China’s Moonshot AI in talks with Microsoft, Amazon, and Google to bring Kimi K3 to US clouds

China’s Moonshot AI is negotiating with Microsoft, Amazon, and Google over deals that could put its Kimi K3 model inside America’s biggest cloud platforms, creating an unusual collision between Washington’s AI restrictions and the economics of selling AI.

The Beijing-based startup is seeking up to 30% of revenue generated from Kimi K3 services offered through Microsoft Azure, Amazon Web Services, and Google Cloud, Reuters reported. The discussions remain in the early stages, and there is no guarantee agreements will be reached.

“China’s Moonshot AI is ​negotiating revenue-sharing agreements with Microsoft, Amazon, and Alphabet’s Google would allow the U.S. cloud giants to host its ‌blockbuster Kimi K3 model,” Reuters reported, citing three people familiar with the private talks.

If completed, the arrangements could become the first major revenue-sharing deals between a Chinese AI model developer and a top U.S. cloud provider.

The news comes just a month after Moonshot AI released the open weights for Kimi K3, a 2.8-trillion-parameter model and the largest open-weight AI model released to date.

The timing is striking. U.S. Treasury Secretary Scott Bessent said last month that Moonshot could be added to a trade blacklist. U.S. officials have accused the company of improperly using Anthropic’s Fable model to help develop Kimi K3 and illegally obtaining Nvidia chips.

Moonshot has rejected claims that Kimi K3’s performance came from distillation. The company told China’s National Business Daily last month that improvements came from original changes to the model’s underlying architecture.

Kimi K3’s economics may be hard for U.S. clouds to ignore

The talks point to a larger tension emerging in AI. Washington has spent years restricting China’s access to advanced chips, yet Chinese developers are producing increasingly competitive models that American cloud companies may have a financial incentive to offer their customers.

Kimi K3 is an open-weight model with 2.8 trillion parameters. Customers can download and modify it, but running a model of that size requires enormous computing resources. That makes cloud distribution particularly valuable. For many companies, accessing K3 through Azure, AWS, or Google Cloud is far more practical than building the infrastructure to run it themselves.

Independent benchmarks have added credibility to the model. Arena.ai ranks Kimi K3 first on a benchmark measuring web interface-building capabilities. Artificial Analysis reports performance comparable to OpenAI’s GPT-5.5 and Anthropic’s Claude Opus 4.8, particularly on tests involving complex, multi-step tasks.

Price could matter as much as benchmark scores. Chinese AI developers have increasingly competed by offering capable models at substantially lower costs than many Western alternatives, pressuring the economics of proprietary AI.

Some details could still derail the negotiations. Reuters reported that Moonshot and the cloud providers have yet to resolve questions around revenue allocation, data access, and auditing token consumption. Token measurement matters directly to the proposed business model since usage determines how much revenue each side receives.

Moonshot has already tested the approach elsewhere. Chinese IT services company Chinasoft International disclosed a revenue-sharing agreement with Moonshot in July, though the companies did not reveal the split. Reuters reported that Moonshot has similar arrangements with smaller cloud platforms.

Founded in 2023 by Carnegie Mellon-trained AI researcher Yang Zhilin, Moonshot has quickly become one of China’s most closely watched AI startups. Alibaba is among its backers. The company raised more than $2 billion in May and is preparing for a potential Hong Kong listing, according to previous reports.

That makes these negotiations bigger than another cloud distribution deal. If Microsoft, Amazon, or Google reaches an agreement with Moonshot, it would show how difficult it is to separate the U.S.-China AI rivalry from commercial reality.

Washington may see Chinese AI models as a strategic risk. American cloud providers may increasingly see them as products their customers want to use.