OpenAI and Microsoft Tear Up Their Exclusive Marriage — A New Multi-Cloud Era Begins

When Microsoft agreed to pour billions of dollars into a then-obscure San Francisco nonprofit back in 2019, the two parties struck a deal that was unlike anything the tech industry had seen: exclusive cloud rights, a mysterious AGI trigger clause, and a revenue-sharing arrangement that would make Microsoft one of the most financially entangled backers in Silicon Valley history. On April 27, 2026, both companies announced they were tearing most of that up — and the terms of the new arrangement reveal just how dramatically OpenAI’s position has shifted.

The End of Azure Exclusivity

The most consequential change in the restructured deal is straightforward: OpenAI is no longer bound to sell its models exclusively through Microsoft’s Azure cloud. Under the original terms, Microsoft was the preferred — and in practice, near-exclusive — distribution channel for OpenAI’s frontier models. That arrangement gave Azure a significant competitive moat, effectively meaning that any enterprise seeking direct access to GPT-series models had to route through Microsoft’s cloud infrastructure.

That moat is now gone. OpenAI can now offer its products to customers across any cloud provider, including Amazon Web Services and Google Cloud — the two largest cloud platforms in the world. The change removes what had become a growing source of tension as OpenAI scaled toward commercial viability and began building relationships with hyperscalers beyond Microsoft.

The timing is not accidental. In February 2026, Amazon announced it would invest up to $50 billion in OpenAI and expand its existing AWS agreement to $138 billion in total cloud commitments over eight years. Under the old exclusive arrangement, that deal risked putting Microsoft in legal jeopardy — the original partnership contained clauses that could have been interpreted as preventing OpenAI from signing such agreements with direct cloud competitors. The April 27 restructure resolves that ambiguity entirely, giving OpenAI a clean path forward with multiple cloud partners.

The AGI Clause Is Gone

Perhaps the most philosophically unusual element of the original deal was the so-called “AGI clause” — a provision that tied Microsoft’s commercial rights to a determination by OpenAI’s non-profit board about whether the company had achieved artificial general intelligence. If OpenAI’s board declared that AGI had been reached, Microsoft’s commercial license would revert to the non-profit, effectively stripping billions of dollars in IP access from what was supposed to be a core technology partner.

Both companies had come to regard the clause as increasingly awkward. The definition of AGI remains contested and philosophically murky, making the trigger condition legally uncertain. As OpenAI’s valuations have soared — the company was valued at $300 billion as recently as late 2025 — the clause represented a growing liability for both parties: for Microsoft, because its rights could vanish at any moment based on an internal board decision; for OpenAI, because the clause complicated its governance structure at a time when the company was converting to a for-profit model.

The restructured deal eliminates the AGI clause entirely. Microsoft’s license now covers OpenAI’s models and products through 2032 on a non-exclusive basis, with no trigger condition tied to AGI declarations.

Revenue Sharing Simplified — And Capped

The original deal required Microsoft to pay a revenue share to OpenAI on Azure-generated revenue from OpenAI products, while OpenAI simultaneously paid a share back to Microsoft. The arrangement was designed to align incentives when OpenAI was a pre-revenue nonprofit. As the company has grown into one of the most commercially valuable AI vendors on the planet, the bilateral revenue share had become increasingly difficult to justify.

Under the new terms, Microsoft will stop paying OpenAI a revenue share entirely. OpenAI will continue to pay Microsoft a revenue share through 2030, reflecting Microsoft’s ongoing role as a capital provider and preferred cloud partner — but that payment is now subject to a total cap, removing the open-ended financial exposure that had characterized the earlier arrangement.

Microsoft retains its equity stake in OpenAI, continuing to participate in the company’s upside as a major shareholder. OpenAI products will still ship first on Azure unless Microsoft cannot or chooses not to support the necessary technical capabilities — preserving Azure’s competitive advantage on the launch cadence while eliminating its commercial exclusivity.

A Rebalancing of Power

Read together, these changes tell a coherent story: OpenAI has outgrown the terms of a deal structured when it was a research nonprofit with no commercial products and no cloud infrastructure of its own. The company’s 2025 transition to a for-profit public benefit corporation, its $300 billion valuation, its expanding relationships with Amazon and Google, and its ambitions to build proprietary data center capacity all create tension with an exclusive, clause-laden partnership designed for a very different era.

The restructure benefits both sides in different ways. For OpenAI, it unlocks commercial flexibility that had been structurally unavailable since 2019: the ability to court enterprise customers on AWS and Google Cloud, to build a multi-cloud AI distribution strategy, and to remove the governance risk associated with the AGI trigger. For Microsoft, eliminating the revenue-share payment removes a multi-billion dollar liability from its books while preserving the equity upside and the Azure first-ship advantage.

The broader implication is for the cloud market. Azure has benefited enormously from its OpenAI exclusivity — analysts estimated the arrangement added measurable points of market share growth to Microsoft’s cloud division in 2024 and 2025 as enterprises flocked to Azure precisely because it was the only place to run frontier OpenAI models at scale. That advantage disappears with the new deal. AWS and Google Cloud, which have both been racing to close the frontier model gap through their own investments in Anthropic and other AI labs, will now be able to offer direct OpenAI model access alongside their own AI products.

What Stays the Same

Not everything changed. Microsoft remains OpenAI’s preferred cloud infrastructure partner, and Azure retains first-launch rights for new OpenAI models. The equity relationship is intact, meaning Microsoft continues to have a financial stake in OpenAI’s commercial success. The two companies have also indicated they will continue to collaborate on enterprise products — including Microsoft’s Copilot suite, which is deeply integrated with OpenAI’s model stack.

The partnership that once looked like a corporate acquisition in all but name has evolved into something closer to a preferred vendor relationship with equity participation. That may be a less cinematic arrangement than the original “exclusive marriage” narrative, but it is almost certainly more stable — and better suited to an AI landscape that has grown far more complex, competitive, and consequential than anyone anticipated in 2019.

The next test will come as AWS begins publicly marketing OpenAI model access. When enterprise buyers can run GPT-5-class models on three different hyperscaler clouds simultaneously, the question of where the real competitive moat lies — the model, the cloud, or the application layer — will get a definitive real-world answer.