BEIJING/SINGAPORE, Aug 26 (Reuters) - China's Moonshot AI is negotiating revenue-sharing agreements with Microsoft, Amazon and Alphabet's Google that would allow the U.S. cloud giants to host its blockbuster Kimi K3 model, three people familiar with the talks said.
Any deal could mark the first big revenue-sharing pact between a Chinese AI firm and a major U.S. cloud company.
The discussions highlight how China's leading AI models, often far cheaper than Western offerings, are gaining traction in the U.S., despite national security concerns in Washington that have led to bans on exports of AI chips to China. They are also taking place despite critical comments about Moonshot from senior U.S. officials.
IPO-bound Moonshot is seeking up to a 30% share of revenue generated from K3-related services on Microsoft's Azure, Amazon Web Services and Google Cloud, according to the sources who declined to be identified because the discussions are private.
That would be in line with terms that sources have said the startup has outlined for major customers using the open-weight model.
The discussions are at an early stage and there is no certainty they will result in agreements, the sources said.
Moonshot did not respond to a request for comment about the potential deals. Microsoft, Google, and AWS declined to comment.
Moonshot has come under fire from U.S. Treasury Secretary Scott Bessent who said last month that he might add it to a trade blacklist. U.S. officials have accused the Beijing-based company of stealing from Anthropic's most sophisticated model, Fable, to help create Kimi K3 and illegally acquiring Nvidia chips.
Moonshot has rejected suggestions that Kimi K3's performance was achieved via distillation, telling China's National Business Daily last month that its performance gains came from original changes to underlying architecture.
UNRESOLVED ISSUES IN THE TALKS
According to one of the sources, unresolved issues in Moonshot's negotiations with the U.S. cloud companies include how revenue might be split, data access and auditing token usage.
Tokens are units of text processed by AI models, and measuring their use is central to calculating revenue under usage-based billing.
Kimi K3 has posted strong results in third-party evaluations.
Arena.ai has ranked it first in a benchmark assessing web interface-building capabilities, while Artificial Analysis says it delivers performance comparable to OpenAI's GPT-5.5 and Anthropic's Claude Opus 4.8, particularly on tests measuring complex, multi-step tasks.