A claim circulating in Korean tech and policy circles โ that the combined revenue of China's AI developers amounts to barely 10 percent of what OpenAI and Anthropic generate together โ has been put to the test. A review of 10 primary sources, including Rhodium Group's tracking of funding into the Chinese AI sector, found the framing broadly consistent with the revenue gap between the two US labs and China's crowded field of LLM startups, though the analysis stopped short of outright confirmation.
The statement, translated from Korean, asserts that even if the sales of all Chinese AI developers were added together, the total would reach only about one-tenth of OpenAI and Anthropic's combined revenue. The review found nothing in the primary sources that contradicts the core of that comparison: the leading US labs command a revenue base that remains an order of magnitude larger than anything reported by China's AI developers, most of which are private companies that disclose little financial detail.
Rhodium Group's widely cited charting of Chinese AI funding underscores the other side of the picture โ capital continues to flow into China's AI sector at scale, even as commercial revenue lags far behind that of the US frontrunners.

์ถ์ฒ: ์กฐ์ฌ ์ถ์ฒ
Three complications kept the assessment below full confirmation. First, both sides of the ledger are growing quickly, so any ratio is a snapshot of a fast-moving market rather than a stable fact. Second, revenue definitions vary โ API sales, subscription plans and enterprise contracts are counted differently across firms โ which blurs the comparison. Third, Chinese AI developers release limited audited figures, forcing analysts to rely on estimates and partial disclosures.
The reviewers concluded that the claim captures a real and large gap, but that its precision is an artifact of a single moment in time. In a market where funding rounds and product launches arrive monthly, the ratio could already have shifted since the data was assembled.
For readers tracking Korea's AI and semiconductor ecosystem, the revenue gap matters because it shapes where computing demand concentrates. Spending by the dominant US labs โ which buy accelerators, memory and data-center capacity at scale โ remains the reference point for suppliers, while China's AI developers are still in the early commercial stage even as investment pours in.
The conclusion: the claim is Mostly True โ the revenue gap is real, but the fast-growing snapshot behind the 10 percent figure means the exact ratio may already be out of date.