MiniMax officially commenced trading at the Hong Kong Stock Exchange at 9:30 AM this morning! As of this writing, the stock price has skyrocketed nearly 80%! MiniMax has attracted top global long-term funds to increase their positions in Hong Kong, leading the rise of Chinese tech stocks and charting an independent course. Remarkably, MiniMax has joined the global top tier in full-modal AI capabilities using only an estimated 1% of OpenAI's expenditure (approximately $500 million). Concurrently, the company's workforce boasts an average age of just 29, as a group of young talents has achieved financial freedom through robust AI technology while contributing highly cost-effective open-source models to the world.
At 9:30 AM on January 9th, MiniMax listed on the Hong Kong Stock Exchange.
When the stock code '0100', which possesses a distinct binary aesthetic, appeared on the screen, the most striking aspect of the scene was not the instantaneous fluctuation of the stock price, but the group of young people standing beneath the ceremonial gong.
The figure on the right is MiniMax's 36-year-old founder, Yan Junjie. According to the stock allocation results announced by MiniMax, if all additional shares reserved for price stabilization (i.e., full exercise of the 'greenshoe' option) are sold, the total offering amounted to approximately 33.58 million shares.
The final price was set at the maximum of the planned range, HK$165 per share, raising a total of approximately HK$5.54 billion.
Market demand was exceptionally strong: The retail portion for individual investors was oversubscribed by a staggering 1,837 times;
The international offering for major institutions was also oversubscribed by 37 times.
Trading began at 9:30 AM today, and as of this writing, MiniMax's stock price has surged nearly 80%.
At this moment symbolizing a capital market coronation, CEO Yan Junjie and his young team have achieved a breakthrough against conventional industry wisdom.
The team is remarkably young, yet its technology is exceptionally mature. At this moment symbolizing a capital market coronation, CEO Yan Junjie and his young team have achieved a breakthrough against conventional industry wisdom.
Over the past three years, the 'Hundred Models War' has left a harsh stereotype in the market: AI is an arms race for giants, with a ticket to entry costing tens of billions of dollars.
OpenAI burned through billions of dollars to achieve the brilliance of ChatGPT, while Silicon Valley unicorns rely on the endless financial backing of Microsoft and Google.
But MiniMax offers a different answer.
This company, founded less than four years ago, has spent a cumulative total of approximately $500 million (merely 1% of OpenAI's expenditure) to propel its models for text, speech, and video into the international top tier.
On its first day of trading, the capital market expressed its stance with real money.
Investors understood this counter-intuitive narrative: even in a track dominated by giants, a Chinese company can successfully navigate an independent path from 'technological leadership' to 'commercial viability' by relying on extreme engineering efficiency and correct technological bets.
The listing of MiniMax marks a definitive turning point for China's AI industry.
The market is no longer merely paying for a speculative traffic story, but is, for the first time, able to systematically price a genuinely globally competitive 'AI productivity company'.
Shattering the 'Bottomless Pit' The Efficiency Truth Behind the 1% Cost Investors in the primary market had been gripped by prolonged anxiety: Is the large model business still a good one?
With soaring computing power costs and ambiguous application scenarios, domestic large models have become mired in a homogenized price war.
Kai-Fu Lee once predicted that the Chinese large model market would likely consolidate to a number that can be 'counted on one hand'.
This 'winner-takes-all' rhetoric made the survival space for startups appear precarious.
MiniMax's prospectus serves as a powerful rebuttal to this pessimistic view.
It was widely believed that achieving internationally leading full-modal capabilities required a ticket price of at least tens of billions of dollars.
MiniMax accomplished it with an expenditure of only about $500 million.
This exceptional capital efficiency stems from the team's precise foresight regarding the technological path from the very beginning.
Looking back to 2023, it was a time when everyone was frantically chasing ChatGPT's text capabilities.
But Yan Junjie and his team made a decision that, at the time, seemed quite risky and was even questioned for being 'unfocused': to simultaneously advance the research and development of text, speech, and video modalities.
This was by no means a blind attempt to be comprehensive, but a deduction based on first principles.
True AGI must, like humans, be capable of processing multimodal information involving hearing, speaking, seeing, and writing simultaneously.
Relying solely on single-modal capabilities would inevitably create insurmountable shortcomings for the goal of pursuing the upper limits of intelligence and AGI.
MiniMax chose a difficult path but achieved an overtaking maneuver at the finish line.
Parallel R&D across all modalities enabled 80% resource reuse in the underlying architecture, avoiding repetitive 'patchwork' development.
