Rhodium: U.S. AI leaders out-earn China's entire model sector by 10-to-1
Rhodium Group estimates that all Chinese AI models together generate only about 10% of the revenue reported by OpenAI and Anthropic. The research firm notes that low revenue does not necessarily align with lower valuations, flagging Moonshot and DeepSeek as having exorbitant valuations relative to revenue. Z.ai raised its revenue forecasts on Wednesday but still trails far behind its U.S. counterparts.
Rhodium Group's estimates, based on annual recurring revenue figures, place OpenAI at roughly $40 billion and Anthropic at $65 billion, while China's major model developers collectively reach only a fraction of that sum. Among Chinese firms, ByteDance leads with $4 billion in ARR, followed by Alibaba at $2.4 billion, with Z.ai reporting $1.8 billion on Wednesday. Smaller players trail significantly: Moonshot at $1 billion, MiniMax at $800 million, and DeepSeek at $500 million.
The revenue gap stands in sharp contrast to valuation multiples. Rhodium flagged Moonshot and DeepSeek at estimated revenue ratios of 50x and 163x respectively, versus 34x for OpenAI and 21x for Anthropic. Several of these companies are reportedly pursuing public listings, with Anthropic expected in the U.S. next month, OpenAI targeting next year, and both Moonshot and DeepSeek reportedly preparing Hong Kong offerings.
This revenue disparity could reshape how investors assess AI companies globally, potentially cooling enthusiasm for Chinese startups whose valuations far outpace their earnings. If the gap persists, it may pressure Chinese firms to demonstrate clearer monetization paths or face funding constraints. Conversely, the high valuations could signal confidence in future growth, particularly as China's domestic market expands. The outcome may influence capital allocation decisions across the sector, affecting which companies secure the resources needed to compete in an increasingly crowded field.