While domestic large model companies are seizing global market share, managing infringement risks remains a critical challenge. Following the listing of "the first global large model stock" Zhipu (2513.HK) on the Hong Kong Stock Exchange on January 8, 2026, another major model company, MiniMax (0100.HK), debuted on January 9. On January 9, MiniMax opened at HK$235.4 per share, a surge of 42.7% compared to its issue price of HK$165. By the market close, its share price reached HK$345, marking a remarkable 109.09% increase and propelling its total market capitalization to HK$106.6 billion, surpassing Zhipu's HK$69.8 billion market cap on the same day. MiniMax issued 29.1976 million shares in this offering, raising HK$4.818 billion, which also exceeded the HK$4.348 billion raised by Zhipu. By January 13, MiniMax's stock continued its ascent, closing up 15.36% for the day and pushing its total market value to HK$123.1 billion, approximately equivalent to 110.1 billion yuan. The successful listing delivered substantial gains for MiniMax's prestigious investor lineup. Entities including Alibaba's Alisoft China, Tencent's Image Framework, and IDG held stakes prior to the IPO. Specifically, as of the listing on January 9, Alibaba, via Alisoft China, held a 12.52% equity stake, making it the company's second-largest shareholder. Mingshi Capital, the most frequent participant across MiniMax's funding rounds, saw this as a major success. Huang Mingming, founding partner of Mingshi Capital, stated, "MiniMax's bell-ringing in Hong Kong represents a 'home run' level investment return for Mingshi. The immense surprise stems not merely from the return itself, but from our profound realization that the disruptive changes brought by AI are significantly underestimated by the wider market." According to post-hearing documents, the funds raised by MiniMax are allocated for R&D over the next five years, encompassing the development of its large model and AI-native products, alongside working capital and general corporate purposes. In recent years, fueled by advancements in large model intelligence and growth among individual users, developers, and enterprise clients, MiniMax has experienced rapid revenue expansion. From 2023 to 2024, its total revenue jumped from $3.46 million to $30.523 million. In the first three quarters of 2025, revenue further increased to $53.437 million. Estimates based on market share suggest the company's 2025 revenue reached approximately $66 million. However, substantial R&D investments have kept MiniMax in a loss-making position. From 2022 to 2024, its adjusted net loss (non-IFRS) widened from $12 million to $244 million. The company anticipates its net loss for 2025 will increase significantly. Rapid growth in its C-end user base has also exposed MiniMax to risks. In September 2025, the company faced lawsuits from several major US film studios, including Disney and Universal Studios, over alleged copyright infringement. Post-hearing materials indicate that, under a worst-case scenario where plaintiffs prevail completely, potential statutory damages could reach $75 million. Additionally, several executive directors resigned in 2024 due to business arrangements. MiniMax had not responded to related inquiries by the time of publication.
Over 70% of Revenue Comes from Individuals As a global AI large model company, MiniMax has seen its revenue grow swiftly in recent years. In 2024, MiniMax's total revenue surged 782% year-on-year to $30.523 million (approximately 214 million yuan); meanwhile, Zhipu's 2024 revenue grew 151% to 312 million yuan, representing a larger revenue base than MiniMax during the same period. For the first three quarters of 2025, MiniMax's total revenue increased 175% year-on-year to $53.437 million. MiniMax's revenue primarily stems from two streams: AI-native products for individual users, and its open platform alongside other AI-based enterprise services. Revenue generation began in 2023, initially dominated by the open platform and enterprise services, which accounted for over 70% that year. From 2024 through the first three quarters of 2025, AI-native products became the primary revenue source, contributing over 70% of total income. Unlike MiniMax's current reliance on individual customers, Zhipu's revenue is predominantly institutional. It provides large model services via its MaaS platform to clients including private enterprises, public sector entities, and individuals (end-users and independent developers), with institutions forming the core. Its revenue channels are cloud deployment and on-premise deployment, the latter being the main source, consistently contributing over 80% of revenue from 2022 to H1 2025. In 2024, the top five verticals for on-premise deployment revenue were internet & technology, public services, telecommunications, traditional enterprises, and consumer electronics. "What ultimately convinced us about MiniMax was its pathway from technology to engineering to productization, coupled with Yan Junjie's (Founder and Board Chairman of MiniMax) profound insights and pragmatic approach," Huang Mingming remarked. "From our earliest discussions, Yan Junjie emphasized his vision to create an intelligent agent with voice, image, and dialogue capabilities, using a single model adaptable to all scenarios, while