Microsoft, OpenAI Revise Partnership, Allowing Multi‑Cloud AI Model Access

Microsoft unveiled a major amendment to its partnership with OpenAI on Monday, a move that fundamentally changes how the two giants will collaborate on artificial‑intelligence services. The new agreement lifts the exclusivity that once bound OpenAI’s models to Azure, permitting the startup to make its products available on any cloud platform, including Amazon Web Services. In practice, OpenAI can now roll out its latest models, such as Codex, to AWS minutes after they appear on Azure.

While the deal opens the door for rivals, Microsoft retains a solid financial foothold. The tech behemoth will continue to receive 20 percent of the revenue OpenAI generates from ChatGPT and its API platform, a share that now also includes earnings from sales on competing clouds. The revenue‑sharing arrangement is one‑way, meaning Microsoft no longer pays a cut of its own Azure OpenAI earnings back to OpenAI.

Microsoft’s role as OpenAI’s “primary cloud partner” remains intact. Azure will still be the first launch venue for new OpenAI products, and Microsoft will ship the latest models ahead of rivals. However, the non‑exclusive license now runs through 2032, extending the previous 2030 deadline by two years. The amendment also eliminates a controversial clause that tied Microsoft’s access to OpenAI’s models to the development of artificial general intelligence (AGI). Without that clause, Microsoft will not lose access to future breakthrough models if OpenAI ever declares AGI.

The financial stakes are significant. Microsoft holds roughly 27 percent of OpenAI’s for‑profit arm, cementing its status as a major shareholder. The revised revenue‑share model ensures that Microsoft continues to benefit from OpenAI’s expanding market, even as the startup diversifies its cloud partnerships.

Sources say the renegotiation follows a period of friction marked by executive disagreements, contract reshuffling, and concerns over AI infrastructure. OpenAI’s leadership had previously warned that the Microsoft deal limited the company’s ability to reach enterprises that favor other cloud providers, particularly Amazon’s Bedrock service. The tension peaked when OpenAI considered a $3 billion acquisition of the AI coding tool Windsurf, a move that threatened to upset the existing partnership.

Both companies appear to have reached a pragmatic compromise. By allowing OpenAI to sell on rival clouds, Microsoft reduces the risk of losing the startup to competitors while still securing a revenue stream and maintaining early‑access privileges. OpenAI, for its part, gains the flexibility to serve a broader enterprise base without sacrificing its primary relationship with Microsoft.

The amendment also signals a shift in how the two firms view future AI development. Microsoft has been quietly advancing its own models, such as the MAI‑Transcribe‑1 speech‑to‑text system, and it has explored using Anthropic’s Claude and Google’s Gemini models for specific tasks. The new deal suggests a more diversified AI strategy, where Microsoft blends internally built models with OpenAI’s offerings and third‑party solutions to stay competitive.

Industry observers note that the updated contract could reshape the broader AI cloud market. Amazon, which recently announced a $50 billion partnership with OpenAI, now has a clear path to host the same models that Microsoft once owned exclusively. Google may also see opportunities to integrate OpenAI’s technology into its Gemini Enterprise Agent Platform in the future.

Overall, the revised partnership reflects a balance between collaboration and competition. Microsoft secures a continued revenue share and early‑access rights, while OpenAI expands its market reach across cloud providers. Both firms walk away with a clearer, financially driven relationship that sidesteps the earlier, more contentious clauses that had clouded their cooperation.