量不符合SPPI条件的金融资产，无论业务模式如何，均分类及计量为以公允价值计量且其变动计入损益。

本集团管理金融资产的业务模式，是指其管理金融资产以产生现金流量的方式。业务模式决定现金流量是来源于收取合同现金流量、出售金融资产，抑或两者兼而有之。分类及计量为摊销成本的金融资产，持有于以收取合同现金流量为目标的业务模式下；而分类及计量为以公允价值计量且其变动计入其他综合收益的金融资产，则持有于以收取合同现金流量及出售为双重目标的业务模式下。不持有于上述业务模式下的金融资产，分类及计量为以公允价值计量且其变动计入损益。

须在市场通常由法规或惯例规定的期限内交割资产的金融资产买卖，于交易日（即本集团承诺买入或卖出该资产之日）予以确认。

– I-23 –

附录一

会计师报告

Financial assets for which credit risk has not increased significantly since initial recognition or that have low credit risk at the reporting date. ECLs are measured at an amount equal to the 12-month ECL.

Stage 2

–

Financial assets for which credit risk has increased significantly since initial recognition (unless they have low credit risk at the reporting date) but that do not have objective evidence of impairment. ECLs are measured at an amount equal to the lifetime ECL.

Stage 3

–

Financial assets that have objective evidence of impairment at the reporting date. ECLs are measured at an amount equal to the lifetime ECL.

Simplified approach

For trade receivables and contract assets that do not contain a significant financing component or when the Group applies the practical expedient of not adjusting the effect of a significant financing component, the Group applies the simplified approach in calculating ECLs. Under the simplified approach, the Group does not track changes in credit risk, but instead recognises a loss allowance based on lifetime ECLs at each reporting date. The Group has established a provision matrix that is based on its historical credit loss experience, adjusted for forward-looking factors specific to the debtors and the economic environment.

For trade receivables and contract assets that contain a significant financing component, the Group chooses as its accounting policy to adopt the simplified approach in calculating ECLs with policies as described above.

后续计量

金融资产的后续计量取决于其分类，具体如下：

以摊余成本计量的金融资产（债务工具）

以摊余成本计量的金融资产随后使用实际利率法计量，并须进行减值测试。当资产被终止确认、修改或减值时，相关损益在利润表中确认。

指定以公允价值计量且其变动计入其他综合收益的金融资产（权益投资）

初始确认时，当权益投资符合《国际会计准则第32号——金融工具：列报》中关于权益的定义且并非以交易为目的持有时，本集团可以不可撤销地选择将其分类为指定以公允价值计量且其变动计入其他综合收益的权益投资。该分类按逐项工具确定。

此类金融资产的损益永远不会重新分类至利润表。当收款权利已确立时，股息作为其他收入在利润表中确认，但本集团从此类收益中受益作为收回金融资产部分成本的情况除外，在此情况下，相关收益记录于其他综合收益。指定以公允价值计量且其变动计入其他综合收益的权益投资不须进行减值评估。

以公允价值计量且其变动计入损益的金融资产

以公允价值计量且其变动计入损益的金融资产在财务状况表中以公允价值列示，公允价值的净变动在利润表中确认。

金融资产的终止确认

金融资产（或在适用情况下，金融资产的一部分或一组类似金融资产的一部分）主要在以下情况下终止确认（即从本集团合并财务状况表中移除）：

• 从该资产收取现金流量的权利已到期；或

• 本集团已转让从该资产收取现金流量的权利，或已承担在"过手"安排下向第三方及时足额支付所收取现金流量的义务；且（a）本集团已转让该资产绝大部分的风险和报酬，或（b）本集团既未转让也未保留该资产绝大部分的风险和报酬，但已转让该资产的控制权。

当本集团已转让从资产收取现金流量的权利或已签订过手安排时，本集团评估其是否以及在多大程度上保留了该资产所有权的风险和报酬。当本集团既未转让也未保留该资产绝大部分的风险和报酬，且未转让该资产的控制权时，本集团继续在本集团持续涉入的范围内确认已转让资产。在此情况下，本集团亦确认相关负债。已转让资产和相关负债按反映本集团所保留的权利和义务的基础进行计量。

以对已转让资产的担保形式存在的持续涉入，按该资产原账面金额与本集团可能须偿还的最高对价金额两者中的较低者计量。

金融资产减值

本集团对所有非以公允价值计量且其变动计入损益的债务工具确认预期信用损失（"预期信用损失"）拨备。预期信用损失基于合同规定的合同现金流量与本集团预期收到的所有现金流量之间的差额，按原始实际利率的近似值折现。预期现金流量将包括出售所持抵押品或其他作为合同条款组成部分的信用增级所产生的现金流量。

– I-24 –

附录一

会计师报告

一般方法

预期信用损失分两个阶段确认。对于自初始确认以来信用风险未显著增加的信用风险敞口，按未来12个月内可能发生的违约事件所导致的信用损失计提预期信用损失拨备（12个月预期信用损失）。对于自初始确认以来信用风险已显著增加的信用风险敞口，无论违约时间如何，均须对该敞口剩余存续期内预期的信用损失计提损失拨备（整个存续期预期信用损失）。

在每个报告日，本集团评估金融工具的信用风险自初始确认以来是否已显著增加。在进行评估时，本集团将该金融工具在报告日发生违约的风险与初始确认日发生违约的风险进行比较，并考虑在不付出过多成本或努力的情况下可获得的合理且有据可查的信息，包括历史及前瞻性信息。本集团认为，当合同款项逾期超过90天时，信用风险已显著增加。

本集团认为，当合同款项逾期超过一年时，金融资产已发生违约。然而，在某些情况下，当内部或外部信息表明本集团在考虑本集团所持有的任何信用增级之前不太可能全额收回未偿合同金额时，本集团亦可能认为金融资产已发生违约。

以公允价值计量且其变动计入其他综合收益的债务投资及以摊余成本计量的金融资产须按一般方法进行减值，并按以下阶段分类以计量预期信用损失，但适用以下详述简化方法的应收账款除外。

第一阶段 – 自初始确认以来信用风险未显著增加，或在报告日具有低信用风险的金融资产。预期信用损失按等于12个月预期信用损失的金额计量。

第二阶段 – 自初始确认以来信用风险已显著增加（除非在报告日具有低信用风险）但尚无客观减值证据的金融资产。预期信用损失按等于整个存续期预期信用损失的金额计量。

第三阶段 – 在报告日具有客观减值证据的金融资产。预期信用损失按等于整个存续期预期信用损失的金额计量。

简化方法

对于不含重大融资成分的应收账款和合同资产，或当本集团采用不调整重大融资成分影响的实务权宜之计时，本集团在计算预期信用损失时采用简化方法。在简化方法下，本集团不追踪信用风险的变化，而是在每个报告日根据整个存续期预期信用损失确认损失拨备。本集团已建立基于其历史信用损失经验的拨备矩阵，并针对债务人及经济环境的特定前瞻性因素进行调整。

对于含有重大融资成分的应收账款和合同资产，本集团选择采用上述简化方法计算预期信用损失作为其会计政策。

Stage 1

–

Financial instruments for which credit risk has not increased significantly since initial recognition and for which the loss allowance is measured at an amount equal to 12-month ECLs

Stage 2

–

Financial instruments for which credit risk has increased significantly since initial recognition but that are not credit-impaired financial assets and for which the loss allowance is measured at an amount equal to lifetime ECLs

Stage 3

–

Financial assets that are credit-impaired at the reporting date (but that are not purchased or originated credit-impaired) and for which the loss allowance is measured at an amount equal to lifetime ECLs

简化方法（Simplified approach）

对于不含重大融资成分的贸易应收款项，或当本集团采用不调整重大融资成分影响的实务权宜措施时，本集团采用简化方法计算预期信用损失。在简化方法下，本集团不追踪信用风险的变化，而是在每个报告日期根据整个存续期的预期信用损失确认损失准备。本集团已根据其历史信用损失经验建立了一个备抵矩阵，并针对债务人及经济环境的前瞻性因素进行了调整。

金融负债

初始确认与计量

金融负债在初始确认时，视情况分类为以公允价值计量且其变动计入损益的金融负债、借款，或应付款项。

所有金融负债均初始以公允价值计量，对于借款及应付款项，则扣除可直接归属的交易成本后列示。

本集团的金融负债包括贸易及票据应付款项、其他应付款项、应计费用及其他负债、可转换可赎回优先股、计息银行借款及租赁负债。

– I-25 –

附录一

会计师报告

后续计量

金融负债的后续计量取决于其分类，具体如下：

以摊余成本计量的金融负债（贸易及票据应付款项、其他应付款项、应计费用及其他负债（不包括可转换债券）、计息银行借款及租赁负债）

初始确认后，其他应付款项及应计费用以及租赁负债随后以摊余成本计量，采用实际利率法；若折现影响不重大，则按成本列示。负债终止确认时以及通过实际利率摊销过程所产生的利得和损失在损益表中确认。

摊余成本的计算须考虑收购时的折扣或溢价，以及构成实际利率组成部分的费用或成本。实际利率摊销计入损益表中的财务费用。

以公允价值计量且其变动计入损益的金融负债（可转换可赎回优先股及可转换债券）

以公允价值计量且其变动计入损益的金融负债，包括在初始确认时指定为以公允价值计量且其变动计入损益的金融负债。本公司发行的可转换可赎回优先股及可转换债券于初始确认时被指定为以公允价值计量且其变动计入损益。其初始以公允价值确认，任何可直接归属的交易成本在损益中作为财务费用确认。其产生的利得或损失在损益表中确认，但因本公司自身信用风险产生的利得或损失除外，该部分在其他综合收益中列示，且不得重新分类至损益表。在损益表中确认的公允价值变动净利得或净损失不包括就该等金融负债收取的任何利息。

金融负债的终止确认

当金融负债项下的义务履行、撤销或到期时，该金融负债予以终止确认。

当现有金融负债被同一贷款方以实质上不同条款的另一金融负债所替代，或现有负债的条款经实质性修改时，此类交换或修改作为原负债的终止确认及新负债的确认处理，相关账面金额之间的差额在损益表中确认。

现金及现金等价物

财务状况表中的现金及现金等价物包括手头现金及银行存款，以及一般到期日在三个月以内、可随时转换为已知金额现金、价值变动风险甚微且为满足短期现金承担而持有的短期高流动性存款。

就合并现金流量表而言，现金及现金等价物包括如上所定义的手头现金及银行存款及短期存款，扣除可按要求偿还且构成本集团现金管理组成部分的银行透支。

拨备

当过去事项导致现时义务（法定或推定）产生，且很可能需要流出资源以履行该义务，并能对义务金额作出可靠估计时，确认一项拨备。

当折现影响重大时，就拨备所确认的金额为报告期末预期履行义务所需未来支出的现值。因时间推移而产生的折现现值增加额计入损益表中的财务费用。

– I-26 –

附录一

会计师报告

Sale of goods

Revenue from the sale of goods is recognised at the point in time when control of the goods is transferred to the customer, generally when the goods are delivered to the customer. Revenue is measured at the transaction price agreed under the contract.

When the Group sells goods to a customer with a right of return, revenue is recognised to the extent that it is highly probable that a significant reversal in the amount of cumulative revenue recognised will not occur. Therefore, the amount of revenue recognised is adjusted for expected returns, which are estimated based on the historical data. A refund liability (included in other payables) is recognised for the goods that are expected to be returned. At the same time, the Group has a right to recover the goods when customers exercise their right of return so an asset (included in other receivables) and a corresponding adjustment to cost of sales is recognised for the right to recover the returned goods. The asset is measured at the former carrying amount of the goods less any expected costs to recover the goods, including any potential decreases in the value of the returned goods.

(b) Rendering of services

Revenue from rendering of services is recognised over time as the services are performed. Revenue is measured at the transaction price agreed under the contract.

(c) Loyalty programme

The Group operates a customer loyalty programme under which customers accumulate points for each dollar spent. Points are redeemable for free goods or discounts on future purchases. A performance obligation exists with respect to the points awarded since they provide a right to customers that they would not receive without entering into the contract. The transaction price is allocated between the goods sold and the points awarded with reference to their relative stand-alone selling prices.

Revenue allocated to the points is deferred and recognised when the points are redeemed or when they expire.

Contract liabilities

A contract liability is the obligation to transfer goods or services to a customer for which the Group has received consideration (or an amount of consideration is due) from the customer. If a customer pays consideration before the Group transfers goods or services to the customer, a contract liability is recognised when the payment is made or the payment is due (whichever is earlier). Contract liabilities are recognised as revenue when the Group performs under the contract.