pursuing commercialization for both C-end and B-end," Huang added. To achieve this, Yan believed breakthroughs were needed in three core areas: multimodal fusion, low-cost training, and global implementation; subsequent C-end and B-end product launches were based on this foundational vision. Revenue from MiniMax's AI-native products includes subscriptions from applications like the MiniMax intelligent Agent app, Conch AI (visual generation platform), MiniMax Voice (audio generation tool), and Talkie/Starfield (full-modal interaction platform). The open platform and other AI-based enterprise services generate revenue by providing enterprise clients and developers with access to usage-based billing for these platforms. Regionally, over 80% of Zhipu's revenue originates from mainland China. In contrast, MiniMax's 2024 revenue was more globally distributed: 30.2% from mainland China, 37.5% from Singapore, and 16.4% from the United States. Globally, Chinese large model technology companies still hold a relatively low market share. The large model sector comprises technology companies and application companies. According to China Insights Consultancy data, based on 2024 global model-based revenue (primarily from subscription services and API calls/licensing), MiniMax ranked as the tenth largest large model technology company globally, with a 0.3% market share. The top eight companies in the 2024 global ranking were all US-based, collectively holding about 66% market share. The global large model market is projected to reach $22 billion in 2025, with MiniMax expected to capture approximately 0.3%. Based on this projection, MiniMax's 2025 revenue is estimated around $66 million, representing a roughly 116% increase compared to its 2024 revenue of $30.523 million. In Huang Mingming's view, "The recognition of MiniMax by global mainstream long-term investors signifies acknowledgment of Chinese AI companies' capabilities, charting a viable path for Chinese large model firms in global competition. We firmly believe more severely undervalued Chinese tech enterprises will similarly shine on the global stage."
Sustained Losses Amid High R&D Despite revenue growth, MiniMax's losses have also escalated. From 2022 to 2024, MiniMax's adjusted net losses (non-IFRS) were $12 million, $89 million, and $244 million respectively, totaling $345 million (approximately 2.4 billion yuan). In the first three quarters of 2025, its unaudited net loss was $186 million, a 9% increase year-on-year. Another large model firm, Zhipu, reported cumulative adjusted losses of about 3.2 billion yuan from 2022 to 2024, higher than MiniMax's losses for the same period. In H1 2025, Zhipu's losses were approximately 1.8 billion yuan, up about 70% year-on-year. High R&D expenditure is a primary driver of these ongoing losses. From 2022 to 2024, MiniMax's R&D spending was $11 million, $70 million, and $189 million respectively, summing to approximately $270 million (about 1.89 billion yuan). In the first three quarters of 2025, R&D开支 reached $180 million (around 1.26 billion yuan), a 30% year-on-year increase. Comparatively, Zhipu's cumulative R&D开支 from 2022 to 2024 were approximately 2.337 billion yuan, exceeding MiniMax's figures. In H1 2025, Zhipu's R&D开支 hit 1.595 billion yuan, surging over 80% year-on-year. Alongside high R&D, a significant increase in administrative开支, even as sales and distribution开支 fell sharply, contributed to the expanded loss in the first three quarters of 2025. During this period, administrative开支 rose about 130% to $22.074 million. Rising employee costs and share-based payment expenses were key factors behind the administrative开支 growth. Employee costs surged 169% year-on-year to $9.997 million, while share-based payment expenses increased over 50% to $2.109 million in the first three quarters of 2025. MiniMax expects its 2025 net loss to increase significantly, primarily due to anticipated R&D开支 and fair value losses on financial liabilities. Zhipu also forecasts a substantial rise in its 2025 net loss. Regarding profitability, MiniMax's gross margin trails Zhipu's. From 2023 to 2024, MiniMax's gross margins were -24.7% and 12.2% respectively. While the open platform and enterprise services maintained gross margins over 63%, negative margins from C-end AI-native products dragged down overall profitability. However, in the first three quarters of 2025, as AI-native product margins turned positive, the company's overall gross margin improved by 20.7 percentage points year-on-year to 23.3%, the highest since 2023. Zhipu's gross margins consistently exceeded 54% from 2022 to 2024, reaching 50% in H1 2025, a slight improvement year-on-year. MiniMax attributes its improving gross margin to continuous enhancements in model intelligence and significant increases in inference efficiency. "Pursuing 'extreme computing power strategies and engineering capabilities,' MiniMax achieved performance within 5% of leading US models using only 1% of their cost. Its B-end gross margin exceeding 60% is far above the industry average," Huang Mingming noted. He predicts that leveraging its strengthening foundational model advantages, MiniMax will gradually enter new market segments and capture significant share, with future business advancing equally in both C-end and B-end domains.
