Microsoft and OpenAI have restructured one of the most consequential commercial partnerships in the history of the technology industry. The revised agreement, announced April 28, ends Microsoft’s exclusive right to distribute OpenAI models — freeing OpenAI to sell its products on Amazon Web Services, Google Cloud, and Oracle. Within 24 hours of the announcement, AWS launched a preview of OpenAI’s models on Amazon Bedrock. The deal also removes the AGI escape clause that had been a long-standing source of tension between the two companies.
What Changed in the Deal
The original Microsoft-OpenAI agreement gave Microsoft an exclusive license to all of OpenAI’s models and products through an indefinite period tied to the achievement of artificial general intelligence — a provision that effectively locked OpenAI into Azure as its sole commercial cloud channel until a condition that may never occur. The revised terms replace that open-ended arrangement with a clear structure:
- Microsoft’s license is now non-exclusive — OpenAI can sell through any cloud provider
- Azure remains OpenAI’s primary cloud partner — products ship on Azure first unless Microsoft “cannot or chooses not” to support the capability
- The AGI escape clause is removed — the agreement now runs through 2032 regardless of AI capability milestones
- Microsoft no longer pays OpenAI a revenue share for distributing its models on Azure
- OpenAI continues paying Microsoft a 20% revenue share through 2030, now subject to an undisclosed cap
- Microsoft retains non-exclusive IP rights to OpenAI models and products through 2032
The deal resolved a legal standoff that had been building for months. When OpenAI announced its $50 billion Amazon investment in February, Microsoft publicly disputed that AWS would have exclusive access to certain OpenAI products. The Financial Times reported in March that Microsoft was considering legal action. The revised agreement makes both the legal threat and the technical dispute moot.
AWS Moved Within 24 Hours
Amazon CEO Andy Jassy posted on LinkedIn the same day as the announcement: “We’re excited to make OpenAI’s models available directly to customers on Bedrock in the coming weeks.” The following day, at an event in San Francisco, AWS launched a preview of OpenAI models on Amazon Bedrock — alongside the unveiling of Amazon Bedrock Managed Agents powered by OpenAI.
The Bedrock product is the enterprise agent platform previously described as a “Stateful Runtime Environment” — allowing businesses to build agents that maintain context across sessions, remember previous interactions, and operate continuously within AWS’s security and compliance infrastructure. OpenAI CEO Sam Altman appeared via recorded video at the launch event, noting his “schedule got taken away” from him — a reference to the Elon Musk trial proceeding across the Bay in Oakland at the same time.
OpenAI’s internal communications, surfaced in reporting, indicate that inbound demand for the AWS offering had been described as “frankly staggering” — and that the Microsoft exclusivity had been “limiting our ability to meet enterprises where they are.” Most large enterprises run their core infrastructure on AWS, not Azure. OpenAI’s previous Azure exclusivity meant those enterprises had to choose between their existing cloud infrastructure and access to OpenAI’s most capable models via API.
Why Both Sides Won
This deal is unusual in that it genuinely works for both companies. OpenAI’s win is clear: it gains access to AWS’s enterprise customer base without abandoning Azure, and it removes the AGI clause that had been described as an open-ended risk. Microsoft’s win is less obvious but equally real: it is relieved of paying OpenAI a revenue share on Azure, which materially improves Azure’s profit margins from OpenAI-related traffic. Microsoft retains a 20% revenue share on OpenAI’s total revenue — which grows as OpenAI sells through AWS and other clouds, not just Azure. And Microsoft continues as a major OpenAI shareholder, so its equity upside from OpenAI’s growth is unaffected by the cloud channel change.
Barclays analysts noted that Microsoft “no longer needs to underwrite all of OpenAI’s data center costs” while retaining the financial upside from OpenAI’s success across all channels. The deal is a clean separation of cloud hosting economics from equity and revenue participation.
What It Means for Enterprise AI Buyers
The practical implication for developers and enterprises is significant: OpenAI models are now available through the cloud provider most enterprises already use. AWS customers no longer need to route API calls to Azure to access ChatGPT or Codex. Bedrock Managed Agents gives them a managed infrastructure for building production agents that stay within AWS’s security perimeter — directly competitive with Anthropic’s Claude Managed Agents, which has been available on Bedrock since last year.
Conclusion
The Microsoft-OpenAI restructuring is one of the most significant commercial AI deals of 2026 — not for the model capabilities it enables, but for the distribution it unlocks. OpenAI is now everywhere its models need to be. Browse our full directory to explore ChatGPT, Claude, and every AI tool now available across the major cloud platforms competing for enterprise AI workloads.