Other income

Interest income is recognised on an accrual basis using the effective interest method by applying the rate that exactly discounts the estimated future cash receipts through the expected life of the financial instrument or a shorter period, when appropriate, to the net carrying amount of the financial asset.

Dividend income is recognised when the shareholders' right to receive payment has been established, it is probable that the economic benefits associated with the dividend will flow to the Group and the amount of the dividend can be measured reliably.

– I-27 –

APPENDIX I ACCOUNTANT'S REPORT

政府补助 Government Grants

政府补助在有合理保证能收到补助款项且集团能遵守所有相关条件的情况下，按公允价值予以确认。当补助与某项费用项目相关时，从相关费用中扣除，并在与该补助具体相关的费用所在的同一期间予以确认。政府补助应确认为递延收入，并在其后确认相关成本或损失时按上述方式予以确认。

Government grants are recognised at their fair value where there is reasonable assurance that the grant will be received and all attaching conditions will be complied with. When the grant relates to an expense item, it is deducted from the related expense and recognised in the same period as the expenses specifically relevant to the grants. The government grants shall be recognised as deferred income and recognised as described above when the relevant costs or losses are recognised subsequently.

当补助与某项资产相关时，其公允价值计入递延收入账户，并在相关资产的预计使用年限内以等额年度分期转入损益表，或从资产账面价值中扣除，并通过减少折旧费用的方式转入损益表。

Where the grant relates to an asset, the fair value is credited to a deferred income account and is released to the statement of profit or loss over the expected useful life of the relevant asset by equal annual instalments or deducted from the carrying amount of the asset and released to the statement of profit or loss by way of a reduced depreciation charge.

收入确认 Revenue Recognition

来自客户合同的收入 Revenue from contracts with customers

来自客户合同的收入在商品或服务的控制权转移给客户时予以确认，金额反映集团预期因转让相关商品或服务而有权收取的对价。

Revenue from contracts with customers is recognised when control of goods or services is transferred to the customers at an amount that reflects the consideration to which the Group expects to be entitled in exchange for those goods or services.

(a) 销售商品

来自AI原生产品的收入 会员订阅 本集团向个人用户提供会员订阅服务，订阅会员可获得本集团AI原生产品高级功能的访问权限。会员订阅费须预先支付，且不予退还。收入按照服务提供情况，在会员期间内按比例确认。

虚拟道具 本集团亦向个人用户提供其AI原生产品中的虚拟道具，以提升用户体验。用户可选择预先购买额外积分以充值账户，并购买上述虚拟道具。对于消耗型虚拟道具，本集团的履约义务为向用户提供一次性服务。该履约义务于虚拟道具被消耗时即告履行完毕。据此，本集团在某一时间点确认相关收入。对于非消耗型虚拟道具，本集团的履约义务为向购买虚拟道具的用户持续提供服务。该履约义务在付费用户的使用期间内持续履行。据此，本集团在上述付费用户的估计平均使用期间内按比例确认收入。

在线营销服务 此外，本集团在其部分AI原生应用（包括通过中介平台）上向企业客户提供基于效果的在线营销服务。来自在线营销服务的收入主要在用户查看或点击广告时于某一时间点予以确认。

（b）

来自开放平台及其他基于AI的企业服务的收入

本集团通过其开放平台向企业客户提供访问其核心AI模型的服务。上述服务的履约义务在客户以令牌（tokens）调用API时于某一时间点即告履行完毕。每月月末，对价根据已消耗的令牌数量固定确定，不存在可变对价。

本集团亦向企业客户提供其他基于AI的企业服务。上述服务的对价为固定金额，来自其他基于AI的企业服务的收入通常在客户验收服务时于某一时间点予以确认。

– I-28 –

附录一

会计师报告

其他收入 利息收入按权责发生制，采用实际利率法确认，即应用能够将预计未来现金收入在金融工具预期存续期（如适当，可采用较短期间）内折现至金融资产账面净值的利率。

合同负债 当本集团在转让相关商品或服务之前收到客户款项或客户款项到期应付（以较早者为准）时，确认合同负债。当本集团履行合同义务（即将相关商品或服务的控制权转让给客户）时，合同负债转为收入予以确认。

以股份为基础的支付 本公司设有股份期权计划。本集团员工（包括董事）以股份为基础的支付形式获得报酬，即员工以提供服务换取权益工具（"以权益结算的交易"）。以权益结算的与员工之间的交易成本，参照授予日的公允价值计量。公允价值由外部估值师采用二叉树模型确定，详情载于历史财务资料附注26。

以权益结算的交易成本在业绩及╱或服务条件的履行期间内确认为员工福利费用，并相应增加权益。于相关期间届满前各报告期末，就以权益结算的交易所确认的累计费用，反映归属期已届满的程度，以及本集团对最终归属的权益工具数量的最佳估计。该期间损益表中的借记或贷记金额，代表该期间期初至期末所确认的累计费用的变动额。

服务条件和非市场业绩条件在确定授予日奖励公允价值时不予考虑，但相关条件达成的可能性须作为本集团对最终归属的权益工具数量的最佳估计的组成部分加以评估。附于奖励但不附带相关服务要求的任何其他条件，均视为非归属条件。非归属条件在奖励公允价值中予以反映，并导致奖励立即计入费用，除非同时存在服务条件及╱或业绩条件。

对于因未满足非市场业绩条件及╱或服务条件而最终未能归属的奖励，不确认任何费用。

若以权益结算的奖励条款经修改，则至少应按照原始条款未经修改的情况确认费用（若原始条款已满足）。此外，对于增加以股份为基础的支付总公允价值的任何修改，或在修改日计量对员工有利的任何修改，均应额外确认费用。若以权益结算的奖励被取消，则视同其于取消日归属，尚未就该奖励确认的任何费用须立即予以确认。

其他员工福利 退休金计划 本集团在中国内地经营的附属公司的员工须参加由当地市政府运营的统筹退休金计划。本集团须按员工工资成本的一定比例向统筹退休金计划缴纳供款。上述供款按照统筹退休金计划的规定，于应付时计入损益表。

– I-29 –

附录一

会计师报告

住房公积金及其他社会保险 根据中国相关法律法规，本集团已为员工参加法定社会保险缴纳计划。该计划包括住房公积金、基本医疗保险、失业保险、工伤保险及生育保险。本集团按月缴纳住房公积金及其他社会保险。相关缴款按应计制计入损益。本集团就上述基金所承担的负债仅限于相关期间内应缴纳的款项。

借款费用 直接归属于符合条件资产（即需要相当长时间才能达到预定可使用或可销售状态的资产）的购置、建造或生产的借款费用，应予资本化，计入该资产成本。当该资产基本达到预定可使用或可销售状态时，停止资本化相关借款费用。所有其他借款费用在发生当期计入损益。借款费用包括主体因借款而产生的利息及其他费用。

外币 历史财务信息以美元列示，美元为本公司的功能货币。本集团各实体自行确定其功能货币，各实体历史财务信息所包含的项目均以该功能货币计量。本集团各实体记录的外币交易，初始按交易发生日各自功能货币的汇率入账。以外币计价的货币性资产和负债，按各相关期间末的功能货币汇率折算。货币性项目结算或折算产生的差额在损益表中确认。

以历史成本计量的外币非货币性项目，按初始交易日的汇率折算。以公允价值计量的外币非货币性项目，按公允价值计量日的汇率折算。以公允价值计量的非货币性项目折算产生的利得或损失，与该项目公允价值变动利得或损失的确认保持一致处理（即公允价值利得或损失计入其他综合收益或损益的项目，其折算差额亦分别计入其他综合收益或损益）。

在确定与预付对价相关的非货币性资产或非货币性负债终止确认时，就相关资产、费用或收入初始确认所适用汇率而言，初始交易日为本集团初始确认因预付对价所产生的非货币性资产或非货币性负债之日。若存在多笔预付款项或收款，本集团应就每笔预付对价的支付或收取确定各自的交易日。

若干附属公司的功能货币为美元以外的其他货币。于各相关期间末，该等实体的资产及负债按各相关期间末的现行汇率折算为美元，其损益表按与交易日现行汇率相近的汇率折算为美元。

由此产生的汇兑差额在其他综合收益中确认，并累积于汇率波动储备，但归属于非控股权益的差额除外。处置境外经营时，储备中与该境外经营相关的累计金额在损益表中确认。

就合并现金流量表而言，境外附属公司的现金流量按现金流量发生日的汇率折算为美元。境外附属公司在年内经常性发生的现金流量，按年度加权平均汇率折算为美元。

– I-30 –

附录一 3.

会计师报告

重大会计判断及估计

Nine months ended 30 September

The preparation of the Group's Historical Financial Information requires management to make judgements, estimates and assumptions that affect the reported amounts of revenues, expenses, assets and liabilities, and their accompanying disclosures, and the disclosure of contingent liabilities. Uncertainty about these assumptions and estimates could result in outcomes that could require a material adjustment to the carrying amounts of the assets or liabilities affected in the future.

Estimation uncertainty

The key assumptions concerning the future and other key sources of estimation uncertainty at the end of reporting period, that have a significant risk of causing a material adjustment to the carrying amounts of assets and liabilities within the next financial year, are described below:

Fair values of convertible redeemable preferred shares

The fair values of the convertible redeemable preferred shares measured at fair value through profit or loss are determined using the valuation techniques, including the backsolve method and the equity allocation method. Such valuation is based on key parameters about risk-free rate, discounts for lack of marketability ("DLOM") and volatility, which are subject to uncertainty and might materially differ from the actual results. Further details are included in note 24 to the Historical Financial Information.

Share-based payments

The Group operates share option schemes for the purpose of providing incentives for employees and persons contributing to the Group. The fair value of the option is determined using the backsolve method, option pricing and binomial model at the grant dates. Valuation techniques are certified by an independent valuer before being implemented for valuation and are calibrated to ensure that outputs reflect market conditions. Some inputs, such as the discount rate for lack of marketability ("DLOM"), discount rate and volatility, require management estimates. Should any of the estimates and assumptions change, it may lead to a change in the fair value to be recognised in the statement of profit or loss. Further details are contained in note 26 to the Historical Financial Information.

4. OPERATING SEGMENT INFORMATION

For management purposes, the Group is organised into one single business unit that includes primarily the rendering of services based on the AI foundation model. Management reviews the overall results and financial position of the Group as a whole based on the same accounting policies set out in note 2.4 to the Historical Financial Information. Accordingly, the Group has only one single operating segment and no further analysis of the single segment is presented.

Geographical information

(a) Revenue from external customers

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Mainland China | – | 2,797 | 9,217 | 6,768 | 14,400 | | Singapore | – | 1 | 11,455 | 7,664 | 12,980 | | United States | – | 575 | 4,999 | 2,871 | 10,913 | | Others | – | 87 | 4,852 | 2,151 | 15,144 | | Total revenue | – | 3,460 | 30,523 | 19,454 | 53,437 |

The revenue information above is based on the locations of the customers.

– I-31 –

APPENDIX I ACCOUNTANT'S REPORT

(b) Non-current assets

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Mainland China | 689 | 4,022 | 4,170 | 3,880 | | Total non-current assets | 689 | 4,022 | 4,170 | 3,880 |

The non-current asset information above is based on the locations of the assets and include Property, plant and equipment and Right-of-use assets.

Information about major customers

Revenues from customers, including a group of entities which are known to be under common control, which individually accounted for over 10% of the Group's total revenue during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 are as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Customer A | N/A | 1,286 | N/A | N/A | N/A | | Customer B | N/A | 426 | N/A | N/A | N/A | | Customer C | N/A | N/A | 9,438 | 6,504 | 7,828 |

5. REVENUE, OTHER INCOME AND GAINS

An analysis of revenue from contracts with customers is as follows:

(a) Disaggregation of revenue from contracts with customers

Revenue during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 is as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | AI-native products | – | 758 | 21,805 | 13,529 | 38,020 | | Open Platform and other AI-based enterprise services | – | 2,702 | 8,718 | 5,925 | 15,417 | | Revenue from services provided | – | 3,460 | 30,523 | 19,454 | 53,437 |

– I-32 –

APPENDIX I ACCOUNTANT'S REPORT

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Timing of revenue recognition | | | | | | | Services transferred at a point in time | – | 2,702 | 25,695 | 15,825 | 30,322 | | Services transferred over time | – | 758 | 4,828 | 3,629 | 23,115 | | Total | – | 3,460 | 30,523 | 19,454 | 53,437 |

(b) Performance obligations

Information about the Group's performance obligations is described in note 2.4 to the consolidated financial statement under "Revenue recognition". The Group also obtained advance payment from the Membership subscription and the Virtual items.

