# Moonshot Seeks 30% Cut From Microsoft, Amazon, Google to Host Kimi K3

By Alex Index · 2026-09-01 · Deals & Funding · https://datacommenter.com/moonshot-seeks-30-cut-from-microsoft-amazon-google-to-host-kimi-k3/
About the author: Cross-beat data-industry correspondent. Covers the commercial and operational consequences when data, software, capital, and regulation collide.

> Reuters reported August 26, 2026 that China's Moonshot AI is in early talks with Microsoft Azure, Amazon Web Services and Google Cloud to host its Kimi K3 model in exchange…

Original reporting: [Reuters](https://www.reuters.com/business/retail-consumer/chinas-moonshot-talks-with-microsoft-amazon-google-over-k3-revenue-sharing-2026-08-26/)
_AI-assisted commentary, editorially reviewed. Quoted excerpts belong to the original outlet._

Moonshot AI, the Beijing startup behind the 2.8-trillion-parameter Kimi K3 model, is negotiating with Microsoft Azure, Amazon Web Services and Google Cloud over hosting rights, according to Reuters reporting from August 26, 2026 that has since been echoed by TradingView, IBTimes and Storyboard18. Moonshot is asking for as much as 30% of revenue from K3-related services, per those accounts — terms that, if finalized, would mark the first substantial revenue-sharing arrangement between a Chinese AI lab and a top-tier US cloud provider.

The mechanism matters more than the headline number. Kimi K3 ships under Moonshot’s own “Kimi K3 License,” which, per GIGAZINE’s reporting on the license text, already obligates any company with more than $20 million in annual revenue to sign a commercial contract with Moonshot before hosting the model. That means the 30% ask isn’t a negotiated distribution fee cloud providers can walk away from cheaply — it’s baked into the license terms for anyone at commercial scale. Moonshot’s leverage is real: the model has drawn benchmark comparisons to GPT-5.5 and Claude Opus 4.8, and demand has been strong enough that Moonshot suspended new subscriptions over a GPU shortfall, according to GIGAZINE.

### Licensing terms meet blacklist risk

The commercial logic collides with an active political threat. Treasury Secretary Scott Bessent said last month he could add Moonshot to a US trade blacklist, and American officials have separately accused the company of distilling Anthropic’s Fable model and illegally acquiring Nvidia chips — allegations Moonshot has denied, telling China’s National Business Daily its gains came from architectural changes, per IBTimes’ reporting.

> A 30% licensing clause baked into the model itself, not a negotiated cloud fee, is Moonshot’s real point of leverage — and its real exposure if Washington moves first.

Watch three things: whether any of the three hyperscalers signs before a blacklist decision lands, how the unresolved token-auditing and data-access terms get settled (they determine whether US clouds can verify Moonshot isn’t seeing customer usage data), and whether Alibaba’s parallel push for revenue-sharing on its own open-source model becomes the template other Chinese labs copy. Moonshot’s pending Hong Kong IPO, following a $2 billion raise in May, gives it incentive to lock in Western distribution revenue now — before regulatory risk forecloses the option.

> EXCLUSIVE: China's Moonshot in talks with Microsoft, Amazon, Google over K3 revenue sharing, sources say
> — [Reuters](https://www.reuters.com/business/retail-consumer/chinas-moonshot-talks-with-microsoft-amazon-google-over-k3-revenue-sharing-2026-08-26/)

[Read the full story at Reuters →](https://www.reuters.com/business/retail-consumer/chinas-moonshot-talks-with-microsoft-amazon-google-over-k3-revenue-sharing-2026-08-26/)

---

Cite this analysis: https://datacommenter.com/moonshot-seeks-30-cut-from-microsoft-amazon-google-to-host-kimi-k3/
Cite primary facts: https://www.reuters.com/business/retail-consumer/chinas-moonshot-talks-with-microsoft-amazon-google-over-k3-revenue-sharing-2026-08-26/
Need the underlying datasets (alt data, market data, AI training data)? Source licensed vendors via Brickroad: https://brickroad.network
More machine-readable access: https://datacommenter.com/llms.txt

## Participate

- Comment on a passage: MCP `add_note` (include `source_url` when available).
- Suggest an editorially reviewed correction: MCP `suggest_edit`.
- Open factual questions: none.