MiniMax and Zhipu AI, two of China’s most prominent generative artificial intelligence startups, have offered the first detailed look at their business models through recent listing filings. The disclosures highlight the financial challenges facing Chinese AI developers as they prepare for potential initial public offerings in Hong Kong.
Both companies are backed by major domestic technology groups, including Alibaba Group Holding and Tencent Holdings, and are widely viewed as leading domestic alternatives to OpenAI. They are among the remaining players after an intense period of competition in China’s AI sector, often referred to as the “Battle of One Hundred Models,” which saw dozens of startups exit or consolidate.
According to the filings shared by Bloomberg, Zhipu generated 312.4 million yuan, or about $44.4 million, in revenue in 2024. MiniMax reported $30.5 million in revenue for the same period. While growth has been steady, those figures remain small compared with U.S. rivals. OpenAI is estimated to be generating about $13 billion in annualized revenue, while Anthropic is projected to reach roughly $9 billion this year.
The contrast underscores the scale advantage enjoyed by Silicon Valley AI labs, which benefit from deeper enterprise adoption, higher pricing power, and broader international reach. Zhipu and MiniMax are each valued at around $4 billion, far below the valuations attached to leading U.S. developers.
Diverging Strategies and Rising Costs
The two companies have pursued different paths to commercialization. Zhipu, founded by researchers from Tsinghua University, has focused on building advanced foundation models while delivering customized AI systems for government-linked and state-owned clients. This approach has helped drive revenue but has also tied growth closely to public sector demand.
MiniMax has emphasized consumer-facing and international products, including AI companion applications and video editing tools. The strategy aims to diversify revenue sources and reduce reliance on domestic enterprise clients, though monetization remains limited at this stage.
Both companies reported widening losses as research and development spending increased. Investment in model training, compute infrastructure, and talent continues to weigh heavily on profitability. The filings show that operating expenses have grown faster than revenue, reflecting the capital-intensive nature of large language model development.
Gross margins provide further insight into their business models. Zhipu currently reports higher margins, supported by enterprise contracts and bespoke deployments. MiniMax’s margins trail but have been improving as the company refines pricing and scales select products.
IPO Push Tests Investor Appetite
The disclosures come as MiniMax and Zhipu race to become the first Chinese generative AI startups to go public. Analysts expect each company could raise several hundred million dollars in Hong Kong listings, depending on market conditions.
The IPO push will test investor appetite for AI firms with heavy losses and long paths to profitability, particularly as global enthusiasm for AI stocks becomes more selective. Unlike their U.S. counterparts, Chinese AI startups face tighter domestic pricing, regulatory constraints, and limited access to foreign markets.
Even so, the filings suggest both companies believe scale and state backing can eventually narrow the gap. For investors, the documents offer a clearer view of the economic realities behind China’s push to build domestic AI champions in a market still dominated by U.S. players.
Meanwhile, OpenAI has also begun preparing for a potential IPO that could value the company at up to $1 trillion, underscoring the scale gap between Chinese AI startups and their Silicon Valley counterparts and highlighting how global capital markets continue to favor companies with larger revenue bases and international reach.