Lawsuit from Disney and Others Over Infringement Amid rapid performance growth, MiniMax faced lawsuits and claims from film studios like Disney over potential copyright infringement. In September 2025, several major US film studios (plaintiffs), including Disney, Universal Studios, and Warner Bros. Discovery, filed a civil lawsuit in the US District Court for the Central District of California. The suit alleges that MiniMax's visual generation platform, Conch AI, infringed upon their copyrights. The plaintiffs primarily seek actual or statutory damages, injunctive relief, attorney's fees, and other equitable remedies. The allegations include direct and indirect copyright infringement. Regarding direct infringement, plaintiffs claim MiniMax, through Conch AI, created and displayed videos and images featuring well-known movie and animation characters owned by them, arguing the company should be considered directly copying, displaying, and distributing their properties. Indirect infringement accusations encompass contributory and vicarious liability. Plaintiffs allege that even for user-generated content, MiniMax should be held liable under these principles, contending the company knew or should have known users could create content depicting plaintiffs' characters and that MiniMax benefited from such use. Information shows Conch AI fully integrates MiniMax's Hailuo-02 model and rapidly became one of the world's most popular AI image and video creation platforms through organic word-of-mouth. It enables real-time generation of high-quality images and videos via web and app interfaces. In the first three quarters of 2025, Conch AI was the company's second-largest revenue source by product segment, contributing over 30%. During this period, Conch AI had 42.35 million users, including 310,000 paying users. While these claims are commercial disputes in nature, after considering advice from its US litigation counsel, MiniMax's board believes the allegations lack factual basis in all material respects and are unsupported by sufficient evidence. Regarding direct infringement, MiniMax contends: Conch AI's content generation is user-triggered, not company action; plaintiffs are overextending their copyright protection scope; and materials used in model development qualify as "fair use." For indirect infringement, MiniMax refutes the charges because: Conch AI's design purpose and primary use are for normal, legal content; the company does not gain direct financial benefit from alleged infringements; and there is no known or intentional misconduct by the company. The potential financial impact from the plaintiffs' claims is substantial. The plaintiffs allege entitlement to statutory damages of up to $150,000 per infringed work, the maximum under US copyright law applicable only if infringement is deemed "willful." The complaint's appendix lists registration information for approximately 500 films and TV programs involved. Post-hearing materials indicate that, assuming plaintiffs fully succeed in a worst-case scenario, monetary claims could reach $75 million in statutory damages. "The vast majority of Conch AI's outputs are user-created content entirely unrelated to the plaintiffs' characters," MiniMax stated in its documents. Considering US counsel's advice, the board believes the likelihood of plaintiffs achieving complete victory, with a finding of willful infringement for 500 registered works resulting in the maximum $75 million penalty, is extremely low. Reasons include: courts are highly unlikely to recognize infringement for all ~500 identified works; and awarding the maximum $150,000 per work is highly improbable. The documents note that for non-willful infringement, statutory damages are typically capped around $30,000 per work, while the higher ~$150,000 threshold requires a finding of willfulness, a standard rarely met in similar cases. "A 2019 UCLA Law Review study examining approximately 1,000 federal copyright cases from 2005-2008 found that while about 80% of plaintiffs alleged willful infringement, courts found willfulness in only about 2% of plaintiff-favorable cases, and awarded the maximum $150,000 in merely 0.2%." Notably, this study covered cases from 2005-2008, and whether more recent statistics exist is unknown. MiniMax's directors believe the claims will not materially adversely affect the company's business, operating results, or financial position, citing reasons such as: the limited impact of worst-case damages relative to liquid resources; the low likelihood of court-ordered injunctive relief; and the limited impact of plaintiffs' IP on Conch AI's operations and commercial viability. To prevent improper or illegal user inputs, MiniMax's terms of service explicitly prohibit users from inputting illegal, non-compliant, or inappropriate content, or using its products for illegal activities or purposes. The company has also established complaint and reporting mechanisms and deployed both automated and manual review processes at input and output stages. MiniMax stated it will defend against the complaint following US civil procedure. "As the case is in its early stages, the company cannot precisely predict its timeline, outcome, potential damages, or associated costs, nor can it guarantee a favorable result."
Comments