The seven-year alliance that gave Microsoft a lock on all of OpenAI’s technology has officially been cracked open.
On April 27, 2026, OpenAI and Microsoft announced a revised partnership that ends the exclusivity arrangement that has existed at the core of their relationship since 2019, freeing OpenAI to serve all of its products to customers on any cloud provider, including Amazon Web Services (AWS) and Google Cloud.
Why the Deal Changed
The ChatGPT-maker says its growing commercial ambitions had been running into a structural ceiling, according to an internal memo seen by CNBC. As enterprise demand scaled, the Azure-only model meant OpenAI could not serve customers who operated on other cloud platforms.
However, the trigger for this deal came in February 2026, when OpenAI announced that Amazon would invest up to $50 billion in the company and agreed to expand its existing cloud agreement with AWS by $100 billion over eight years.
And now the revised deal converts Microsoft’s license to OpenAI’s intellectual property from exclusive to non-exclusive through 2032. OpenAI still has to ship its models first on Azure per the revised terms and Microsoft no longer has to pay a revenue share to OpenAI for models served through Azure. It, however, continues to receive 20% of OpenAI’s revenue until 2030, although that is now subject to a cap that both companies did not publicly disclose.
Microsoft still maintains a major equity position in OpenAI.
Amazon Moves Fast
The day after the announcement, OpenAI and AWS confirmed that GPT-5.5, GPT-4, the Codex coding agent, and Bedrock Managed Agents entered a limited preview on Amazon Bedrock.
This AWS relationship also goes deeper than model access. Under a $38 billion, seven-year cloud agreement signed in November 2025, AWS will provide Amazon EC2 UltraServers with hundreds of thousands of Nvidia GPUs and capacity to scale to tens of millions of CPUs, with all capacity targeted for deployment before the end of 2026.
Additionally, AWS holds exclusive third-party distribution rights for OpenAI Frontier. Together, the commercial agreements between the two companies now exceed $100 billion over eight years.
Google and the Broader Market Shift
Reports suggest OpenAI has already been using Google Cloud infrastructure for certain workloads, which highlights how AI companies are increasingly spreading compute across multiple providers.
Analyst Gil Luria of D.A. Davidson noted that AWS and Google Cloud enterprise customers had been limited in their ability to integrate OpenAI’s products because of the exclusive relationship, and will now be more likely to consider OpenAI alongside Anthropic and other providers.
There is also a regulatory dimension to the restructuring. Antitrust scrutiny of the original arrangement had been building on another continent, with the UK Competition and Markets Authority as well as the European Commission both opening reviews in 2024. Removing the exclusivity language deal now renders useless the most aggressive antitrust theory, that Microsoft was using its OpenAI stake to shut out cloud competition.
And now, the OpenAI that once ran and operated on a single AI relationship now runs on many.