A review of its cornerstone investor list reveals a prominent lineup of international long-term capital.
These 'old money' institutions, managing hundreds of billions of dollars, are notoriously selective about Hong Kong IPOs.
Their rare, concerted action to heavily invest in a young Chinese AI company underscores the appeal of its scarcity as an 'independent player'.
They are not seeking another monetization tool dependent on a giant's traffic pool; they are searching for the next global technology enterprise with independent viability.
MiniMax, with its '1% cost', has demonstrated it possesses the right DNA.
Valuation Logic Reshaped Building Global Productivity, Not a Traffic Business Attempting to understand MiniMax's value using the old internet traffic logic is challenging.
It lacks the staggering daily active users of ByteDance or the massive GMV growth rate of Pinduoduo.
The market is assigning it a new valuation based on 'intelligent productivity premium'.
MiniMax is one of the very few Chinese tech companies truly 'born global' from the outset.
Its prospectus shows that over 70% of its revenue comes from overseas.
This indicates the global competitiveness of the user experience supported by its technology.
While most domestic large models are still engaged in cutthroat price competition within China, vying for B2B private deployment orders and driving API prices below cost, MiniMax is already generating high-value US dollar revenue across the ocean.
As of September 30, 2025, MiniMax's products reached over 200 countries and regions globally, serving more than 212 million users.
Even more impressive is its healthy business闭环 (closed loop).
In the SaaS and AI industry, days sales outstanding (DSO) typically range from 60 to 90 days, meaning companies often have to fund operations themselves, facing immense cash flow pressure.
MiniMax has compressed this figure to an astonishing 38 days.
This number reveals a fundamental difference in business models: MiniMax is not selling 'custom projects' that require lengthy bidding processes and relationship-building, but rather 'standardized productivity' that users and enterprises urgently need and are willing to pay for immediately.
This high turnover and efficient collection directly enhance profit quality.
MiniMax's gross margin for its B2B business is as high as 69.4%.
This reflects a profit structure typical of a software company, not an outsourcing firm.
It demonstrates that MiniMax has built a solid commercial moat globally through technological premium.
In the Hong Kong market, which values profitability and cash flow, MiniMax provides the most compelling response: an AI company does not need to rely on promises to sustain a high valuation; it can achieve substantive growth now, with both B2B and B2C segments driving a 160% and 181% year-over-year revenue increase, respectively.
Reviewing the Technological Path Strategic Resolve for Full Modality Reviewing MiniMax's technological journey reveals a victory for 'long-termism'.
In 2023, while the entire industry was obsessed with scaling text parameters, MiniMax persisted in advancing text, speech, and video concurrently.
This approach, which seemed like dispersing efforts at the time, has now proven to be the necessary path towards AGI.
In 2024, its Speech 02 model ranked first in comprehensive performance, with users generating over 220 million hours of speech.
This incredibly natural, emotionally nuanced voice interaction became a key component of the user experience in its C-end products, Xingye/Talkie.
Breakthroughs in video followed closely.
Hailuo 02, released in June 2025, ranked second globally in the authoritative AA Video Arena, trailing only the industry's top closed-source model.
Creators used it to generate over 590 million videos, with the model understanding physical laws and creative intent without complex prompts.
In the text domain, the M2 series (including M2 and M2.1), whose release began in October 2025, shattered the stereotype that 'Chinese models only excel in Chinese leaderboards'.
It immediately ranked among the global top five and was number one among open-source models.
On the global model aggregation platform OpenRouter, M2 quickly climbed to become the most used domestic model by token volume.
M2.1, released at the end of last month, went a step further, directly claiming the top spot among open-source models!
This series of achievements spanning speech, video, and text validates MiniMax's unique understanding of Scalability—it is not merely the stacking of parameters in a single modality, but the use of a unified underlying architecture to simultaneously support the emergence of intelligence across all modalities.
Scalability isn't just about stacking GPUs and data centers; it's about the extensibility of the technical architecture.
Using fewer parameters and a more efficient Mixture-of-Experts (MoE) architecture, MiniMax achieved results comparable to trillion-parameter models.
Against the macro backdrop of constrained computing power, MiniMax has maintained a Moore's Law-like pace of technological iteration—advancing steadily year after year without missing a beat.
This proves that in a high-uncertainty field like AI, blindly following trends is the most dangerous approach; only by making early layouts based on judgments about the endgame can one remain at the table after the tide recedes.
