OpenAI removes Microsoft’s legal risk tied to $50B Amazon partnership

Microsoft and OpenAI have announced a revised agreement governing their long-standing partnership, introducing updated terms that reshape how the two companies collaborate. While some commentary has framed the change as a win for OpenAI, the outcome provides advantages for both sides.

A central element of the updated agreement is that it resolves a major legal complication tied to OpenAI’s separate partnership with Amazon, which includes a potential investment of up to $50 billion.

Under the revised terms, Microsoft will no longer retain exclusive access to all of OpenAI’s intellectual property until the achievement of artificial general intelligence (AGI). Instead, the partnership now operates on a defined timeline, granting Microsoft a nonexclusive license to OpenAI’s models and products through 2032.

Despite this shift, Microsoft will continue to be described as OpenAI’s “primary cloud partner,” meaning that a significant portion of OpenAI’s infrastructure is still expected to run on Azure over the coming years. OpenAI has also committed to substantial cloud spending with Microsoft, including a previously announced agreement to purchase an additional $250 billion in Azure services.

The companies stated that OpenAI’s products will launch “first on Azure,” unless Microsoft is unable or chooses not to support required capabilities. At the same time, OpenAI is now free to distribute its products across multiple cloud providers, removing earlier restrictions that limited such flexibility.

This change directly addresses the legal tension that arose following OpenAI’s agreement with Amazon. In February, OpenAI disclosed that Amazon planned to invest up to $50 billion, consisting of an initial $15 billion and an additional $35 billion contingent on certain conditions. As part of that arrangement, OpenAI agreed to collaborate on “stateful runtime technology” through AWS Bedrock, a platform that supports AI services and models.

Stateful runtime technology enables AI systems, particularly agents, to retain memory and context over extended interactions. OpenAI also agreed to grant AWS exclusive rights to host its Frontier agent development platform, a move that became a point of conflict.

Under the previous Microsoft agreement, OpenAI was restricted from offering certain products — especially those accessed via APIs — exclusively through other cloud providers. Microsoft had maintained exclusive rights to host such services on Azure, including products like Frontier. On the same day OpenAI announced its AWS deal, Microsoft publicly emphasised that its exclusivity terms remained in effect, stating that Azure would continue to host all stateless API access to OpenAI models, even in collaborations involving third parties.

Reports at the time indicated that Microsoft had considered legal action to enforce those terms.

The newly negotiated agreement removes Microsoft’s exclusivity, eliminating the risk of a legal dispute over OpenAI’s AWS partnership. Andy Jassy welcomed the development, noting that it would make OpenAI’s models available through AWS Bedrock.

While the revised terms benefit OpenAI by expanding its flexibility, Microsoft also secures important advantages. The company will no longer be required to share revenue with OpenAI, while OpenAI will continue to make revenue-sharing payments to Microsoft through 2030, subject to a cap.

Although the exact financial impact is not disclosed, Microsoft has previously reported significant returns from its investment in OpenAI, including $7.5 billion in a single quarter. Microsoft also remains a major shareholder, holding approximately 27% of OpenAI’s for-profit entity, meaning it continues to benefit from the company’s overall growth — including revenue generated through competing cloud platforms.

The primary trade-off for Microsoft is the loss of exclusivity, which could have driven additional demand for Azure services. However, the company has also strengthened relationships with other AI providers, including Anthropic, positioning itself to support alternative models and services.

For enterprise customers, the updated agreement introduces greater flexibility, allowing organisations to choose their preferred AI models and cloud environments, while major providers compete to deliver services.

Recent developments in the Microsoft–OpenAI relationship illustrate how the partnership has evolved:

• In October, the two companies updated their agreement to help OpenAI address legal challenges related to its corporate structure, allowing certain products to operate on non-Azure platforms.
• In November, OpenAI and Amazon entered into a multi-year agreement involving $38 billion in AWS cloud services.
• In February, Amazon announced a potential $50 billion investment in OpenAI, tied to exclusive technology development and hosting arrangements, prompting Microsoft to reaffirm its contractual rights.
• In March, reports indicated that Microsoft was considering legal action over the dispute.
• In April, Microsoft and OpenAI finalised an agreement that sets a defined timeline for their partnership, removes exclusivity restrictions, and enables OpenAI to operate across multiple cloud providers while maintaining Microsoft as a key partner and shareholder.