Moonshot AI is seeking up to 30% of revenue from managed Kimi K3 services on Azure, AWS and Google Cloud. The talks would give enterprises access to Kimi K3 through familiar cloud procurement and infrastructure, while unresolved issues include the revenue split, data access and token usage auditing.
Published byEdited with GPT-5.6 LunaImages generated with GPT Image 1.5
Research answer

Create a landscape editorial hero image for this Studio Global article: What is Moonshot AI seeking in its negotiations with Microsoft, Amazon, and Google to deploy its 2.8-trillion-parameter open-weight Kimi K3. Article summary: Moonshot is seeking to have Kimi K3 offered as a managed, usage-billed service on Azure, AWS, and Google Cloud, while retaining up to 30% of the resulting service revenue. If concluded, it could be the first major direct. Topic tags: general, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts wi
Moonshot AI is negotiating with Microsoft, Amazon and Google over agreements that would allow Kimi K3 to be offered as a managed, usage-billed service through Azure, Amazon Web Services and Google Cloud. The company is reportedly seeking up to 30% of the revenue generated by K3-related services. The discussions are still preliminary and may not produce agreements. 1
2
The proposed deals are about more than making Kimi K3 downloadable. Moonshot wants the model distributed through the enterprise cloud platforms where companies already purchase computing capacity, manage data and deploy AI applications.
Under the reported structure, the hyperscalers would host and sell access to K3-related services, while Moonshot would receive a share of the resulting revenue. The precise commercial terms have not been settled. Negotiators are still working through the revenue split, customer-data access and how token usage would be measured and audited. 1
4
If completed, the arrangement could represent the first significant direct revenue-sharing deal between a Chinese AI company and major U.S. cloud providers. That distinction matters because it would connect a Chinese-developed open-weight model to the commercial distribution systems of three of the world’s largest hyperscalers. 1
3
“Open-weight” does not necessarily mean that every company can freely resell hosted access without commercial conditions. Moonshot’s Kimi K3 License requires a model-as-a-service operator to enter a separate agreement once the licensee and its affiliates exceed $20 million in annual revenue.
That provision creates a path for broad experimentation while preserving a commercial negotiation for large inference providers. For enterprises, a managed cloud service can also be more practical than downloading and operating a model themselves: customers can buy access through an existing cloud account rather than building a separate serving, billing and monitoring stack.
Kimi K3 has been described in the supplied reporting as a 2.8-trillion-parameter model. A model of that scale strengthens the case for managed infrastructure, although the evidence provided does not substantiate every technical claim attached to K3, including specific accelerator requirements or cost comparisons. 12
The cloud deals would give Moonshot a route from model release to recurring service revenue. Instead of relying solely on companies downloading the weights or building products around them, Moonshot could participate financially when customers consume K3 through established cloud marketplaces and APIs.
That would also extend the company’s existing revenue-sharing approach beyond China. Moonshot reportedly had a revenue-sharing and joint-innovation arrangement with Chinasoft International, although the split was not disclosed. A deal with Azure, AWS or Google Cloud would put the model in front of a much broader overseas enterprise audience. 1
The talks also arrive as Moonshot prepares for a potential public listing. Reuters reported that the company had raised more than $2 billion and was preparing for a possible Hong Kong listing. A signed cloud contract could support an argument that Moonshot has a repeatable commercial model, but the evidence provided does not establish a valuation of $50 billion. 1
The negotiations have a geopolitical complication. They come amid U.S. scrutiny of Moonshot over allegations involving benefits obtained from Anthropic’s Fable and the acquisition of Nvidia chips. Moonshot has rejected the distillation allegation and attributed K3’s performance to its own architectural work. 1
That creates a tension at the center of the proposed business: Moonshot is seeking formal, auditable distribution through U.S. cloud companies while facing allegations that could affect export-control or national-security decisions. The supplied evidence does not verify the more specific claim that the disputed hardware consisted of Nvidia GB300 servers, so that detail should not be treated as established.
For Microsoft, Amazon and Google, any partnership would therefore involve more than evaluating K3’s technical performance. They would also need to assess licensing, compliance, data handling and the policy risks of hosting a Chinese-developed model for enterprise customers.
The talks point to a changing business model for open-weight AI. Models can be released broadly to encourage adoption while their developers retain leverage over high-volume hosted inference, particularly when the license requires large model-as-a-service providers to negotiate commercial terms.
That approach sits between two familiar strategies: giving away model weights to maximize distribution and selling a fully controlled proprietary API. A cloud revenue-sharing agreement could offer Moonshot the reach of the first strategy with some of the recurring economics of the second.
The reported discussions also align with broader efforts by Chinese AI companies to monetize open models through commercial partnerships. Reuters separately reported that Alibaba was considering revenue-sharing terms with major users of its next open-source model. That suggests the Kimi K3 structure may become a template worth watching, although it is too early to know whether any of these proposed arrangements will close.
For now, the key fact is not that Kimi K3 has secured a place on Azure, AWS or Google Cloud. It has not. The important development is that Moonshot is trying to make open-weight distribution, cloud infrastructure and revenue sharing work together—despite the technical, contractual and geopolitical obstacles.
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Moonshot AI is seeking up to 30% of revenue from managed Kimi K3 services on Azure, AWS and Google Cloud.
Moonshot AI is seeking up to 30% of revenue from managed Kimi K3 services on Azure, AWS and Google Cloud. The talks would give enterprises access to Kimi K3 through familiar cloud procurement and infrastructure, while unresolved issues include the revenue split, data access and token usage auditing.
Kimi K3’s license requires a separate commercial agreement for qualifying model as a service providers whose annual revenue exceeds $20 million, making cloud partnerships central to Moonshot’s monetization strategy.
