OpenAI and Microsoft announced on April 27 that they have renegotiated the foundational terms of their partnership, ending a period of mounting tension that had threatened to produce a legal dispute over OpenAI's expanding relationships with Microsoft's biggest cloud competitors.
The headline change is the end of exclusivity. Under the previous arrangement, Microsoft held exclusive rights to OpenAI's intellectual property and to hosting OpenAI's products through APIs. That restriction was becoming untenable. In February, OpenAI announced an investment of up to $50 billion from Amazon, including an agreement to give AWS exclusive rights to host Frontier, OpenAI's new agent platform. Microsoft responded publicly that the deal violated their existing terms, and the Financial Times later reported that Microsoft had considered legal action to enforce them.
The new agreement resolves that conflict directly. Microsoft's license to OpenAI's IP through 2032 is now non-exclusive. OpenAI can serve any of its products on any cloud platform. Microsoft remains the "primary" cloud partner, with OpenAI products shipping first on Azure unless Microsoft cannot or chooses not to support the necessary capabilities, but "first" no longer means "exclusive." Amazon CEO Andy Jassy celebrated the deal on Monday, noting that OpenAI's models would now be available to AWS customers through the Bedrock service in the coming weeks.
The financial terms shifted as well. Microsoft will no longer pay a revenue share to OpenAI, a flow that had existed for years tied to Azure usage. OpenAI will continue paying a 20% revenue share to Microsoft through 2030, but that payment is now subject to a total cap. Given that Microsoft reported $7.5 billion from its OpenAI investment in a single quarter last year, the cap represents a meaningful constraint on the ceiling of what Microsoft collects. Microsoft retains approximately a 27% stake in OpenAI's for-profit entity, so it continues to benefit financially from OpenAI's growth regardless of which cloud platform runs its products.
The renegotiation reflects a broader realignment in the AI industry. OpenAI is no longer a startup entirely dependent on a single corporate backer. It has attracted investment commitments in the hundreds of billions of dollars from Amazon, SoftBank, and others, and is building out its own data center infrastructure. The question of which cloud platform runs its products is increasingly a business decision, not a survival question. At the same time, Microsoft has developed a closer relationship with Anthropic and is investing heavily in its own AI capabilities, reducing its dependency on OpenAI as its primary AI differentiator.
What the deal also clarifies is a timeline question that had hung over the partnership from the beginning. The previous arrangement had been structured around OpenAI's pursuit of artificial general intelligence, with Microsoft's exclusivity rights lasting until OpenAI determined it had achieved AGI. That framing made the partnership open-ended in a way that suited neither company as the commercial stakes grew. The new deal replaces it with a calendar endpoint: Microsoft's license runs to 2032, full stop, without reference to AGI milestones.
For enterprise customers, the outcome is straightforwardly positive. Companies can now access OpenAI's models through Azure, AWS, or Google Cloud without being forced to choose a platform based on which company holds the exclusive contract. The competitive dynamic this creates, with cloud providers bidding to offer the best infrastructure for AI workloads, is likely to produce better pricing and more flexibility over time. Whether the underlying relationship between OpenAI and Microsoft is as "strong and central" as both companies continue to insist is a separate question that the next few years of cloud competition will answer.