The Company elected to use the practical expedient to not disclose the remaining performance obligations, as substantially all of the Company's contracts have duration of one year or less.

(c)

Revenue recognised in relation to contract liabilities

The amounts of revenue recognised during the years ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 that were included in the contract liabilities at the beginning of those periods were nil, nil, USD559,000, USD487,000 (unaudited) and USD1,358,000, respectively.

The amounts of transaction prices allocated to the remaining performance obligations (unsatisfied or partially unsatisfied) as at the end of each of the Relevant Periods were nil, USD559,000, USD1,553,000 and USD4,657,000. The revenue attributable to these remaining performance obligations is expected to be recognised within one year.

**Other income and gains, net**

An analysis of other income and gains is as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Interest income | 39 | 7,785 | 20,448 | 17,199 | 7,876 | | Foreign exchange gains, net | 175 | 311 | 2 | 1,415 | 1,600 | | Fair value gain on financial assets at fair value through profit or loss | 941 | 788 | 15,710 | 6,682 | 20,414 | | Others | – | 58 | (9) | (18) | 1,342 | | **Total** | **1,155** | **8,942** | **36,151** | **25,278** | **31,232** |

– I-33 –

---

APPENDIX I — ACCOUNTANT'S REPORT

6. FINANCE COSTS

An analysis of finance costs is as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Interest on interest-bearing bank borrowings | – | – | 355 | 199 | 404 | | Interest on lease liabilities | 14 | 61 | 154 | 117 | 107 | | **Total** | **14** | **61** | **509** | **316** | **511** |

---

7. LOSS BEFORE TAX

The Group's loss before tax is arrived at after charging/(crediting):

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Cost of services provided (excluding employment benefit) | – | 4,314 | 26,785 | 18,944 | 40,348 | | Depreciation of property, plant and equipment | 25 | 180 | 451 | 325 | 582 | | Depreciation of right-of-use assets | 182 | 631 | 1,450 | 1,072 | 1,478 | | Listing expenses | – | – | – | – | 3,675 | | Research and development costs (excluding employee benefit expenses, depreciation and amortisation costs) | 5,011 | 49,465 | 143,807 | 105,410 | 145,434 | | Employee benefit expenses: | | | | | | | — Wages and salaries | 5,676 | 19,762 | 44,036 | 30,676 | 38,851 | | — Pension scheme contributions | 188 | 1,106 | 2,402 | 1,803 | 2,009 | | — Share-based payment expenses | 106 | 2,068 | 4,548 | 4,442 | 7,338 | | Fair value loss on financial liabilities | 60,509 | 176,826 | 214,172 | 128,063 | 313,477 | | Impairment losses on financial assets, net | – | 3 | 88 | 68 | 22 | | Fair value gains on financial assets at fair value through profit or loss | (941) | (788) | (15,710) | (6,682) | (20,414) |

– I-34 –

---

APPENDIX I — ACCOUNTANT'S REPORT

8. DIRECTORS' AND CHIEF EXECUTIVE'S REMUNERATION

Directors' and chief executive's remuneration for the year/period, disclosed pursuant to the Listing Rules, section 383(1)(a), (b), (c) and (f) of the Hong Kong Companies Ordinance and Part 2 of the Companies (Disclosure of Information about Benefits of Directors) Regulation, is as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Salaries, allowances and benefits in kind | 358 | 625 | 1,144 | 832 | 1,011 | | Performance related bonuses | 131 | 247 | 170 | – | – | | Pension scheme contributions | 9 | 24 | 39 | 31 | 23 | | Equity-settled share option expense | 963 | 1,278 | 2,275 | 1,658 | 1,243 | | **Total** | **1,461** | **2,174** | **3,628** | **2,521** | **2,277** |

During the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025, certain directors were granted share options, in respect of their services to the Group, under the share option scheme of the Company, further details of which are set out in note 26 to the Historical Financial Information. The fair value of such options, which has been recognised in the statement of profit or loss over the vesting period, was determined as at the date of grant and the amount included in the financial statements for the current year is included in the above directors' and chief executive's remuneration disclosures.

**(a) Independent non-executive directors**

During the Relevant Periods, Mr. Huang Guobin, Mr. Wang Pengcheng and Mr. Zhu Huaxing were appointed as independent non-executive directors of the Company from the Listing Date.

There was no emolument payable to the independent non-executive directors during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025.

**(b) Executive directors, a non-executive director and the chief executive**

**2022**

| | Salaries, allowances and benefits in kind USD'000 | Performance related bonuses USD'000 | Pension scheme contributions USD'000 | Equity-settled share option expense USD'000 | Total remuneration USD'000 | |---|---|---|---|---|---| | **Executive directors:** | | | | | | | Ms. Yun Yeyi (i) | 94 | 80 | 4 | 963 | 1,141 | | Mr. Yang Bin (ii) | 180 | 51 | 2 | – | 233 | | Ms. Wang Meng (iii) | 84 | – | 3 | – | 87 | | **Total** | **358** | **131** | **9** | **963** | **1,461** |

– I-35 –

---

APPENDIX I — ACCOUNTANT'S REPORT

**2023**

| | Salaries, allowances and benefits in kind USD'000 | Performance related bonuses USD'000 | Pension scheme contributions USD'000 | Equity-settled share option expense USD'000 | Total remuneration USD'000 | |---|---|---|---|---|---| | **Executive directors:** | | | | | | | Ms. Yun Yeyi (i) | 179 | 94 | 7 | 1,278 | 1,558 | | Mr. Yang Bin (ii) | 216 | 43 | 7 | – | 266 | | Mr. Zhang Mozhi (iv) | 159 | 55 | 7 | – | 221 | | Mr. Yan Junjie (v) | 71 | 55 | 3 | – | 129 | | **Total** | **625** | **247** | **24** | **1,278** | **2,174** |

Pension scheme contributions

Equity-settled share option expense

Total remuneration

USD'000

USD'000

USD'000

USD'000

USD'000

2024 Executive directors: Ms. Yun Yeyi (i) ў ў ў ў ў Mr. Wei Wei (vi) ў ў ў ў ў Mr. Zhang Qianchuan (vii) ў ў ў ў ў Mr. Yan Junjie (v) ў ў ў ў Mr. Zhang Mozhi (iv) ў ў Mr. Yang Bin (ii) ў ў ў ў ў Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў

30 September 2024 (unaudited) Executive directors: Ms. Yun Yeyi (i) ў ў ў ў ў Mr. Wei Wei (vi) ў ў ў ў ў Mr. Zhang Qianchuan (vii) ў ў ў ў ў Mr. Yang Bin (ii) ў ў ў ў ў Mr. Yan Junjie (v) ў ў ў ў Mr. Zhang Mozhi (iv) ў ў Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў

267 212

86 –

7 7

1,282 736

1,642 955

123 173 168 201

– 42 42 –

4 7 7 7

257 – – –

384 222 217 208

1,144

170

39

2,275

3,628

Salaries, allowances and benefits in kind

Performance related bonuses

Pension scheme contributions

Equity-settled share option expense

Total remuneration

USD'000

USD'000

USD'000

USD'000

USD'000

126 159

– –

5 5

960 441

1,091 605

123 166 130 128

– – – –

4 6 5 6

257 – – –

384 172 135 134

832

–

31

1,658

2,521

Salaries, allowances and benefits in kind

Performance related bonuses

Pension scheme contributions

Equity-settled share option expense

Total remuneration

USD'000

USD'000

USD'000

USD'000

USD'000

30 September 2025 ў ў ў ў ў Executive directors: Ms. Yun Yeyi (i) ў ў ў ў ў Mr. Yan Junjie (v) ў ў ў ў Mr. Zhao Pengyu (viii) ў Mr. Zhou Yucong (viii) ў

518 139 166 188

– – – –

5 7 5 6

956 – 125 162

1,479 146 296 356

Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў

1,011

–

23

1,243

2,277

– I-36 –

APPENDIX I

ACCOUNTANT'S REPORT

(i)

Ms. Yun Yeyi served as a director and chief operating officer of the Company since December 2022, and was re-designated as an executive Director in June 2025.

(ii)

Mr. Yang Bin served as a director of the Company since December 2022, and tendered his resignation in August 2024 due to the commercial arrangement.

(iii)

Ms. Wang Meng served as a director of the Company since December 2021, and tendered her resignation in December 2022 due to the commercial arrangement.

(iv)

Mr. Zhang Mozhi served as a director of the Company since October 2023, and tendered his resignation in December 2024 due to the commercial arrangement.

(v)

Mr. Yan Junjie served as a director, the chief executive officer and chief technology officer since October 2023, and was re-designated as our executive Director in June 2025.

(vi)

Mr. Wei Wei served as a director of the Company since March 2024, and tendered his resignation in December 2024 due to the commercial arrangement.

(vii) Mr. Zhang Qianchuan served as a director of the Company since March 2024, and tendered his resignation in December 2024 due to the commercial arrangement.

(viii) Mr. Zhao Pengyu and Mr. Zhou Yucong served as a director of the Company since June 2025.

There was no arrangement under which a director or the chief executive waived or agreed to waive any remuneration during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025.

(c)

Directors' retirement benefits and termination benefits

Mr. Yang Bin received termination benefits amounting to USD35,000 in the year ended 31 December 2024.

No other director's retirement or termination benefit subsisted at the end of each year disclosed or at any time during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025.

9.

FIVE HIGHEST PAID EMPLOYEES

The five highest paid employees during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 included two, three, two, three and two directors, details of whose remuneration are set out in above. Details of the remuneration for the remaining three, two, three, two and three highest paid employees who are neither a director nor chief executive of the Company during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 are as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | | Salaries, allowances and benefits in kind | 337 | 303 | 610 | 230 | 235 | | Performance related bonuses | 150 | 156 | 220 | – | – | | Pension scheme contributions | 17 | 10 | 21 | 13 | 10 | | Termination benefits | – | – | – | – | 229 | | Equity-settled share option expense | 39 | 734 | 365 | 693 | 4,221 | | Total | 543 | 1,203 | 1,216 | 936 | 4,695 |

– I-37 –

APPENDIX I

ACCOUNTANT'S REPORT

The number of non-director and non-chief executive highest paid employees whose remuneration fell within the following bands is as follows:

| | Number of employees | | | | | |---|---|---|---|---|---| | | Year ended 31 December | | | Nine months ended 30 September | | | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | Nil to HK$2,000,000 | 3 | – | – | – | – | | HK$2,000,001 to HK$4,000,000 | – | 1 | 3 | 1 | 2 | | HK$4,000,001 to HK$6,000,000 | – | 1 | – | 1 | – | | HK$6,000,001 to HK$30,000,000 | – | – | – | – | 1 | | Total | 3 | 2 | 3 | 2 | 3 |

During the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025, share options were granted to a non-director and non-chief executive highest paid employee in respect of his services to the Group, further details of which are included in the disclosures in note 26 to the Historical Financial Information. The fair value of such options, which has been recognised in the statement of profit or loss over the vesting period, was determined as at the date of grant and the amount included in the financial statements for the current year is included in the above non-director and non-chief executive highest paid employees' remuneration disclosures.

10.

INCOME TAX

The Group is subject to income tax on an entity basis on profits arising in or derived from the jurisdictions in which members of the Group are domiciled and operate.

**Cayman Islands**

Under the current laws of the Cayman Islands, the Company and its subsidiaries are not subject to tax on income or capital gains.

**Hong Kong**

The subsidiary incorporated in Hong Kong is subject to Hong Kong profits tax at the rate of 16.5% on any estimated assessable profits arising in Hong Kong during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025. The first HK$2,000,000 of assessable profits of this subsidiary were taxed at 8.25% and the remaining assessable profits were taxed at 16.5% during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025.

**Singapore**

The subsidiaries incorporated in Singapore are subject to Singapore profits tax at the rate of 17% on any estimated assessable profits arising in Singapore during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025.

– I-38 –

APPENDIX I

ACCOUNTANT'S REPORT

**Mainland China**

The provision for corporate income tax ("CIT") in Mainland China is based on the statutory rate of 25% of the assessable profits as determined in accordance with the PRC Corporate Income Tax Law which was approved and became effective on 1 January 2008.

Beijing Jizhi was qualified as a High and New Technology Enterprise in 2023 and is entitled to a preferential CIT rate of 15% from 2023 to 2025. This qualification is subject to review by the relevant tax authority in the PRC every three years.

Shanghai Jizhi was qualified as a High and New Technology Enterprise in 2024 and is entitled to a preferential CIT rate of 15% from 2024 to 2026. This qualification is subject to review by the relevant tax authority in the PRC every three years.

