Analysts at Rhodium Group pointed out that the problem of monetizing the development of artificial intelligence startups against the background of billions of dollars of expenditure is quite serious, but in this sense, the Chinese competitors of the US market leader are ten times lower than them.
Image source: Moonshot AI
The authors of the analysis report pointed out that from March to August this year, the advanced artificial intelligence models developed by seven Chinese companies brought them approximately US$10.7 billion in annual recurring revenue (ARR). During the same period, American Anthropic and OpenAI’s combined annual recurring revenue exceeded $100 billion. In fact, the amounts received differ almost by an order of magnitude, and while on the American side we are talking about the activities of only two companies, on the Chinese side their number reaches seven. More precisely, the report covers the activities of DeepSeek, Moonshot AI, Z.ai, MiniMax, Alibaba Group Holding, ByteDance and Kuaishou Technology – the latter of which operates AI video generator Kling.
The annual recurring revenue metric is calculated as the annualized value of subscription revenue for AI model developers based on data for a selected period. In this case, this is the months from March to August of that year. Extrapolating monthly revenue from AI model subscription sales to annual revenue yields the required value.
As of July, ByteDance’s ARR value has reached approximately US$4 billion, establishing itself as the leader in China’s artificial intelligence model market. Alibaba ranked second with $2.4 billion in August. Meanwhile, U.S. OpenAI hit the $40 billion milestone in August, and Anthropic had already demonstrated adjusted annual revenue worth $65 billion in July. That is, the study’s authors took the companies’ peak monthly subscription revenue between March and August and then multiplied by 12 to extrapolate. DeepSeek and Kuaishou Technology turned out to be counter-leaders, reaching annual adjusted subscription revenue levels of $500 million at the end of June and end of March respectively. MiniMax topped them with $800 million in late August, the same month that became Moonshot AI’s most successful month of the accounting period, with the company reporting $1 billion in annual revenue.
Z.ai representatives told investors this week that they had reached $1.8 billion in ARR. Meanwhile, representatives from Rhodium Group found that the startup’s annual revenue from providing customers with access via APIs reached $1.6 billion in August.
Image source: DeepSeek
This disproportionality compared to the performance of U.S. competitors has not reduced the number of people willing to invest in Chinese AI startups; fortunately, their capital is now significantly lower than that of their U.S. counterparts. In the case of Moonshot, we can talk about a capitalization of around $50 billion; the Chinese startup also recently filed for an IPO in Hong Kong. It is understood that DeepSeek may be valued at as high as US$74 billion before its listing. South China Morning Post. Anthropic and OpenAI also plan to go public. The former’s capital size may reach US$2 trillion, while the latter is expected to reach US$1.2 trillion based on the results of the last round of IPO financing.
Chinese startups have significantly higher capitalization-to-ARR ratios than U.S. startups, which may indicate high investor interest in their assets. For example, Moonshot and DeepSeek reach this ratio of 50 and 163 respectively, while OpenAI settles for a multiplier of 34 and Anthropic is limited entirely to 21. The capital investment payback period of Chinese AI startups is significantly higher, indicating that there are problems with the efficiency of monetizing their services at this stage. On the other hand, the revenue of many Chinese companies is increasing rapidly. At the end of the first half, Z.ai grew by 400% to US$142 million, and MiniMax grew by 283% to US$116.6 million.
Analysts at Macquarie Group predict Z.ai and MiniMax may remain unprofitable until 2030 due to higher infrastructure spending. At this stage of the AI boom, investors tend to accept the risk of not being profitable in the long term and bank on good financial returns in the distant future. Additionally, many analysts agree that Chinese AI model developers are rapidly closing the gap with Western rivals. The initially chosen policy of a decentralized open weight model made it difficult for Chinese companies to monetize their business quickly, but they began to look for business models suitable to solve this problem. For example, Moonshot and Alibaba set out in their agreements with customers the possibility of receiving a portion of the customer’s revenue if certain conditions are met. “Commission” can reach 30% of the core revenue of such companies’ customers.
If you find an error, select it with your mouse and press CTRL+ENTER.