The race for capital in China's AI sector is accelerating at a breakneck pace. On May 31st, MiniMax, which had only been listed on the Hong Kong Stock Exchange for just over 140 days, announced its board had resolved to explore a preliminary plan to issue RMB shares and list on Shanghai's STAR Market. The day prior, the China Securities Regulatory Commission's filing system revealed that MiniMax had signed a tutoring agreement with CITIC Securities on May 29th, officially launching its A-share IPO process. The move from its Hong Kong listing in January to this A-share announcement in late May spanned just over four months, underscoring the "speed" that has defined this general AI company, founded just over four years ago. On June 1st, KNOWLEDGE ATLAS (ASX: 02513) announced in Hong Kong that it would apply for a listing on the STAR Market, aiming to raise 15 billion yuan. This means the two companies, which listed in Hong Kong in January and were dubbed the "AI Giants," are set to converge again in the A-share market. However, behind the buzz lies a series of unavoidable challenges for MiniMax: a looming "flood" of share lock-up expiries in July, a nearly 430 billion Hong Kong dollar valuation gap with KNOWLEDGE ATLAS, and persistent questions about its model capabilities. Before its A-share listing, MiniMax faces far more than just the question of whether its valuation narrative remains sustainable.
Pursuing the A-Share Market: MiniMax's Capital Window
The numbers are staggering. On January 9th, MiniMax listed at HK$165 per share, closing its first day up 109% and pushing its market cap past one trillion Hong Kong dollars. Its share price then surged, hitting an intraday high of HK$1,330 on March 18th, briefly making it the highest-priced stock on the Hong Kong exchange. Even after a pullback, its share price closed at HK$708 on June 1st, a cumulative gain of over 300% from its IPO price, with a total market cap of approximately 222 billion Hong Kong dollars. From its founding in late 2021 to its Hong Kong listing, MiniMax took just over four years, setting a global record for an AI company's speed from inception to public listing. Now, just 140-plus days after going public, it is launching an A-share listing, a pace that is turning heads. Analyst Shen Meng noted that MiniMax's move to return to the A-share market aims to capitalize on the ongoing AI hype to replenish liquidity. Currently, domestic AI assets are in a phase where capital markets are most willing to assign high valuations. MiniMax's timing is driven by two factors: on one hand, such assets are notably scarce in the A-share market; on the other, the large model industry remains a capital-intensive arena requiring continuous investment in computing power, R&D, talent, data, and inference costs. For MiniMax, a Hong Kong listing does not mark the end of fundraising but rather potentially the start of a new round of capital consumption. The reasons for the A-share move are even more straightforward: to raise more funds for the large model war, to provide an additional exit for early shareholders, and to leverage the A-share platform to boost brand recognition and user scale. From the company's perspective, raising more capital while market sentiment is high is beneficial for development. However, from an investor's standpoint, Hong Kong has already assigned MiniMax a high valuation; can the A-share market offer an even higher premium? If not, what is the new growth story for its A-share listing?
The Giants' Divergence: A Nearly 3-Fold Valuation Gap
When viewed in the broader context of the Hong Kong-listed "AI Giants," the story becomes more compelling. In January, KNOWLEDGE ATLAS and MiniMax listed in Hong Kong, with KNOWLEDGE ATLAS taking the title of "first large model stock" by a day. On the path back to A-shares, KNOWLEDGE ATLAS was also a step ahead, having appointed Guotai Junan Securities and China International Capital Corporation (CICC) as tutors for its STAR Market listing in February. Overall, compared to other players like Kimi and StepFun, MiniMax and KNOWLEDGE ATLAS started on nearly the same footing in the secondary market. However, a significant gap has emerged, particularly in market capitalization. As of the June 1st close, KNOWLEDGE ATLAS had a total market cap of 650.9 billion Hong Kong dollars, while MiniMax stood at 222 billion Hong Kong dollars—a difference of 428.9 billion, with KNOWLEDGE ATLAS valued at 2.9 times that of MiniMax. Analysts widely attribute this divergence to different business models. KNOWLEDGE ATLAS leans more towards a general-purpose large model foundation and a Model-as-a-Service (MaaS) model, targeting government and enterprise clients, developer ecosystems, and model infrastructure. MiniMax emphasizes multimodal AI-native applications, attempting to build a closed loop with consumer-facing products and business services. KNOWLEDGE ATLAS fits more neatly into the market's "AI infrastructure" framework. Labels like foundational models, government/enterprise services, MaaS platforms, and domestic substitution