A reconciliation of the tax expense applicable to loss before tax using the statutory rate for the countries or jurisdictions in which the Company and its subsidiaries are domiciled and operate to the tax expense at the applicable tax rate is as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Loss before tax | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | Tax calculated at statutory tax rates of each entities' jurisdictions | (3,474) | (23,469) | (67,941) | (47,372) | (54,175) | | Effect of preferential tax rates | – | 2,707 | 22,254 | 16,169 | 20,909 | | Expenses not deductible for tax | 1 | 30 | 32 | 27 | 24 | | Additional deductible allowance for qualified research and development costs | (1,082) | (5,989) | (9,573) | (7,732) | (10,004) | | Temporary difference and tax losses not recognised | 4,555 | 26,721 | 55,228 | 38,908 | 43,246 | | Tax charge at the Group's effective rate | – | – | – | – | – |

Deferred tax assets have not been recognized during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 in respect of these losses as they have arisen in subsidiaries that have been loss-making for some time and it is not considered probable that taxable profits will be available against which the tax losses can be utilised.

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Tax losses | 16,702 | 135,099 | 404,349 | 332,354 | 637,650 | | Temporary differences | 1,514 | 7,206 | 62,629 | 41,712 | 91,718 | | Total | 18,216 | 142,305 | 466,978 | 374,066 | 729,368 |

– I-39 –

APPENDIX I

ACCOUNTANT'S REPORT

The tax losses incurred from the Company's subsidiaries in Mainland China that are not recognised as deferred tax assets will expire from 2027 to 2035. Tax losses of the Group's subsidiaries incorporated in Hong Kong and Singapore will be carried forward indefinitely. Deductible losses that are not recognised for deferred income tax assets will expire in the following years:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | 2027 | 16,702 | 14,677 | – | – | – | | 2028 | – | 70,023 | 3,136 | 3,136 | 3,136 | | 2029 | – | – | 685 | 1,161 | 685 | | 2030 | – | – | – | – | 1,725 | | 2031 | – | – | – | – | – | | 2032 | – | 2,025 | 16,702 | 16,702 | 16,702 | | 2033 | – | 27,070 | 93,957 | 93,957 | 93,957 | | 2034 | – | – | 222,542 | 161,686 | 222,542 | | 2035 | – | – | – | – | 209,095 | | Indefinitely | – | 21,304 | 67,327 | 55,712 | 89,808 | | **Total** | **16,702** | **135,099** | **404,349** | **332,354** | **637,650** |

**11. DIVIDENDS**

The Board did not recommend the payment of any dividend during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025.

**12. LOSS PER SHARE ATTRIBUTABLE TO ORDINARY EQUITY HOLDERS OF THE PARENT**

**(a) Basic loss per share**

Basic loss per share during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 are calculated by dividing the loss attributable to owners of the parent by the weighted average number of ordinary shares in issue during the respective periods.

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | Loss attributable to owners of the parent (expressed in USD'000) | (73,728) | (269,246) | (465,238) | (304,342) | (512,013) | | Weighted average number of ordinary shares in issue during the year/period used in the basic loss per share calculation | 100,000,000 | 105,334,213 | 108,650,075 | 108,650,075 | 108,650,075 | | Basic loss per share (expressed in USD) | | | | | |

**(b)**

(0.74)

(2.56)

(4.28)

(2.80)

(4.71)

Diluted loss per share

As the Group incurred losses during the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025, the potential ordinary shares were not included in the calculation of diluted loss per share as their inclusion would be anti-dilutive. Accordingly, diluted loss per share for the year ended 31 December 2022, 2023 and 2024 and the nine months ended 30 September 2024 and 2025 are the same as basic loss per share of the respective periods.

– I-40 –

APPENDIX I 13.

ACCOUNTANT'S REPORT

PROPERTY, PLANT AND EQUIPMENT

| | As at 31 December | | | |---|---|---|---| | | Leasehold improvements USD'000 | Office equipment USD'000 | Total USD'000 |

31 December 2022 At 1 January 2022 (unaudited): | Cost | – | – | – | | Accumulated depreciation | – | – | – | | Net carrying amount | – | – | – |

| At 1 January 2022, net of accumulated depreciation (unaudited) | – | – | – | | Additions | – | 256 | 256 | | Depreciation provided during the year | – | (25) | (25) |

| At 31 December 2022, net of accumulated depreciation | – | 231 | 231 |

At 31 December 2022: | Cost | – | 256 | 256 | | Accumulated depreciation | – | (25) | (25) | | Net carrying amount | – | 231 | 231 |

| | As at 31 December | | | |---|---|---|---| | | Leasehold improvements USD'000 | Office equipment USD'000 | Total USD'000 |

31 December 2023 At 1 January 2023: | Cost | – | 256 | 256 | | Accumulated depreciation | – | (25) | (25) | | Net carrying amount | – | 231 | 231 |

| At 1 January 2023, net of accumulated depreciation | – | 231 | 231 | | Additions | 256 | 402 | 658 | | Depreciation provided during the year | (46) | (134) | (180) |

| At 31 December 2023, net of accumulated depreciation | 210 | 499 | 709 |

At 31 December 2023: | Cost | 256 | 658 | 914 | | Accumulated depreciation | (46) | (159) | (205) | | Net carrying amount | 210 | 499 | 709 |

– I-41 –

APPENDIX I

ACCOUNTANT'S REPORT

| | As at 31 December | | | |---|---|---|---| | | Leasehold improvements USD'000 | Office equipment USD'000 | Total USD'000 |

31 December 2024 At 1 January 2024: | Cost | 256 | 658 | 914 | | Accumulated depreciation | (46) | (159) | (205) | | Net carrying amount | 210 | 499 | 709 |

| At 1 January 2024, net of accumulated depreciation | 210 | 499 | 709 | | Additions | 352 | 483 | 835 | | Depreciation provided during the year | (168) | (283) | (451) |

| At 31 December 2024, net of accumulated depreciation | 394 | 699 | 1,093 |

At 31 December 2024: | Cost | 608 | 1,141 | 1,749 | | Accumulated depreciation | (214) | (442) | (656) | | Net carrying amount | 394 | 699 | 1,093 |

| | As at 30 September | | | |---|---|---|---| | | Leasehold improvements USD'000 | Office equipment USD'000 | Total USD'000 |

30 September 2025 At 1 January 2025: | Cost | 608 | 1,141 | 1,749 | | Accumulated depreciation | (214) | (442) | (656) | | Net carrying amount | 394 | 699 | 1,093 |

| At 1 January 2025, net of accumulated depreciation | 394 | 699 | 1,093 | | Additions | 445 | 178 | 623 | | Depreciation provided during the period | (313) | (269) | (582) |

| At 30 September 2025, net of accumulated depreciation | 526 | 608 | 1,134 |

At 30 September 2025: | Cost | 1,053 | 1,319 | 2,372 | | Accumulated depreciation | (527) | (711) | (1,238) | | Net carrying amount | 526 | 608 | 1,134 |

During the Relevant Periods, there was no impairment provided for the Group's property, plant and equipment.

– I-42 –

APPENDIX I 14.

ACCOUNTANT'S REPORT

LEASES

The Group as a lessee The Group has lease contracts for buildings used in its operations, and the lease terms are generally between 2 and 3 years. Generally, the Group is restricted from assigning and subleasing the leased assets outside the Group.

(a) Right-of-use assets

The carrying amounts of the Group's right-of-use assets and the movements during the Relevant Periods are as follows:

| | Buildings USD'000 | |---|---|

(b)

| As at 1 January 2022 (unaudited) | – | | Additions | 640 | | Depreciation charge | (182) |

| As at 31 December 2022 | 458 | | Additions | 3,718 | | Depreciation charge | (631) | | Disposal as a result of early cancellation of lease | (232) |

| As at 31 December 2023 | 3,313 |

Additions ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Depreciation charge ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Disposal as a result of early cancellation of lease ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

1,249 (1,450) (35)

As at 31 December 2024 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Additions ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Depreciation charge ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Disposal as a result of early cancellation of lease ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

3,077 1,815 (1,478) (668)

As at 30 September 2025 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

2,746

The carrying amount of lease liabilities and the movements during the Relevant Periods are as follows:

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Carrying amount at the beginning of the year/period | – | 440 | 3,160 | 3,023 | | New leases | 640 | 3,718 | 1,249 | 1,815 | | Disposal as a result of early cancellation of lease | – | (302) | (34) | (843) | | Accretion of interest recognised during the year/period | 14 | 61 | 154 | 107 | | Payments | (214) | (757) | (1,506) | (1,471) | | Carrying amount at the end of the year/period | 440 | 3,160 | 3,023 | 2,631 | | Analysed into: | | | | | | Current portion | 349 | 1,248 | 1,964 | 1,694 | | Non-current portion | 91 | 1,912 | 1,059 | 937 |

– I-43 –

APPENDIX I ACCOUNTANT'S REPORT

(c) The amounts recognised in profit or loss in relation to leases are as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Interest on lease liabilities | 14 | 61 | 154 | 117 | 107 | | Depreciation charge of right-of-use assets | 182 | 631 | 1,450 | 1,072 | 1,478 | | Expense relating to short-term leases | 42 | 76 | – | – | 7 | | Total amount recognised in profit or loss | 238 | 768 | 1,604 | 1,189 | 1,592 |

15. TRADE RECEIVABLES

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Trade receivables | – | 1,341 | 7,073 | 8,176 | | Impairment | – | (3) | (91) | (113) | | Net carrying amount | – | 1,338 | 6,982 | 8,063 |

Amounts due from the related parties included in the Group's trade receivables were nil, USD41,000, USD41,000 and USD113,000 as at 31 December 2022, 2023 and 2024 and 30 September 2025, respectively, which are recoverable within one year.

The Group's trading terms with its customers are mainly on credit. The credit term is generally from 30 to 60 days. The Group seeks to maintain strict control over its outstanding receivables and has a credit control process to minimise credit risk. The Group does not hold any collateral or other credit enhancements over its trade receivable balances. Trade receivables are non-interest-bearing.

An ageing analysis of the trade receivables as at the end of each of the Relevant Periods, based on the invoice date and net of loss allowance, is as follows:

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Within one year | – | 1,338 | 6,982 | 8,063 |

The movements in the loss allowance for impairment of trade receivables are as follows:

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | At beginning of the year/period | – | – | 3 | 91 | | Additions | – | 3 | 88 | 22 | | At end of the year/period | – | 3 | 91 | 113 |

– I-44 –

APPENDIX I ACCOUNTANT'S REPORT

Set out below is the information about the credit risk exposure on the Group's trade receivables using a provision matrix:

As at 31 December 2023

| | Within one year | Total | |---|---|---| | Expected credit loss rate | 0.2% | 0.2% | | Gross carrying amount (USD'000) | 1,341 | 1,341 | | Expected credit losses (USD'000) | 3 | 3 |

As at 31 December 2024

| | Within one year | Total | |---|---|---| | Expected credit loss rate | 1.3% | 1.3% | | Gross carrying amount (USD'000) | 7,073 | 7,073 | | Expected credit losses (USD'000) | 91 | 91 |

As at 30 September 2025

| | Within one year | Total | |---|---|---| | Expected credit loss rate | 1.4% | 1.4% | | Gross carrying amount (USD'000) | 8,176 | 8,176 | | Expected credit losses (USD'000) | 113 | 113 |

16. PREPAYMENTS, OTHER RECEIVABLES AND OTHER ASSETS

The Group

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | **Current** | | | | | | Prepayments | 56 | 119 | 394 | 451 | | Value-added-tax recoverable (i) | 387 | 3,854 | 7,144 | 1,500 | | Deferred listing expenses | – | – | – | 422 | | Deposits and other receivables (ii) | 126 | 405 | 5,932 | 9,438 | | Total | 569 | 4,378 | 13,470 | 11,811 | | **Non-current** | | | | | | Deposits and other receivables (ii) | – | 435 | 561 | 731 | | Total | – | 435 | 561 | 731 |

(i) The Group's domestic sales of services are subject to PRC value-added-tax ("VAT"). Input VAT on purchases can be deducted from output VAT payable. The VAT recoverable is mainly the net difference between output and deductible input VAT.

