Chinese AI Leaders Struggle to Profit Despite Technical Gains
Heavy losses and a capital gap with U.S. rivals threaten the sustainability of China's AI boom.
China's most prominent artificial intelligence companies are posting steep losses even as their models reach performance levels that rival American systems, raising questions about the commercial viability of the country's AI ambitions.
Zhipu AI, known commercially as Z.ai, exemplifies the paradox. The company reported losses of approximately $330 million in the first half of 2025 and about $411 million in 2024, according to independent financial reports, even as revenue grew from 57.4 million yuan in 2022 to 312 million yuan in 2024. The pattern extends across the sector: Moonshot AI raised $2 billion in May 2026 at a $20 billion valuation, while DeepSeek secured roughly $7.4 billion (50 billion RMB) in June 2026 at a valuation between $50 billion and $59 billion.
The Capital Gap
Yet these figures pale next to U.S. counterparts. Anthropic raised $65 billion in Series H funding in May 2026 alone, at a $965 billion valuation. The disparity underscores a widening capital chasm that could determine which nation's AI industry achieves sustainable scale.
"Open models are a powerful distribution strategy, but they are not a complete business model," said Wei Sun, principal AI analyst at Counterpoint Research.
Chinese firms face a perfect storm of constraints. A culture favoring open-source and open-weight models has accelerated technical development while eroding the ability to charge premium prices. U.S. export controls on advanced chips force companies to rent remote data centers or optimize for chip efficiency, driving up operational costs. Meanwhile, domestic consumers remain highly price-sensitive, limiting revenue potential.
Compute Constraints Bite
The infrastructure crunch hit Moonshot AI hard this year. When the company launched its Kimi K3 model, a shortage of computer chips forced it to stop accepting new users within two days of release. The incident highlighted how even well-funded Chinese startups lack guaranteed access to the compute needed to scale.
"In two or three years we will still be trying to figure out how large models can earn money," said Richard Lin, vice president at Datastrato.
State Support or IPOs Ahead
The struggle to find a sustainable business model suggests that technical parity with U.S. AI does not guarantee commercial success. If Chinese firms cannot monetize independently, they may rely more heavily on state funding or public listings to survive. That shift would transform the AI race from a commercial competition into a state-subsidized arms race.
The dynamics also create unexpected interdependence. Some Silicon Valley startups have begun integrating cheap Chinese open-weight models into their stacks, a trend U.S. policymakers are now debating whether to restrict. For now, China's AI powerhouses remain caught between technical achievement and financial reality, racing to prove that their models can do more than impress benchmarks—they must also pay the bills.