Chinese AI firms' P/S ratios surpass those of their U.S. peers by decades.

ChainCatcher report, according to Tommy’s analysis, Chinese open-source AI companies exhibit a significant gap between valuation and revenue conversion, with their price-to-sales (P/S) ratios far exceeding those of leading U.S. peers. Data shows that Zhipu, the developer of the GLM 5.2 model, currently has a market capitalization of approximately $137 billion, yet its fiscal year 2025 revenue is estimated at just $107 million, resulting in a P/S ratio of 1,280x. MiniMax has a market cap of about $23 billion, with a P/S ratio of roughly 290x. In contrast, leading U.S. AI labs have more grounded valuations: OpenAI (valued at ~$852 billion) and Anthropic (valued at ~$965 billion) have P/S ratios of only 34x and 21x, respectively. Tommy argues that due to overseas users’ concerns over data privacy, they are reluctant to send data directly to Chinese AI companies, preventing massive demand from translating into actual API revenue—resulting in substantial profits flowing instead to third-party overseas inference providers (such as OpenRouter). To justify their current high valuations, Chinese AI companies urgently need to demonstrate their data non-retention mechanisms and capture market share through low pricing, or explore revenue-sharing and exclusive licensing partnerships with overseas inference platforms to expand their real-world revenue streams.