(ii)

– I-45 –

APPENDIX I

ACCOUNTANT'S REPORT

The Company | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

17.

| | | | | | |---|---|---|---|---| | Current | | | | | | Amounts due from subsidiaries | 14,100 | 103,895 | 358,591 | 663,220 | | Deferred listing expenses | – | – | – | 422 | | Other receivables | – | – | 1,500 | – | | **Total** | **14,100** | **103,895** | **360,091** | **663,642** |

INVESTMENTS

The Group | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

| | | | | | |---|---|---|---|---| | **Non-current assets** | | | | | | Financial assets at fair value through profit or loss | | | | | | Wealth management products, at fair value | – | – | 95,331 | 70,228 | | Financial assets at fair value through other comprehensive income | | | | | | Investment in a listed entity | – | – | 4,836 | 6,440 | | **Total** | **–** | **–** | **100,167** | **76,668** | | **Current assets** | | | | | | Financial assets at fair value through profit or loss | | | | | | Wealth management products, at fair value | 65,791 | 15,802 | 295,220 | 644,154 | | Financial assets at amortised costs | – | – | 147,444 | – | | **Total** | **65,791** | **15,802** | **442,664** | **644,154** |

The Company | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

| | | | | | |---|---|---|---|---| | **Non-current assets** | | | | | | Financial assets at fair value through profit or loss | | | | | | Wealth management products, at fair value | – | – | 95,331 | 70,228 | | Investments in subsidiaries | 1,069 | 4,415 | 11,238 | 19,819 | | **Total** | **1,069** | **4,415** | **106,569** | **90,047** |

– I-46 –

APPENDIX I

ACCOUNTANT'S REPORT

18. | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

| | | | | | |---|---|---|---|---| | **Current assets** | | | | | | Financial assets at fair value through profit or loss | | | | | | Wealth management products, at fair value | 65,791 | 10,152 | 295,220 | 639,899 | | Financial assets at amortised costs | – | – | 147,444 | – | | **Total** | **65,791** | **10,152** | **442,664** | **639,899** |

CASH AND CASH EQUIVALENTS

The Group | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

| | | | | | |---|---|---|---|---| | Cash and bank balances | 6,912 | 18,188 | 316,024 | 293,762 | | Time deposits | – | 279,844 | 26,546 | 94,023 | | **Subtotal** | **6,912** | **298,032** | **342,570** | **387,785** | | Less: Pledged for bank borrowings | – | – | 15,491 | 6,154 | | Pledged for guaranteeing bank acceptance bills payable | – | – | 11,802 | 18,943 | | Other restricted cash* | 2,221 | 39 | 38 | 41 | | Non-pledged time deposits with original maturity of more than three months when acquired | – | 91,698 | 26,327 | – | | **Cash and cash equivalents** | **4,691** | **206,295** | **288,912** | **362,647** |

Denominated in: | | | | | | |---|---|---|---|---| | USD | 1,884 | 190,194 | 130,440 | 280,728 | | RMB | 2,807 | 15,675 | 157,219 | 79,235 | | SGD | – | 426 | 1,253 | 2,684 | | **Total** | **4,691** | **206,295** | **288,912** | **362,647** |

*Other restricted cash primarily reflects bank balances subject to institutional clearing timelines and is automatically released following standard interbank processing protocols.

The RMB is not freely convertible into other currencies, however, under Mainland China's Foreign Exchange Control Regulations and Administration of Settlement, and Sale and Payment of Foreign Exchange Regulations, the Group is permitted to exchange RMB for other currencies through banks authorised to conduct foreign exchange business.

Cash at banks earns interest at floating rates based on daily bank deposit rates. Short term time deposits are made for varying periods of between one day and three months, depending on the immediate cash requirements of the Group, and earn interest at the respective short term time deposit rates. The bank balances and time deposits are deposited with creditworthy banks with no recent history of default.

– I-47 –

APPENDIX I

ACCOUNTANT'S REPORT

The Company | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

19. | | | | | | |---|---|---|---|---| | Cash and bank balances | 1,784 | 3,488 | 246,792 | 164,315 | | Time deposits | – | 279,744 | 26,546 | 86,397 | | **Subtotal** | **1,784** | **283,232** | **273,338** | **250,712** | | Less: Pledged for guaranteeing bank acceptance bills payable | – | – | 11,802 | – | | Non-pledged time deposits with original maturity of more than three months when acquired | – | 91,598 | 26,327 | – | | **Cash and cash equivalents** | **1,784** | **191,634** | **235,209** | **250,712** |

Denominated in: | | | | | | |---|---|---|---|---| | USD | 1,784 | 187,330 | 126,657 | 250,676 | | RMB | – | 4,304 | 108,552 | 36 | | **Total** | **1,784** | **191,634** | **235,209** | **250,712** |

INTEREST-BEARING BANK BORROWINGS | | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

| | | | | | |---|---|---|---|---| | **Current** | | | | | | Bank loans — unsecured | – | – | – | 14,070 | | Bank loans — secured | – | – | 19,455 | 5,032 | | **Total** | **–** | **–** | **19,455** | **19,102** |

APPENDIX I

ACCOUNTANT'S REPORT

The weighted average interest rates for the years ended 2024 and the nine months ended September 30, 2025 were 3.14% and 3.00% respectively.

20. TRADE AND BILLS PAYABLES

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Within 1 year | 2,394 | 17,242 | 51,159 | 70,219 | | Over 1 year | – | – | 53 | – | | Total | 2,394 | 17,242 | 51,212 | 70,219 |

Amounts due to related parties included in the Group's trade and bills payables were USD10,000, USD592,000, USD4,022,000 and USD26,509,000 as at 31 December 2022, 2023 and 2024 and 30 September 2025, respectively.

Trade and bills payables are non-interest-bearing and normally settled on terms of 30 to 90 days.

– I-48 –

APPENDIX I

ACCOUNTANT'S REPORT

21. OTHER PAYABLES, ACCRUALS AND OTHER LIABILITIES

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Payroll payables | 1,976 | 5,469 | 10,596 | 9,052 | | Other tax payables | 82 | 303 | 644 | 1,209 | | Convertible bonds (i) | – | – | 14,722 | – | | Other payables and accruals | 268 | 8,969 | 25,550 | 7,061 | | Total | 2,326 | 14,741 | 51,512 | 17,322 |

(i) In May 2024, the Company issued interest-free convertible bonds with a principal amount of RMB100,000,000 (USD13,910,000 equivalent). The convertible bonds will mature in five years since issuance unless having been redeemed, repurchased or converted prior to such date. As at 31 December 2024, the convertible bonds were carried at the then fair values of USD14,722,000. The Company repurchased the convertible bonds at RMB105,000,000 (USD14,668,000 equivalent) as renegotiated with the holders in May 2025.

22. CONTRACT LIABILITIES

Details of contract liabilities are as follows:

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Short-term advances received from customers: | | | | | | Rendering of services | – | 559 | 1,553 | 4,657 |

Contract liabilities include advances received from customers for delivery of enterprise services and membership subscription. The increase in contract liabilities was mainly due to the increase in advances received from customers for services in future.

23. OTHER NON-CURRENT LIABILITIES

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2025 USD'000 | | Government grants | – | 1,218 | 1,200 | 1,467 |

24. CONVERTIBLE REDEEMABLE PREFERRED SHARES

The convertible redeemable preferred shareholders have been granted certain special rights in relation to the Group, including but not limited to redemption rights, conversion rights at any time after investing, the pre-emptive rights, right of co-sale, liquidation preferences, rights of first refusal, information rights and director appointment rights. The redemption rights have been suspended immediately prior to the first filing of the listing application and all other special rights will be terminated upon Listing.

Presentation and classification

The Group does not bifurcate any embedded derivatives from the convertible redeemable preferred shares and designates the entire instruments as financial liabilities at fair value through profit or loss. The convertible redeemable preferred shares were classified as current liabilities since the conversion options were not classified as equity and are exercisable at any time at the shareholders' options. The change in fair value is charged to profit or loss except for the portion attributable to credit risk change that shall be charged to other comprehensive income, if any. Management considered that fair value change in the convertible redeemable preferred shares attributable to changes of credit risk was not significant.

– I-49 –

APPENDIX I

ACCOUNTANT'S REPORT

The movements of the convertible redeemable preferred shares are set out below:

| | Series Angel | | Series Pre-A | | Series A | | Series A+ | | Series Pre-B | | Series Pre-B+ | | Series Pre-B++ | | Total | |---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---| | | Number of shares | USD'000 | Number of shares | USD'000 | Number of shares | USD'000 | Number of shares | USD'000 | Number of shares | USD'000 | Number of shares | USD'000 | Number of shares | USD'000 | USD'000 | | As at 1 January 2022 (unaudited) | 18,343,195 | – | – | – | – | – | – | – | – | – | – | – | – | – | – | | Issue | – | – | – | 50,000 | – | 257,000 | – | – | – | – | – | – | – | – | 307,000 | | Changes in fair value | – | – | – | 15,941 | – | 87,306 | – | – | – | – | – | – | – | – | 176,826 | | As at 31 December 2022 | 18,343,195 | 34,666 | – | 65,941 | 37,172,913 | 344,306 | 62,537,371 | – | – | – | – | – | – | – | 34,666 | | Issue | – | – | – | – | – | – | – | 50,000 | – | – | – | – | – | – | 50,000 | | Changes in fair value | – | – | – | 27,390 | – | 41,232 | – | 10,218 | – | – | – | – | – | – | 60,509 | | As at 31 December 2023 | 18,343,195 | 79,234 | – | 93,331 | 37,172,913 | 385,538 | 62,537,371 | 60,218 | – | – | – | – | – | – | 629,001 | | Issue | – | – | – | – | – | – | – | – | – | – | – | – | 25,787,040 | 390,451 | 739,588 | | Changes in fair value | – | 51,912 | – | 23,315 | – | – | – | 1,960 | – | – | – | 93,282 | – | 4,725 | 213,360 | | As at 31 December 2024 | 18,343,195 | 131,146 | – | 116,646 | 37,172,913 | 385,538 | 62,537,371 | 62,178 | – | – | – | 93,282 | 25,787,040 | 395,176 | 1,581,949 | | Issue | – | – | – | – | – | – | 2,915,191 | – | 7,140,526 | 651,872 | – | 87,716 | – | – | 629,001 | | Changes in fair value | – | 43,076 | – | – | – | – | – | – | – | 98,211 | – | 5,566 | – | – | 739,588 | | As at 31 December 2024 | 18,343,195 | 174,222 | – | 143,423 | 37,172,913 | 473,337 | 62,537,371 | 75,405 | 7,140,526 | 750,083 | – | 93,282 | 25,787,040 | 395,176 | 2,321,193 | | Issue | – | – | – | – | – | – | – | – | – | – | 5,677,436 | 35,811 | – | – | 426,262 | | Changes in fair value | – | 39,684 | – | 26,777 | – | – | – | 13,227 | – | 122,925 | – | 17,845 | – | – | 312,982 | | As at 30 September 2025 | 18,343,195 | 213,906 | – | – | 11,834,320 | – | 11,834,320 | – | 11,834,320 | 873,008 | 5,677,436 | 146,938 | 25,787,040 | – | 2,321,193 |

– I-50 –

The Group applied the back-solve method to determine the equity value of the Company and adopted the equity allocation model to determine the fair values of the convertible redeemable preferred shares as at the end of each of the Relevant Periods. Key valuation assumptions used to determine the fair values of the convertible redeemable preferred shares are set below:

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 | | Risk-free interest rate (%) | 4.11% | 4.00% | 4.24% | 3.62% | | Discounts for lack of marketability ("DLOM") (%) | 22% | 23% | 24% | 15% | | Volatility (%) | 54% | 64% | 71% | 62% |

The Group estimated the risk-free interest rate based on the yield of the US Government Bond with maturity close to the expected exit timing as at the valuation date. The DLOM was estimated based on the option-pricing method. Under the option-pricing method, the cost of put option, which can hedge the price change before the privately held share can be sold, was considered as a basis to determine the lack of marketability discount. Volatility was estimated based on annualised standard deviation of daily stock price return of comparable companies for a period from the valuation date and with a similar span as time to expiration.

Set out below is a summary of significant unobservable inputs to the valuation of financial liabilities categorised within Level 3 of the fair value hierarchy, together with a quantitative sensitivity analysis as at the end of each of the Relevant Periods.

Increase/(decrease) in fair value

| Significant unobservable inputs | Increase/(decrease) in the inputs | As at 31 December 2022 USD'000 | As at 31 December 2023 USD'000 | As at 31 December 2024 USD'000 | As at 30 September 2025 USD'000 | |---|---|---|---|---|---| | Risk-free interest rate (%) | +25 bp | (395) | (1,009) | (673) | (753) | | Risk-free interest rate (%) | -25 bp | 363 | 770 | 567 | 756 | | Discounts for lack of marketability ("DLOM") (%) | +1 | (346) | (1,656) | (3,999) | (3,885) | | Discounts for lack of marketability ("DLOM") (%) | -1 | 346 | 1,656 | 3,999 | 3,885 | | Volatility (%) | +1 | 76 | (34) | (148) | 89 | | Volatility (%) | -1 | (80) | (24) | 138 | (99) |

25. DEFICITS

Shares

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | Authorised | | | | | | Ordinary shares | 500,000,000 | 500,000,000 | 500,000,000 | 500,000,000 |

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | Issued | | | | | | Ordinary shares | 100,000,000 | 108,650,075 | 108,650,075 | 108,650,075 |

– I-51 –

APPENDIX I

ACCOUNTANT'S REPORT

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | Fully paid | | | | | | Ordinary shares | – | – | – | – |

The amounts of the Group's reserves and the movements therein for the Relevant Periods are presented in the consolidated statements of changes in equity.

