- Chinese foundation model developer returns to investors just six months after its IPO
- The deal highlights the soaring capital demands of the AI race
Chinese AI startup MiniMax has raised about HK$16 billion ($2 billion) in fresh equity financing, underscoring how rapidly leading foundation model developers are burning through capital as competition intensifies.
The company said on July 10 that the financing attracted more than 100 institutional investors, including sovereign wealth funds, long-only asset managers and multi-strategy funds from Asia, Europe and the United States.
The offering was about seven times oversubscribed, allowing MiniMax to increase the deal size from an initially planned $1.8 billion to more than $2 billion.
The fundraising comes just six months after the company’s Hong Kong IPO, which raised net proceeds of about $680 million.
MiniMax sold 35.6 million new shares at HK$268 each, a 9.9% discount to the previous day’s closing price, alongside HK$6.5 billion of zero-coupon convertible bonds with a conversion price of HK$335. Morgan Stanley and UBS acted as joint arrangers.
A widening gap
The fundraising comes as MiniMax’s market performance has diverged sharply from that of domestic rival Zhipu AI (智谱).
When the two companies listed in Hong Kong six months earlier, they were widely viewed as China’s twin champions in the foundation model race.
Since then, Zhipu has been rewarded by investors for its enterprise-focused strategy, while MiniMax’s consumer AI business has come under pressure amid lower monetization and higher computing costs.
By the July 9 close, Zhipu was valued at roughly HK$700 billion, while MiniMax’s market cap stood at just over HK$100 billion, down from a peak of HK$410 billion, leaving a gap of about HK$600 billion after the two companies went public in Hong Kong.
The contrast reflects two distinct business models. Zhipu generates most of its revenue from enterprise deployments, where customers are willing to pay premium prices.
MiniMax, by contrast, has focused on consumer AI applications, with overseas markets accounting for more than 70% of its revenue but requiring heavy investment to support large user bases.
The new AI arms race
MiniMax’s latest financing illustrates how the AI industry has entered a new phase in which access to capital is becoming as important as advances in model performance.
According to Chinese tech media, the company is developing a 2.7-trillion-parameter model named M3 Pro, and building a domestic AI computing cluster to support future training and inference workloads.
As model sizes continue to grow, computing infrastructure has become one of the industry’s biggest expenses.
The latest fundraising also suggests that for leading AI developers, an IPO is no longer an exit event but another source of financing for an increasingly capital-intensive business.
Why it matters globally
MiniMax’s latest transaction reflects a broader divergence emerging in China’s AI sector. Investors are increasingly differentiating between companies with proven enterprise pricing power and those pursuing consumer adoption at scale.
It also highlights a contrast with the U.S. AI sector, where commercial leadership has become concentrated among a handful of companies such as OpenAI and Anthropic.
The two firms account for an estimated 89% of the top AI model market by revenue.
China’s AI ecosystem remains far more fragmented, with numerous well-funded laboratories competing for market share, talent and computing resources.
As a result, the next stage of competition may depend less on who builds the most capable model and more on who can sustain the enormous capital investment required to commercialize it.
