A report published on Thursday (17) by the Rhodium Group estimates that Chinese artificial intelligence companies collectively generate around 10% of the annualized recurring revenue (ARR) reported by OpenAI and Anthropic. While OpenAI is valued at $40 billion and Anthropic at $65 billion in ARR, the total revenue of China’s leading companies falls short of $10 billion, according to the same study.
Revenue of China’s leading companies
DeepSeek has the lowest ARR at $500 million. MiniMax and Moonshot follow with $800 million and $1 billion, respectively. Z.ai recently reported a value of $1.8 billion to investors. ByteDance, meanwhile, recorded $4 billion, and Alibaba, $2.4 billion in annual recurring revenue.
Market valuations versus revenue
The study highlights that Chinese startups’ valuations appear disproportionate compared to their current revenue. Moonshot has a valuation-to-revenue multiple of 50 times, while DeepSeek reaches 163 times—figures significantly higher than OpenAI’s 34 times and Anthropic’s 21 times.
Anthropic is expected to go public in the United States next month, while OpenAI has postponed its initial public offering until 2025. Moonshot has reportedly filed a confidential IPO application in Hong Kong, and DeepSeek is also preparing for a stock market debut. When contacted, Moonshot declined to comment on rumors.
Open models, lower costs, and monetization challenges
Chinese labs widely use open-source models, which allow any user with suitable hardware to download and run the technology without transferring resources to the original developer. This feature, combined with lower prices per task, limits revenue generation, according to Artificial Analysis.
Logan Wright, a partner at the Rhodium Group, warns that the lack of adequate funding makes it harder for these top labs to achieve sustainable growth. “They will heavily depend on a favorable stock market climate—historically, this is not an easy bet in China,” he said in a statement to CNBC.
Real-world platform usage contrasts with revenue
Despite the revenue gap, Chinese models already dominate usage on the OpenRouter platform, accounting for 46.4% of all tokens processed, compared to 35.7% for American models. Although Anthropic represents only 14.8% of the volume, it captures more than half of total spending, thanks to a premium pricing strategy.
Conclusion
The Rhodium Group’s study shows that while Chinese AI companies are gaining ground in adoption, their revenue-generating capacity still lags far behind U.S. giants. The challenge now is to transform dominance in usage volume into sustainable sources of profit, especially in a scenario where model openness and price competition are key factors.