Reserves

The Group

(i) Share option reserve

The share option reserve of the Group represents the equity-settled share-based payment as set out in note 26 to the Historical Financial Information.

(ii) Exchange fluctuation reserve

The exchange fluctuation reserve represents exchange differences arising from the translation of the financial statements of group companies whose functional currencies are different from the Group's presentation currency.

(iii) Fair value reserve of financial assets at fair value through other comprehensive income

The fair value reserve of financial assets at fair value through other comprehensive income comprises the cumulative gains of equity investments designated at fair value through other comprehensive income.

The Company

| | Share option reserve USD'000 | Accumulated losses USD'000 | Total USD'000 | |---|---|---|---| | At 31 December 2021 (unaudited) | – | (3,666) | (3,666) | | Loss for the year | – | (59,834) | (59,834) | | Total comprehensive loss for the year | – | (59,834) | (59,834) | | Recognition of share-based payment expenses | 1,069 | – | 1,069 | | At 31 December 2022 | 1,069 | (63,500) | (62,431) |

| | Share option reserve USD'000 | Accumulated losses USD'000 | Total USD'000 | |---|---|---|---| | At 31 December 2022 | 1,069 | (63,500) | (62,431) | | Loss for the year | – | (168,552) | (168,552) | | Total comprehensive loss for the year | – | (168,552) | (168,552) | | Recognition of share-based payment expenses | 3,346 | – | 3,346 | | At 31 December 2023 | 4,415 | (232,052) | (227,637) |

– I-52 –

APPENDIX I

ACCOUNTANT'S REPORT

| | Share option reserve USD'000 | Accumulated losses USD'000 | Total USD'000 | |---|---|---|---| | At 31 December 2023 | 4,415 | (232,052) | (227,637) | | Loss for the year | – | (178,544) | (178,544) | | Total comprehensive loss for the year | – | (178,544) | (178,544) | | Recognition of share-based payment expenses | 6,823 | – | 6,823 | | At 31 December 2024 | 11,238 | (410,596) | (399,358) |

| | Share option reserve USD'000 | Accumulated losses USD'000 | Total USD'000 | |---|---|---|---| | At 31 December 2023 | 4,415 | (232,052) | (227,637) | | Loss for the period (unaudited) | – | (103,525) | (103,525) | | Total comprehensive loss for the period (unaudited) | – | (103,525) | (103,525) | | Recognition of share-based payment expenses (unaudited) | 6,100 | – | 6,100 | | At 30 September 2024 | 10,515 | (335,577) | (325,062) |

| | Share option reserve USD'000 | Accumulated losses USD'000 | Total USD'000 | |---|---|---|---| | At 31 December 2024 | 11,238 | (410,596) | (399,358) | | Loss for the period | – | | |

(410,596) (287,435)

(399,358) (287,435)

本期综合亏损总额 确认股份支付费用

– 8,581

(287,435) –

(287,435) 8,581

于2025年9月30日

19,819

(698,031)

(678,212)

26.

以股份为基础的支付

为向雇员及对集团作出贡献的人士提供激励、吸引及留住高级管理团队及核心人才，本公司于2021年建立以股权结算的股份支付计划（"上市前计划"），并向集团雇员授出购股权。根据上市前计划授出的购股权在授权书所规定的服务期内归属，并于授出日期起十年后届满。每20份购股权于行使时可换算为1股普通股。

于相关期间各期末，可根据上市前计划发行的最大股份数目分别为11,240,661股、19,890,736股、20,890,736股及20,890,736股，分别占已发行普通股的11%、18%、19%及19%。

于相关期间各期末，根据上市前计划授出的购股权须遵守四种不同的归属机制：（i）就四年分级归属时间表而言，购股权总数的15%须于归属开始日期的第一（1）个周年纪念日归属，购股权总数的25%须于归属开始日期的第二（2）个周年纪念日归属，购股权总数的25%须于归属开始日期的第三（3）个周年纪念日归属，其余已授出购股权须于归属开始日期的第四（4）个周年纪念日归属；（ii）就六年分级归属时间表而言，购股权总数的10%须分别于授出日期及归属开始日期的第一（1）个周年纪念日归属，购股权总数的15%须分别于归属开始日期的第二（2）个周年纪念日至第五（5）个周年纪念日归属，其余20%的购股权总数须于归属开始日期的第六（6）个周年纪念日归属；（iii）就多年分级归属时间表而言，购股权总数的10%须分别于本公司首次公开招股（"首次公开招股"）日期的第一（1）个及第二（2）个周年纪念日归属，购股权总数的20%须于首次公开招股日期的第三（3）个周年纪念日至第六（6）个周年纪念日归属；及（iv）就一次性归属时间表而言，100%的购股权须于归属开始日期或授出日期的第一（1）个周年纪念日归属。此外，上市前计划项下的所有购股权在本公司首次公开招股完成前均受到进一步限制，不得行使，该限制将被视为服务条件并影响归属期。

– I-53 –

附录一

会计师报告

下文载列相关期间内根据上市前计划授出的购股权变动详情：

| | 购股权数目 | 加权平均行使价（每份购股权美元） | 加权平均授出日期公平价值（每份购股权美元） | |---|---|---|---| | 于2022年1月1日尚未行使（未经审核） | – | – | – | | 授出 | 94,187,448 | 0.003 | 0.09 | | 已没收 | – | – | – | | 于2022年12月31日尚未行使 | 94,187,448 | 0.003 | 0.09 | | 授出 | 75,091,320 | 0.017 | 0.22 | | 已没收 | (2,954,870) | 0.020 | 0.18 | | 于2023年12月31日尚未行使 | 166,323,898 | 0.009 | 0.10 | | 授出 | 63,349,900 | 0.040 | 0.39 | | 已没收 | (24,207,555) | 0.033 | 0.32 | | 于2024年12月31日尚未行使 | 205,466,243 | 0.016 | 0.20 | | 授出 | 215,386,571 | 0.038 | 0.54 | | 已没收 | (51,306,660) | 0.023 | 0.32 | | 于2025年9月30日尚未行使 | 369,546,154 | 0.028 | 0.39 |

| | 加权平均剩余合约期限（年） | |---|---| | 于2022年12月31日尚未行使 | 9.34 | | 于2023年12月31日尚未行使 | 8.86 | | 于2024年12月31日尚未行使 | 8.25 | | 于2025年9月30日尚未行使 | 8.79 |

相关期间内概无购股权获行使。

相关期间内，集团分别确认股份支付费用1,069,000美元、3,346,000美元、6,823,000美元及8,581,000美元。

相关期间及截至2024年9月30日及2025年9月30日止九个月内授出购股权的公平价值，乃于授出日期采用二项式模型并考虑购股权授出条款及条件估算。下表列示模型所使用的输入值：

| | 截至12月31日止年度 | | | 截至9月30日止九个月 | | |---|---|---|---|---|---| | | 2022年 | 2023年 | 2024年 | 2024年（未经审核） | 2025年 | | 无风险利率（%） | 2.35%-3.83% | 3.47%-4.58% | 3.79%-4.57% | 3.79%-4.39% | 4.15%-4.23% | | 预期波动率（%） | 57.2%-58.0% | 58.5%-60.5% | 60.5%-61.1% | 60.5%-60.9% | 60.3%-64.1% | | 预期期限（年） | 10 | 10 | 10 | 10 | 10 |

27.

综合现金流量表附注

(a) 主要非现金交易

于截至2022年、2023年及2024年12月31日止年度及截至2024年及2025年9月30日止九个月，集团就楼宇租赁安排分别新增使用权资产及租赁负债640,000美元、3,718,000美元、1,249,000美元、1,205,000美元（未经审核）及1,815,000美元（均为非现金性质）。

– I-54 –

附录一

会计师报告

(b) 融资活动所产生负债的变动

| | 计息银行及其他借款（千美元） | 租赁负债（千美元） | 可换股债券（千美元） | 可换股可赎回优先股（千美元） | |---|---|---|---|---|

At 1 January 2022 (unaudited) Changes from financing cash flows Changes in fair value New leases Disposal as a result of early cancellation of lease Interest expense (note 6)

| | Interest-bearing bank and other borrowings USD'000 | Lease liabilities USD'000 | Convertible bonds USD'000 | Convertible redeemable preferred shares USD'000 | |---|---|---|---|---| | | – | – | – | 34,666 | | | – | (214) | – | 50,000 | | | – | – | – | 60,509 | | | – | 640 | – | – | | | – | – | – | – | | | – | 14 | – | – | | At 31 December 2022 | – | 440 | – | 145,175 |

| | Interest-bearing bank and other borrowings USD'000 | Lease liabilities USD'000 | Convertible bonds USD'000 | Convertible redeemable preferred shares USD'000 | |---|---|---|---|---| | At 1 January 2023 | – | 440 | – | 145,175 | | Changes from financing cash flows | – | (757) | – | 307,000 | | Changes in fair value | – | – | – | 176,826 | | New leases | – | 3,718 | – | – | | Disposal as a result of early cancellation of lease | – | – | (302) | – | | Interest expense (note 6) | – | – | 61 | – | | At 31 December 2023 | – | 3,160 | – | 629,001 |

| | Interest-bearing bank and other borrowings USD'000 | Lease liabilities USD'000 | Convertible bonds USD'000 | Convertible redeemable preferred shares USD'000 | |---|---|---|---|---| | At 1 January 2024 | – | 3,160 | – | 629,001 | | Changes from financing cash flows | 19,100 | (1,506) | 13,910 | 739,588 | | Changes in fair value | – | – | 812 | 213,360 | | New leases | – | 1,249 | – | – | | Disposal as a result of early cancellation of lease | – | 355 | (34) | – | | Interest expense (note 6) | – | (34) | 154 | – | | At 31 December 2024 | 19,455 | 3,023 | 14,722 | 1,581,949 |

| | Interest-bearing bank and other borrowings USD'000 | Lease liabilities USD'000 | Convertible bonds USD'000 | Convertible redeemable preferred shares USD'000 | |---|---|---|---|---| | At 1 January 2024 | – | 3,160 | – | 629,001 | | Changes from financing cash flows (unaudited) | 19,759 | (1,214) | 13,910 | 686,372 | | Changes in fair value (unaudited) | – | – | 348 | 127,715 | | New leases (unaudited) | – | 1,205 | – | – | | Disposal as a result of early cancellation of lease (unaudited) | – | (36) | – | – | | Interest expense (note 6) (unaudited) | 199 | 117 | – | – | | At 30 September 2024 (unaudited) | 19,958 | 3,232 | 14,258 | 1,443,088 |

– I-55 –

APPENDIX I ACCOUNTANT'S REPORT

(c)

| | Interest-bearing bank and other borrowings USD'000 | Lease liabilities USD'000 | Convertible bonds USD'000 | Accrued Listing Expense included in other payables USD'000 | Convertible redeemable preferred shares USD'000 | |---|---|---|---|---|---| | At 1 January 2025 | 19,455 | 3,023 | 14,722 | – | 1,581,949 | | Changes from financing cash flows | (757) | (1,471) | (14,668) | (357) | 426,262 | | Changes in fair value | – | – | 495 | – | 312,982 | | New leases | – | 1,815 | – | – | – | | Addition | – | – | – | 422 | – | | Disposal as a result of early cancellation of lease | – | (843) | – | – | – | | Gain on disposal | – | – | (549) | – | – | | Interest expense (note 6) | 404 | 107 | – | – | – | | At 30 September 2025 | 19,102 | 2,631 | – | 65 | 2,321,193 |

Total cash outflow for leases The total cash outflow for leases included in the statement of cash flows is as follows:

| | Year ended 31 December | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | Within operating activities | (42) | (76) | – | – | (7) | | Within financing activities | (214) | (757) | (1,506) | (1,214) | (1,471) | | Total | (256) | (833) | (1,506) | (1,214) | (1,478) |

28. CONTINGENT LIABILITIES

As at 30 September 2025, certain subsidiaries of the Group are respondents in several legal dispute cases in relation to claims of alleged infringement of intellectual property rights. While these cases are still at an early stage and the outcome cannot be estimated with certainty, the directors of the Company, having given due consideration to the legal advice and the relevant facts and circumstances, are of the opinion that no provision has been made in respect of those cases as at 30 September 2025.