Moonshot AI is seeking up to 30% of revenue from managed Kimi K3 services on Azure, AWS and Google Cloud. The talks would give enterprises access to Kimi K3 through familiar cloud procurement and infrastructure, while unresolved issues include the revenue split, data access and token usage auditing.
Published byEdited with GPT-5.6 LunaImages generated with GPT Image 1.5
Research answer

Create a landscape editorial hero image for this Studio Global article: What is Moonshot AI seeking in its negotiations with Microsoft, Amazon, and Google to deploy its 2.8-trillion-parameter open-weight Kimi K3. Article summary: Moonshot is seeking to have Kimi K3 offered as a managed, usage-billed service on Azure, AWS, and Google Cloud, while retaining up to 30% of the resulting service revenue. If concluded, it could be the first major direct. Topic tags: general, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts wi
Moonshot AI is negotiating with Microsoft, Amazon and Google over agreements that would allow Kimi K3 to be offered as a managed, usage-billed service through Azure, Amazon Web Services and Google Cloud. The company is reportedly seeking up to 30% of the revenue generated by K3-related services. The discussions are still preliminary and may not produce agreements. 1
2
The proposed deals are about more than making Kimi K3 downloadable. Moonshot wants the model distributed through the enterprise cloud platforms where companies already purchase computing capacity, manage data and deploy AI applications.
Under the reported structure, the hyperscalers would host and sell access to K3-related services, while Moonshot would receive a share of the resulting revenue. The precise commercial terms have not been settled. Negotiators are still working through the revenue split, customer-data access and how token usage would be measured and audited. 1
4
If completed, the arrangement could represent the first significant direct revenue-sharing deal between a Chinese AI company and major U.S. cloud providers. That distinction matters because it would connect a Chinese-developed open-weight model to the commercial distribution systems of three of the world’s largest hyperscalers. 1
3
“Open-weight” does not necessarily mean that every company can freely resell hosted access without commercial conditions. Moonshot’s Kimi K3 License requires a model-as-a-service operator to enter a separate agreement once the licensee and its affiliates exceed $20 million in annual revenue.
That provision creates a path for broad experimentation while preserving a commercial negotiation for large inference providers. For enterprises, a managed cloud service can also be more practical than downloading and operating a model themselves: customers can buy access through an existing cloud account rather than building a separate serving, billing and monitoring stack.
Kimi K3 has been described in the supplied reporting as a 2.8-trillion-parameter model. A model of that scale strengthens the case for managed infrastructure, although the evidence provided does not substantiate every technical claim attached to K3, including specific accelerator requirements or cost comparisons. 12
The cloud deals would give Moonshot a route from model release to recurring service revenue. Instead of relying solely on companies downloading the weights or building products around them, Moonshot could participate financially when customers consume K3 through established cloud marketplaces and APIs.
That would also extend the company’s existing revenue-sharing approach beyond China. Moonshot reportedly had a revenue-sharing and joint-innovation arrangement with Chinasoft International, although the split was not disclosed. A deal with Azure, AWS or Google Cloud would put the model in front of a much broader overseas enterprise audience. 1
The talks also arrive as Moonshot prepares for a potential public listing. Reuters reported that the company had raised more than $2 billion and was preparing for a possible Hong Kong listing. A signed cloud contract could support an argument that Moonshot has a repeatable commercial model, but the evidence provided does not establish a valuation of $50 billion. 1
The negotiations have a geopolitical complication. They come amid U.S. scrutiny of Moonshot over allegations involving benefits obtained from Anthropic’s Fable and the acquisition of Nvidia chips. Moonshot has rejected the distillation allegation and attributed K3’s performance to its own architectural work. 1
That creates a tension at the center of the proposed business: Moonshot is seeking formal, auditable distribution through U.S. cloud companies while facing allegations that could affect export-control or national-security decisions. The supplied evidence does not verify the more specific claim that the disputed hardware consisted of Nvidia GB300 servers, so that detail should not be treated as established.
For Microsoft, Amazon and Google, any partnership would therefore involve more than evaluating K3’s technical performance. They would also need to assess licensing, compliance, data handling and the policy risks of hosting a Chinese-developed model for enterprise customers.
The talks point to a changing business model for open-weight AI. Models can be released broadly to encourage adoption while their developers retain leverage over high-volume hosted inference, particularly when the license requires large model-as-a-service providers to negotiate commercial terms.
That approach sits between two familiar strategies: giving away model weights to maximize distribution and selling a fully controlled proprietary API. A cloud revenue-sharing agreement could offer Moonshot the reach of the first strategy with some of the recurring economics of the second.
The reported discussions also align with broader efforts by Chinese AI companies to monetize open models through commercial partnerships. Reuters separately reported that Alibaba was considering revenue-sharing terms with major users of its next open-source model. That suggests the Kimi K3 structure may become a template worth watching, although it is too early to know whether any of these proposed arrangements will close.
For now, the key fact is not that Kimi K3 has secured a place on Azure, AWS or Google Cloud. It has not. The important development is that Moonshot is trying to make open-weight distribution, cloud infrastructure and revenue sharing work together—despite the technical, contractual and geopolitical obstacles.
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Moonshot AI is seeking up to 30% of revenue from managed Kimi K3 services on Azure, AWS and Google Cloud.
Moonshot AI is seeking up to 30% of revenue from managed Kimi K3 services on Azure, AWS and Google Cloud. The talks would give enterprises access to Kimi K3 through familiar cloud procurement and infrastructure, while unresolved issues include the revenue split, data access and token usage auditing.
Kimi K3’s license requires a separate commercial agreement for qualifying model as a service providers whose annual revenue exceeds $20 million, making cloud partnerships central to Moonshot’s monetization strategy.