Organizational Innovation The 'AI Native' Experiment Led by Post-95s Technology is the skeleton; organization is the lifeblood.
The youthfulness of MiniMax's lifeblood is astonishing.
With a total staff of 385, the average age is 29, and the average age of directors is 32.
The idea of this group of post-95s steering a global listed company is almost unimaginable in the traditional business world.
But these young people are conducting an unprecedented experiment in AI Native organization.
In traditional large tech firms, 300 people might be allocated to just a medium-sized project, bogged down in layers of reporting and PowerPoint presentations.
At MiniMax, these 385 individuals support hundreds of millions of users and a multi-billion dollar valuation.
The secret to high personnel efficiency lies in the restructuring of workflows.
Internal sources reveal that over 80% of the code within MiniMax is generated by AI.
Beyond significant cost savings, the more crucial objective is to embed AI into the deepest layers of production, enabling it to discover and solve practical problems.
Engineers have transformed into 'architects' and 'trainers' for AI, directing it to write code and perform data analysis.
This working method allows MiniMax to achieve an output efficiency where 'one person does the work of ten'.
In terms of organizational structure, there are no more than three hierarchical levels below the CEO.
The extremely flat design ensures lossless transmission of information.
An idea from a newly hired graduate could potentially reach the CEO's desk within two days and be implemented as a model update within three.
This organizational form is a dimensional reduction attack on traditional bureaucracy.
While competitors are still coordinating across departments, MiniMax's young engineers have already completed a model iteration.
When smart people decide to stop working for giants, they create more than just another new company.
They do not迷信 (blindly believe in) brute-force manpower tactics, focusing instead on talent density.
They have proven that, empowered by AI, small teams are fully capable of challenging giants.
This is also the biggest revelation MiniMax's listing offers the industry: future competition will be a one-sided碾压 (crushing defeat) of high-IQ, high-efficiency organizations over bloated ones.
The 'Coming-of-Age Ceremony' for a New Generation of Chinese Tech Companies The listing of MiniMax is a highly instructive case study in the history of Chinese technology commerce.
It signifies that domestic large models have moved beyond the ivory tower of the laboratory and the虚幻光环 (illusory halo) of PowerPoint presentations, entering the残酷却真实 (brutal yet real) 'moment of truth'.
It proves that even in the current challenging external environment, Chinese companies can still secure a seat at the global AI table through extreme efficiency, correct strategic foresight, and agile global positioning.
This is not just a victory for MiniMax alone, but a strong signal for the industry: China's AI industry is transitioning from a 'burn-rate mode' to a 'viability mode', from 'following and imitating' to 'independently defining'.
Of course, the listing is merely the prologue; the road ahead remains perilous.
Model iterations from Google and OpenAI are accelerating, and the fog of global competition has not yet lifted.
For Yan Junjie and his young team, the listing is just an entry ticket to the next stage of the competition.
The ultimate test for these young people will be how to continuously maintain their Scaling capability in a jungle surrounded by giants.
At the 2024 World Artificial Intelligence Conference (WAIC), Yan Junjie proposed the vision: 'Intelligence with Everyone'.
He hopes that AI can become a readily accessible external brain for productivity for everyone.
On the day of the listing ceremony, Yan Junjie also delivered a speech on behalf of MiniMax:
We believe the true value of an AI company lies in providing a continuous stream of advanced intelligence for people to use. This relies on sustained technological innovation, and even more so on the inclusivity and openness of the entire process. This is also the meaning of our mission, 'Intelligence with Everyone'. Over the past four years, starting from scratch, a group of young people has built globally leading full-modal capabilities and served over 200 million users worldwide. But this is only the beginning. We anticipate that the pace of progress in the AI industry over the next four years will be as fast as the past four, and we will strive to make our own contributions within it. What is more important in this process is to find more fellow travelers. We are proud of this team full of passion and innovative capability. At the same time, we eagerly look forward to more outstanding talents from around the world joining us. Let us continue to strive for the advancement of intelligence and make greater contributions to social and economic development. Standing now in the spotlight of the Hong Kong Stock Exchange, we have reason to harbor greater expectation for this vision.
The story of MiniMax tells us that in an era ruled by algorithms, the most precious computing power comes from humanity's unceasing curiosity and courage.
True greatness in open source is not only about contributing groundbreaking code, but also about demonstrating a possibility to the world—
Even starting from humble beginnings, by persistently doing what is difficult yet right, the flame of wisdom from the East will eventually merge into the brilliant galaxy of humanity's exploration of AGI, illuminating the future of our shared human destiny.