29. RELATED PARTY TRANSACTIONS

Parties are considered to be related if one party has the ability, directly or indirectly, to control the other party or exercise significant influence over the other party in making financial and operational decisions. Parties are also considered to be related if they are subject to common control.

The following significant transactions were carried out between the Group and its related party during the periods presented. In the opinion of the directors of the Company, the related party transactions were carried out in the normal course of business and on terms negotiated between the Group and the respective related parties.

(a) Name and relationship

| Name of related party | Relationship with the Company | |---|---| | Shanghai Jizhi Wujie Technology Co., Ltd. ("上海極智無界科技有限公司") | Entity controlled by Founder | | Shanghai Jizhi Zongheng Technology Co., Ltd. ("上海極智縱橫科技有限公司") | Entity controlled by Founder | | Alibaba and its affiliates ("Alibaba Group") | A shareholder of the Company |

– I-56 –

APPENDIX I ACCOUNTANT'S REPORT

(b) Transactions with related parties

| | Year ended December 31 | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 USD'000 | 2023 USD'000 | 2024 USD'000 | 2024 USD'000 (unaudited) | 2025 USD'000 | | **Purchase of services** | | | | | | | Alibaba Group | 42 | 2,936 | 9,372 | 6,487 | 54,937 | | Shanghai Jizhi Wujie Technology Co., Ltd. ("上海極智無界科技有限公司") | – | – | – | – | 5 | | Shanghai Jizhi Zongheng Technology Co., Ltd. ("上海極智縱橫科技有限公司") | – | – | – | – | 21 | | **Sales of services** | | | | | | | Alibaba Group | – | 39 | 31 | 31 | 220 | | **Other transaction** | | | | | | | Shanghai Jizhi Wujie Technology Co., Ltd. ("上海極智無界科技有限公司") | – | – | – | – | 287 | | Shanghai Jizhi Zongheng Technology Co., Ltd. ("上海極智縱橫科技有限公司") | – | – | – | – | 287 |

Total ……………………………………………………… 1,596,671 99,240 1,695,911

At 30 September 2025 Financial assets

Financial assets at fair value through profit or loss

Financial assets at fair value through other comprehensive income

Financial assets at amortised cost

Total

USD'000

USD'000

USD'000

USD'000

Trade receivables ……………………………… Financial assets included in prepayments, other receivables and other assets ………… Financial assets at fair value through other comprehensive income…………………… Financial assets at fair value through profit or loss ……………………………………… Restricted cash ………………………………… Financial assets at amortised cost ………… Time deposits …………………………………… Cash and cash equivalents ………………… Total ……………………………………………

–

–

35,701

35,701

–

–

21,048

21,048

–

9,471

–

9,471

1,019,645 – – – –

– – – – –

– 14,462 388,805 159,339 712,962

1,019,645 14,462 388,805 159,339 712,962

1,019,645

9,471

1,332,317

2,361,433

Financial liabilities Financial liabilities at fair value through profit or loss

Financial liabilities at amortised cost

Total

USD'000

USD'000

USD'000

Interest-bearing bank and other borrowings ……… Trade and bills payables …………………………… Financial liabilities included in other payables and accruals ………………………………………………… Lease liabilities ………………………………………

– –

57,261 117,047

57,261 117,047

47,952 –

131,546 6,137

179,498 6,137

Total …………………………………………………… 47,952 311,991 359,943

– I-60 –

APPENDIX I ACCOUNTANT'S REPORT

The transactions with related parties were made according to the published prices and conditions offered to the non-related parties of the Company.

(c) Outstanding balances with related parties

| | As at 31 December | | | As at 30 September | |---|---|---|---|---| | | 2022 | 2023 | 2024 | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 |

Alibaba Group | Trade receivables | – | 41 | 41 | 113 | | Trade and bills payables | 10 | 592 | 4,022 | 26,509 |

Shanghai Jizhi Wujie Technology Co., Ltd. ("上海極智無界科技有限公司") | Prepayments, other receivables and other assets | – | – | – | 307 |

Shanghai Jizhi Zongheng Technology Co., Ltd. ("上海極智縱橫科技有限公司") | Prepayments, other receivables and other assets | – | – | – | 309 |

As at the end of each of the Relevant Periods, the Group's outstanding balances with related parties were all unsecured, interest-free and of trade nature.

– I-57 –

APPENDIX I ACCOUNTANT'S REPORT

(d) Compensation of key management personnel of the Group:

| | Year ended December 31 | | | Nine months ended 30 September | | |---|---|---|---|---|---| | | 2022 | 2023 | 2024 | 2024 (unaudited) | 2025 | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | | Short term employee benefits | 489 | 872 | 1,314 | 832 | 1,011 | | Post-employment benefits | 9 | 24 | 39 | 31 | 23 | | Equity-settled share option expense | 963 | 1,278 | 2,275 | 1,658 | 1,243 | | Total | 1,461 | 2,174 | 3,628 | 2,521 | 2,277 |

30. FINANCIAL INSTRUMENTS BY CATEGORY

The carrying amounts of each of the categories of financial instruments as at the end of each of the Relevant Periods are as follows:

At 31 December 2022 Financial assets

| | Financial assets at fair value through profit or loss | Financial assets at amortised cost | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Financial assets at fair value through profit or loss | 65,791 | – | 65,791 | | Financial assets included in prepayments, other receivables and other assets | – | 126 | 126 | | Restricted cash | – | 2,221 | 2,221 | | Cash and cash equivalents | – | 4,691 | 4,691 | | Total | 65,791 | 7,038 | 72,829 |

Financial liabilities

| | Financial liabilities at fair value through profit or loss | Financial liabilities at amortised cost | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Trade and bills payables | – | 2,394 | 2,394 | | Financial liabilities included in other payables, accruals and other liabilities | – | 268 | 268 | | Convertible redeemable preferred shares | 145,175 | – | 145,175 | | Lease liabilities | – | 440 | 440 | | Total | 145,175 | 3,102 | 148,277 |

– I-58 –

APPENDIX I ACCOUNTANT'S REPORT

At 31 December 2023 Financial assets

| | Financial assets at fair value through profit or loss | Financial assets at amortised cost | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Trade receivables | – | 1,338 | 1,338 | | Financial assets included in prepayments, other receivables and other assets | – | 840 | 840 | | Financial assets at fair value through profit or loss | 15,802 | – | 15,802 | | Restricted cash | – | 39 | 39 | | Time deposits | – | 91,698 | 91,698 | | Cash and cash equivalents | – | 206,295 | 206,295 | | Total | 15,802 | 300,210 | 316,012 |

Financial liabilities

| | Financial liabilities at fair value through profit or loss | Financial liabilities at amortised cost | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Trade and bills payables | – | 17,242 | 17,242 | | Financial liabilities included in other payables and accruals | – | 8,969 | 8,969 | | Convertible redeemable preferred shares | 629,001 | – | 629,001 | | Lease liabilities | – | 3,160 | 3,160 | | Total | 629,001 | 29,371 | 658,372 |

At 31 December 2024 Financial assets

| | Financial assets at fair value through profit or loss | Financial assets at fair value through other comprehensive income | Financial assets at amortised cost | Total | |---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | | Trade receivables | – | – | 6,982 | 6,982 | | Financial assets included in prepayments, other receivables and other assets | – | – | 6,493 | 6,493 | | Financial assets at fair value through other comprehensive income | – | 4,836 | – | 4,836 | | Financial assets at fair value through profit or loss | 390,551 | – | – | 390,551 | | Restricted cash | – | – | 27,331 | 27,331 | | Financial assets at amortised cost | – | – | 147,444 | 147,444 | | Time deposits | – | – | 26,327 | 26,327 | | Cash and cash equivalents | – | – | 288,912 | 288,912 | | Total | 390,551 | 4,836 | 503,489 | 898,876 |

– I-59 –

APPENDIX I ACCOUNTANT'S REPORT

Financial liabilities

| | Financial liabilities at fair value through profit or loss | Financial liabilities at amortised cost | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Interest-bearing bank and other borrowings | – | 19,455 | 19,455 | | Trade and bills payables | – | 51,212 | 51,212 | | Financial liabilities included in other payables and accruals | 14,722 | 25,550 | 40,272 | | Convertible redeemable preferred shares | 1,581,949 | – | 1,581,949 | | Lease liabilities | – | 3,023 | 3,023 | | Total | 1,596,671 | 99,240 | 1,695,911 |

At 30 September 2025 Financial assets

| | Financial assets at fair value through profit or loss | Financial assets at fair value through other comprehensive income | Financial assets at amortised cost | Total | |---|---|---|---|---| | | USD'000 | USD'000 | USD'000 | USD'000 | | Trade receivables | – | – | 35,701 | 35,701 | | Financial assets included in prepayments, other receivables and other assets | – | – | 21,048 | 21,048 | | Financial assets at fair value through other comprehensive income | – | 9,471 | – | 9,471 | | Financial assets at fair value through profit or loss | 1,019,645 | – | – | 1,019,645 | | Restricted cash | – | – | 14,462 | 14,462 | | Financial assets at amortised cost | – | – | 388,805 | 388,805 | | Time deposits | – | – | 159,339 | 159,339 | | Cash and cash equivalents | – | – | 712,962 | 712,962 | | Total | 1,019,645 | 9,471 | 1,332,317 | 2,361,433 |

Financial liabilities

| | Financial liabilities at fair value through profit or loss | Financial liabilities at amortised cost | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Interest-bearing bank and other borrowings | – | 57,261 | 57,261 | | Trade and bills payables | – | 117,047 | 117,047 | | Financial liabilities included in other payables and accruals | 47,952 | 131,546 | 179,498 | | Lease liabilities | – | 6,137 | 6,137 | | Total | 47,952 | 311,991 | 359,943 |

– I-60 –

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

6,440

–

–

6,440

–

714,382

–

714,382

Liabilities measured at fair value: As at 31 December 2022 Fair value measurement using

Convertible redeemable preferred sharesў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

–

–

894,265

894,265

As at 31 December 2023 Fair value measurement using

Convertible redeemable preferred sharesў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

–

–

1,596,671

1,596,671

As at 31 December 2024 Fair value measurement using

Convertible redeemable preferred sharesў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

–

–

2,321,193

2,321,193

– I-62 –

APPENDIX I

ACCOUNTANT'S REPORT

At 30 September 2025 Fair value measurement using

Convertible redeemable preferred sharesў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

–

–

2,321,193

2,321,193

During the Relevant Periods, there were no transfers of fair value measurements between Level 1 and Level 2 and no transfers into or out of Level 3 fair value measurements.

Reconciliation of fair value measurements categorised within Level 3 of the fair value hierarchy:

Convertible redeemable preferred shares

USD'000

At 1 January 2022 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Issuance of convertible redeemable preferred shares ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Changes in fair value recognised in profit or loss ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

558,051 165,360 170,854

At 31 December 2022 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Issuance of convertible redeemable preferred shares ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Changes in fair value recognised in profit or loss ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

894,265 399,692 302,714

At 31 December 2023 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Changes in fair value recognised in profit or loss ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

1,596,671 724,522

At 31 December 2024 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў Changes in fair value recognised in profit or loss ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

2,321,193 –

At 30 September 2025 ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

2,321,193

The convertible redeemable preferred shares were converted into ordinary shares upon the Listing. Further details are set out in note 24 to the Historical Financial Information.

Below is a summary of the valuation techniques and significant unobservable inputs used in the Level 3 fair value measurement of convertible redeemable preferred shares during the Relevant Periods:

| Valuation technique | Significant unobservable inputs | Range (weighted average) | |---|---|---| | Back-solve method and equity allocation model | Equity value of the Company | USD558,051,000–USD2,321,193,000 | | | Discount for lack of marketability | 5.0%–15.0% | | | Risk-free rate | 2.1%–4.8% | | | Volatility | 60.0%–75.0% | | | Time to liquidation | 0.5–3.0 years |

– I-63 –

APPENDIX I

ACCOUNTANT'S REPORT

The significant unobservable input used in the fair value measurement of convertible redeemable preferred shares is the equity value of the Company, discount for lack of marketability, risk-free rate, volatility, and time to liquidation. Significant increases/decreases in the equity value of the Company in isolation would result in a significantly higher/lower fair value measurement. Significant increases/decreases in the discount for lack of marketability in isolation would result in a significantly lower/higher fair value measurement.

32.

