Moonshot AI presses for revenue-sharing terms with major cloud rivals, report says

Moonshot AI, a Beijing-based artificial intelligence startup, is reportedly in talks with major cloud providers about revenue-sharing arrangements tied to its Kimi K3-related services. A report from Yahoo Finance, published via Quartz, said the startup wants terms that could give it “up to 30% of revenue” generated from those services running on Microsoft’s Azure, Amazon Web Services (AWS), and Google Cloud.

The proposal, as described in the report, centers on distributing a portion of the customer revenue that flows through each cloud platform for using Moonshot AI’s Kimi K3 offerings. In plain terms, the startup is aiming to align its earnings with how much customers spend to access its model-driven services when deployed on the dominant cloud infrastructures.

Kimi K3, as referenced in the report, is the latest version of Moonshot AI’s model family, and the company’s strategy appears to rely on getting its offerings adopted through mainstream cloud channels. The report characterizes the talks as part of a broader effort to scale distribution for Kimi K3-related services, rather than limiting growth to direct enterprise customers or smaller hosting arrangements.

The report did not identify the specific deal structures under discussion, nor did it provide contract timelines, pricing mechanics, or whether Moonshot would share any usage-related costs with the cloud providers. It also did not confirm whether the proposals are limited to particular regions, customer segments, or model features, such as chat or inference-only access.

For Microsoft, the prospect is notable because Azure remains one of the core platforms where AI model providers try to monetize demand. For Amazon and Google, similar incentives exist, but the same dynamic applies: cloud providers typically seek predictable margins for compute and platform services, while model vendors seek ways to convert technical adoption into revenue and more direct economics.

Moonshot AI’s willingness to tie its potential earnings to cloud-generated sales suggests it is treating distribution through hyperscalers as a primary growth lever. Revenue sharing in this context can also function as a hedge, since it reduces the startup’s dependence on fixed fees or one-off licensing while potentially increasing its stake in how aggressively a cloud provider markets or sells access to the model.

The report also framed these efforts around negotiations with multiple hyperscalers, which could indicate Moonshot is playing providers against one another to reach favorable terms. However, it remains unclear from the coverage how far talks have progressed, whether either side has committed to pilot programs, or whether any agreement includes performance or compliance conditions given how model usage typically intersects with cloud security and governance requirements.

Microsoft did not disclose any details in the published report, and the coverage did not describe official responses from Amazon or Google. That leaves key questions unanswered, including the final percentage, the exact revenue base (for example, whether it is tied to compute, API calls, or bundled service subscriptions), and whether Moonshot’s Kimi K3-related offerings would be offered under the cloud providers’ marketplaces, custom enterprise contracts, or partner programs.