Key Takeaways
- The collective annual recurring revenue of China’s premier AI models equals approximately 10% of OpenAI and Anthropic’s combined earnings
- OpenAI reports $40 billion in ARR while Anthropic reaches $65 billion
- ByteDance tops Chinese competitors with $4 billion ARR; DeepSeek trails at $500 million
- Valuation multiples for Chinese AI ventures run extraordinarily high, with DeepSeek at 163x compared to OpenAI’s 34x
- Capital access limitations and volatile equity markets threaten Chinese AI laboratories’ expansion capabilities
China’s AI companies are experiencing rapid expansion, yet their earnings represent merely a fraction of revenues generated by American counterparts, new data from Rhodium Group reveals.
Analysts discovered that China’s top-tier AI models collectively produce roughly 10% of the annual recurring revenue documented by OpenAI and Anthropic. This substantial disparity persists with little indication of narrowing in the near term.
Rhodium’s assessment places OpenAI’s ARR at $40 billion, with Anthropic reaching $65 billion. These figures significantly overshadow current revenue streams from Chinese enterprises.
ByteDance commands the Chinese market with $4 billion in ARR, trailed by Alibaba at $2.4 billion. Z.ai, previously operating as Zhipu AI, disclosed to investors this week that its ARR achieved $1.8 billion.
Moonshot recorded $1 billion, MiniMax registered $800 million, and DeepSeek reached $500 million. Even when combined, these amounts fall considerably below OpenAI’s individual performance.
Sky-High Valuations Despite Revenue Gap
Notwithstanding smaller revenue streams, Chinese AI ventures command elevated valuation multiples. Rhodium calculated DeepSeek’s valuation-to-revenue proportion at 163x, while Moonshot stands at 50x.
Comparatively, OpenAI maintains a 34x ratio and Anthropic holds 21x. Rhodium’s analysis characterized Moonshot and DeepSeek’s valuations as “exorbitant” when measured against present revenue generation.
This dynamic suggests investors are placing substantial wagers on prospective expansion that remains unrealized in current financial performance.
Capital Access and Growth Obstacles
Logan Wright, a Rhodium Group partner who contributed to the analysis, emphasized that Chinese AI laboratories confront significant financing obstacles moving forward.
“The financing gap means it will be far more difficult for Chinese frontier AI labs to scale sustainably,” Wright said. He added that state funding in China has mostly gone toward chips and servers, not model developers.
Rhodium calculated that government-affiliated capital sources comprised over 60% of equity investment directed toward Chinese AI semiconductor and hardware infrastructure sectors.
Wright further observed these enterprises will depend substantially on equity markets, which have demonstrated historical volatility within China.
Information from Artificial Analysis indicates that premier American models from OpenAI and Anthropic remain predominantly closed-source and command higher per-task pricing than Chinese competitors. Chinese models have captured market share partially through aggressive pricing strategies.
Multiple Chinese AI enterprises are currently pursuing public market debuts. Moonshot has allegedly submitted confidential documentation for a Hong Kong IPO, while DeepSeek reportedly prepares for its own listing.
Among American companies, Anthropic anticipates going public in October. OpenAI has allegedly postponed its listing timeline until next year.
The Rhodium analysis delivers an unambiguous assessment: Chinese AI ventures demonstrate growth momentum, yet the revenue differential with American industry leaders remains substantial, and the trajectory toward bridging this gap carries significant uncertainty.