FINANCIAL RISK MANAGEMENT OBJECTIVES AND POLICIES

The Group's principal financial instruments include interest-bearing bank and other borrowings, convertible redeemable preferred shares, financial assets at fair value through profit or loss, financial assets at fair value through other comprehensive income, restricted cash and cash and cash equivalents. The main purpose of these financial instruments is to raise finance for the Group's operations. The Group has various other financial assets and liabilities such as trade receivables and trade payables, which arise directly from its operations.

The main risks arising from the Group's financial instruments are market risk (including interest rate risk and currency risk), credit risk and liquidity risk. The board of directors reviews and agrees on the policies for managing each of these risks as set out below.

Interest rate risk

The Group's exposure to the risk of changes in market interest rates relates primarily to the Group's long-term debt obligations with floating interest rates.

The following table demonstrates the sensitivity to a reasonably possible change in interest rates, with all other variables held constant, of the Group's profit before tax (through the impact on floating rate borrowings).

| | Increase/decrease in basis points | Effect on profit before tax USD'000 | |---|---|---| | Year ended 31 December 2022 | +50 | (72) | | | -50 | 72 | | Year ended 31 December 2023 | +50 | (187) | | | -50 | 187 | | Year ended 31 December 2024 | +50 | (130) | | | -50 | 130 | | Nine months ended 30 September 2024 | +50 | (97) | | | -50 | 97 | | Nine months ended 30 September 2025 | +50 | (79) | | | -50 | 79 |

– I-64 –

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

6,440

–

–

6,440

–

714,382

–

714,382

Liabilities measured at fair value: As at 31 December 2022 Fair value measurement using Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

Convertible redeemable preferred shares

–

–

145,175

145,175

As at 31 December 2023 Fair value measurement using

Convertible redeemable preferred shares

Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

–

– I-62 –

–

629,001

629,001

APPENDIX I

ACCOUNTANT'S REPORT

As at 31 December 2024 Fair value measurement using Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

Financial liabilities included in other payables and accruals

–

–

14,722

14,722

Convertible redeemable preferred shares

–

–

1,581,949

1,581,949

As at 30 September 2025 Fair value measurement using

Convertible redeemable preferred shares

Quoted prices in active markets (Level 1)

Significant observable inputs (Level 2)

Significant unobservable inputs (Level 3)

Total

USD'000

USD'000

USD'000

USD'000

–

–

2,321,193

2,321,193

During the Relevant Periods, there were no transfers of fair value measurements between Level 1 and Level 2 and no transfers into or out of Level 3 for both financial assets and financial liabilities.

32. FINANCIAL RISK MANAGEMENT OBJECTIVES AND POLICIES

The Group's principal financial instruments comprise bank loans and overdrafts, convertible bonds, convertible redeemable preferred shares, financial assets at fair value through profit or loss, and cash and short-term deposits. The main purpose of these financial instruments is to raise finance for the Group's operations. The Group has various other financial assets and liabilities such as trade receivables and trade and bills payables, which arise directly from its operations.

The main risks arising from the Group's financial instruments are foreign currency risk, credit risk and liquidity risk. The senior management of the Company meets regularly to analyse and formulate measures to manage the Group's exposure to these risks. In addition, the Board holds meetings regularly to analyse and approve the proposals made by the senior management of the Company. Generally, the Group introduces conservative strategies on its risk management. As the Group's exposure to these risks is kept to a minimum, the Group has not used any derivatives and other instruments for hedging purposes. The Group does not hold or issue derivative financial instruments for trading purposes. The Board reviews and agrees policies for managing each of these risks and they are summarised below.

Foreign currency risk

Foreign currency risk is the risk of loss resulting from changes in foreign currency exchange rates. Such exposures arise from sales and purchases by operating units in currencies other than the units' functional currencies. The Group seeks to limit its exposure to foreign currency risk by minimising its net foreign currency position.

– I-63 –

APPENDIX I

ACCOUNTANT'S REPORT

The following table demonstrates the sensitivity at the end of each of the Relevant Periods to a reasonably possible change in foreign currency exchange rates, with all other variables held constant, of the Group's loss before tax (due to changes in the fair value of monetary assets and liabilities) and the Group's deficits.

| | Increase/(decrease) in RMB rate | Increase/(decrease) in loss before tax | Increase/(decrease) in deficits | |---|---|---|---| | | % | USD'000 | USD'000 | | **2022** | | | | | If the RMB weakens against the USD | (5) | 73 | 73 | | If the RMB strengthens against the USD | 5 | (73) | (73) |

| | Increase/(decrease) in RMB rate | Increase/(decrease) in loss before tax | Increase/(decrease) in deficits | |---|---|---|---| | | % | USD'000 | USD'000 | | **2023** | | | | | If the RMB weakens against the USD | (5) | (43) | (43) | | If the RMB strengthens against the USD | 5 | 43 | 43 |

| | Increase/(decrease) in RMB rate | Increase/(decrease) in loss before tax | Increase/(decrease) in deficits | |---|---|---|---| | | % | USD'000 | USD'000 | | **2024** | | | | | If the RMB weakens against the USD | (5) | (5,295) | (5,295) | | If the RMB strengthens against the USD | 5 | 5,295 | 5,295 |

| | Increase/(decrease) in RMB rate | Increase/(decrease) in loss before tax | Increase/(decrease) in deficits | |---|---|---|---| | | % | USD'000 | USD'000 | | **30 September 2025** | | | | | If the RMB weakens against the USD | (5) | (3,241) | (3,241) | | If the RMB strengthens against the USD | 5 | 3,241 | 3,241 |

Credit risk

The Group trades only with recognised and creditworthy third parties. It is the Group's policy that all customers who wish to trade on credit terms are subject to credit verification procedures. In addition, receivable balances are monitored on an ongoing basis and the Group's exposure to bad debts is not significant.

Maximum exposure and year-end staging

The table below shows the credit quality and the maximum exposure to credit risk based on the Group's credit policy, which is mainly based on reasonable and supportable information that is available at the reporting date about past events, current conditions and forecasts of future economic conditions, and year-end staging classification as at the end of each of the Relevant Periods. The amounts presented are gross amounts for financial assets.

| 31 December 2022 | 12-months ECLs | Lifetime ECLs | | | Total | |---|---|---|---|---|---| | | Stage 1 | Stage 2 | Stage 3 | Simplified approach | | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 |

APPENDIX I

ACCOUNTANT'S REPORT

Financial assets included in prepayments and other receivables and other assets – Normal** ў ў ў ў ў ў ў ў ў ў ў ў Restricted cash ў ў ў ў ў ў ў ў ў ў ў ў Cash and cash equivalents ў ў ў ў ў

126 2,221 4,691

– – –

– – –

– – –

126 2,221 4,691

Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў

7,038

–

–

–

7,038

– I-64 –

APPENDIX I

ACCOUNTANT'S REPORT

31 December 2023 | | 12-months ECLs | Lifetime ECLs | | | | |---|---|---|---|---|---| | | Stage 1 | Stage 2 | Stage 3 | Simplified approach | Total | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | | Trade receivables* ў ў ў ў ў ў ў ў ў | – | – | – | 1,338 | 1,338 | | Financial assets included in prepayments and other receivables and other assets – Normal** ў ў ў ў ў ў ў ў ў ў ў ў | 840 | – | – | – | 840 | | Restricted cash ў ў ў ў ў ў ў ў ў ў ў ў | 39 | – | – | – | 39 | | Time deposits ў ў ў ў ў ў ў ў ў ў ў ў | 91,698 | – | – | – | 91,698 | | Cash and cash equivalents ў ў ў ў ў | 206,295 | – | – | – | 206,295 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 298,872 | – | – | 1,338 | 300,210 |

31 December 2024 | | 12-months ECLs | Lifetime ECLs | | | | |---|---|---|---|---|---| | | Stage 1 | Stage 2 | Stage 3 | Simplified approach | Total | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | | Trade receivables* ў ў ў ў ў ў ў ў ў | – | – | – | 6,982 | 6,982 | | Financial assets included in prepayments, other receivables and other assets – Normal** ў ў ў ў ў ў ў ў ў ў ў ў | 6,493 | – | – | – | 6,493 | | Financial assets at amortised cost ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 147,444 | – | – | – | 147,444 | | Restricted cash ў ў ў ў ў ў ў ў ў ў ў ў | 27,331 | – | – | – | 27,331 | | Time deposits ў ў ў ў ў ў ў ў ў ў ў ў | 26,327 | – | – | – | 26,327 | | Cash and cash equivalents ў ў ў ў ў | 288,912 | – | – | – | 288,912 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 496,507 | – | – | 6,982 | 503,489 |

30 September 2025 | | 12-months ECLs | Lifetime ECLs | | | | |---|---|---|---|---|---| | | Stage 1 | Stage 2 | Stage 3 | Simplified approach | Total | | | USD'000 | USD'000 | USD'000 | USD'000 | USD'000 | | Trade receivables* ў ў ў ў ў ў ў ў ў | – | – | – | 8,063 | 8,063 | | Financial assets included in prepayments, other receivables and other assets – Normal** ў ў ў ў ў ў ў ў ў ў ў ў | 10,169 | – | – | – | 10,169 | | Financial assets at amortised cost ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | – | – | – | – | – | | Restricted cash ў ў ў ў ў ў ў ў ў ў ў ў | 25,138 | – | – | – | 25,138 | | Cash and cash equivalents ў ў ў ў ў | 362,647 | – | – | – | 362,647 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 397,954 | – | – | 8,063 | 406,017 |

– I-65 –

APPENDIX I

ACCOUNTANT'S REPORT

*

For trade receivables to which the Group applies the simplified approach for impairment, information based on the provision is disclosed in note 15 to the Historical Financial Information.

**

The credit quality of the financial assets included in prepayments, other receivables and other assets is considered to be "normal" when they are not past due and there is no information indicating that the financial assets had a significant increase in credit risk since initial recognition. Otherwise, the credit quality of the financial assets is considered to be "doubtful".

Further quantitative data in respect of the Group's exposure to credit risk arising from trade receivables and prepayments, other receivables and other assets are respectively disclosed in notes 15 and 16 to the Historical Financial Information.

Liquidity risk

The Group monitors its exposure to liquidity risk by monitoring the current ratio, which is calculated by comparing the current assets with the current liabilities.

The liquidity of the Group is primarily dependent on its ability to maintain adequate cash inflows from operations to meet its debt obligations as they fall due, and its ability to obtain external financing to meet its committed future capital expenditure.

The maturity profile of the Group's financial liabilities as at the end of each of the Relevant Periods, based on the contractual undiscounted payments, is as follows:

As at 31 December 2022 | | Within 1 year | 1 to 5 years | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Lease liabilities ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 349 | 92 | 441 | | Trade and bills payables ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 2,394 | – | 2,394 | | Financial liabilities included in other payables and accruals ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 268 | – | 268 | | Convertible redeemable preferred shares ў ў ў ў ў ў | 85,284 | – | 85,284 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 88,295 | 92 | 88,387 |

As at 31 December 2023 | | Within 1 year | 1 to 5 years | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Lease liabilities ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 1,248 | 1,990 | 3,238 | | Trade and bills payables ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 17,242 | – | 17,242 | | Financial liabilities included in other payables and accruals ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 8,969 | – | 8,969 | | Convertible redeemable preferred shares ў ў ў ў ў ў | 410,775 | – | 410,775 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 438,234 | 1,990 | 440,224 |

– I-66 –

APPENDIX I

ACCOUNTANT'S REPORT

As at 31 December 2024 | | Within 1 year | 1 to 5 years | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Lease liabilities ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 1,964 | 1,086 | 3,050 | | Interest-bearing bank and other borrowings ў ў ў ў | 19,455 | – | 19,455 | | Trade and bills payables ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 51,212 | – | 51,212 | | Financial liabilities included in other payables and accruals ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 40,272 | – | 40,272 | | Convertible redeemable preferred shares ў ў ў ў ў ў | 1,215,508 | – | 1,215,508 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 1,328,411 | 1,086 | 1,329,497 |

As at 30 September 2025 | | Within 1 year | 1 to 5 years | Total | |---|---|---|---| | | USD'000 | USD'000 | USD'000 | | Lease liabilities ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 1,694 | 1,039 | 2,733 | | Interest-bearing bank and other borrowings ў ў ў ў | 19,102 | – | 19,102 | | Trade and bills payables ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 70,219 | – | 70,219 | | Financial liabilities included in other payables and accruals ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 7,061 | – | 7,061 | | Convertible redeemable preferred shares ў ў ў ў ў ў | 1,698,247 | – | 1,698,247 | | Total ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў ў | 1,796,323 | 1,039 | 1,797,362 |

APPENDIX II