align more closely with the hard-tech narrative familiar to A-share investors, especially in the current policy environment, which tends to reward "certainty premiums." MiniMax's story is more application-focused. Products like AI companionship, video generation, voice, and multimodal interaction naturally gain traction with consumers quickly. However, the opportunities and risks in consumer AI are two sides of the same coin: viral growth can be fleeting, user growth may outpace paid conversion, and while user experience is easily perceived, model shortcomings are equally magnified. The divergence between the two companies arguably began on March 18th, the day MiniMax released its new M2.7 model. On that day, MiniMax shares surged over 28% intraday to a record high before closing up 19.85%, while KNOWLEDGE ATLAS, without any specific positive news of its own, also closed up 19.47%. Subsequently, their share price trajectories began to diverge: MiniMax trended downward with volatility, while KNOWLEDGE ATLAS trended upward. Model capability may be a key factor in this widening gap. Some users have been critical, suggesting that while KNOWLEDGE ATLAS's models are strong, MiniMax's offerings have issues. Consequently, the market is discounting MiniMax: if it is a foundational model company, it must prove its model is top-tier; if it is an AI application company, it must prove user scale translates to high-quality revenue; if it is a multimodal platform company, it must demonstrate more than just hit apps, but also a developer ecosystem and enterprise customer stickiness.
Key Risks for Investors to Monitor
For investors, MiniMax faces several significant hurdles in the coming months. The first is the July share lock-up expiry. Estimates suggest only about 5% of MiniMax's total shares are freely tradable, with approximately 65% becoming eligible for trading in July. A large-scale unlock is expected around July 9th, involving about 63% of its Hong Kong-listed shares, with over one-third held by financial investors. MiniMax's current high valuation is partly supported by a "scarcity premium" due to its extremely low float. When the lock-up expires in July, early investors sitting on substantial profits may have a strong incentive to cash out. How the company navigates this will significantly impact its share price performance. The second hurdle is the widening gap between valuation and fundamentals. According to its 2025 financial report, MiniMax reported total revenue of $79.038 million (approximately 569 million yuan), a year-on-year increase of 158.9%. Against its 222 billion Hong Kong dollar market cap, this implies a price-to-sales ratio of about 359 times. Using the Annual Recurring Revenue (ARR) metric favored by large model companies, MiniMax disclosed that its ARR grew over 100% in the past two months. Extrapolating from a disclosed base of over $150 million in ARR as of February, its current ARR likely exceeds $300 million, implying a valuation multiple of about 94 times ARR. For comparison, OpenAI's latest valuation is approximately $852 billion, with an estimated annualized revenue exceeding $20 billion, implying a multiple around 43 times. MiniMax's valuation appears significantly higher, built on assumptions of continued scarcity, sustained ARR doubling, improving gross margins, and achieving unit economics. If a growth slowdown arrives before profitability, or if overseas large model giants go public, MiniMax's valuation could face rapid downward pressure. The third risk is uncertainty stemming from its heavy reliance on overseas markets. Generating over 70% of its revenue internationally is both a badge of honor as a "domestic AI出海标杆" and a vulnerability. In September 2025, Disney and other companies sued its video product, Conch AI, for generating copyrighted movie characters, seeking up to $75 million in damages and demanding a ban on related features. MiniMax argued in its prospectus that Conch AI is merely a neutral generation tool driven by user prompts, and that even an unfavorable ruling would not materially impact its overall finances. However, generative AI copyright compliance risks, combined with legal, regulatory, and geopolitical variables in overseas markets, are amplified by its high international revenue exposure.
Final Assessment
Capital markets weigh both the narrative and the financials. MiniMax currently stands at a critical juncture. On one side, it has a compelling story built on inclusion in the Hang Seng Tech Index, its A-share fundraising plans, and rapid ARR growth. On the other side, it faces the gravitational pull of reality: the July share unlock, a price-to-sales ratio exceeding 350 times, and persistently widening losses. The large model race is a marathon requiring sustained investment over years. MiniMax has run incredibly fast. The central question is whether its operational performance can support its lofty valuation once the scarcity premium fades, major tech platforms leverage their traffic advantages, and the floodgates open for share sales in July.
Comments