OpenAI on AWS Bedrock: Microsoft Exclusivity Is Over
3 min readOpenAI’s models are now available on Amazon Web Services for the first time, just one day after the company restructured its long-running partnership with Microsoft to remove an exclusive cloud hosting arrangement. The move reshapes how developers and enterprises will access frontier AI going forward.
A Deal Seven Years in the Making
When Microsoft first invested $1 billion in OpenAI back in 2019, it secured exclusive rights to host OpenAI’s models on its Azure cloud platform. That arrangement deepened over the years as Microsoft committed tens of billions more to the company. Until this week, any organization wanting to deploy OpenAI models in a managed cloud environment had one option: Azure. For enterprise customers already running infrastructure on AWS, that created real friction.
What Changed This Week
On April 27, 2026, Microsoft and OpenAI announced a restructured agreement that ends the exclusivity. Under the new terms, Microsoft receives a nonexclusive license to OpenAI’s IP and models through 2032. OpenAI will continue paying a revenue share to Microsoft through 2030, subject to a cap. Neither side framed the change as a breakup. Microsoft called it a move toward “flexibility and certainty,” while OpenAI described it as a natural evolution of the partnership.
The very next day, CNBC reported that AWS and OpenAI had formalized an expanded partnership. OpenAI’s flagship frontier models, including the recently released GPT-5.5, are now available through Amazon Bedrock. AWS customers also gain access to OpenAI’s Codex agent for writing and debugging code, all through the same platform. A jointly built agent service is part of the launch as well.
Why This Matters for AI Developers
For developers and engineering teams, this is a meaningful unlock. Companies that run workloads on AWS no longer need to route through Azure to use OpenAI models. That removes architectural friction and opens up new options for building AI-powered applications within existing cloud environments.
For the broader AI industry, the shift signals that OpenAI is moving toward a multi-cloud strategy as it works to reach a wider base of enterprise customers. It also intensifies competition between Azure and AWS as the leading platform for production AI workloads. The battle for AI infrastructure is no longer a foregone conclusion in Microsoft’s favor.
What to Watch Next
The Microsoft-OpenAI relationship is not ending. Microsoft retains a nonexclusive license through 2032 and remains a major backer. But the partnership has matured from a tightly exclusive arrangement into something closer to a standard enterprise deal. Expect Google Cloud and other providers to push hard for their own OpenAI integrations in the coming months. The race to become the preferred home for frontier AI models is now fully open.