Microsoft and OpenAI end exclusivity: OpenAI can now sell AI through AWS and other clouds

The Microsoft-OpenAI alliance has not collapsed, but it has ceased to be exclusive — and this is one of the most significant shifts in generative AI infrastructure in recent years. The companies have renegotiated their partnership terms so that OpenAI can now sell its products through Microsoft's competitors' clouds, including Amazon Web Services and Google Cloud. For the market, this signals that one of the largest AI developers is no longer locked into a single ecosystem, while Microsoft no longer controls the entire commercial flow around OpenAI's models.

Under the new terms, Azure remains OpenAI's primary cloud partner, and OpenAI products must still launch first on Microsoft's platform — if Azure can deliver the required capabilities. But the key change is different: OpenAI has gained the right to serve customers through any cloud provider.

At the same time, Microsoft retains a license to OpenAI's intellectual property until 2032, but this license is now non-exclusive. The money flows have also changed: Microsoft will no longer pay OpenAI a share of the revenue from AI products it sells, while OpenAI will continue to share part of its revenue with Microsoft until 2030, but now within an established cap.

This move continued a gradual renegotiation of relationships that until recently were considered almost inseparable. In autumn 2025, the companies had already loosened several restrictions in the agreement, giving OpenAI more space to attract capital, work with third parties, and purchase computing power.

Pressure on the old structure intensified after OpenAI began expanding its infrastructure and commercial ties with other partners, primarily Amazon. For OpenAI, this is a matter of scale: demand for models, enterprise services, and computing resources is growing faster than a single platform can meet without limitations on capacity, timelines, and price.

It is also important that the deal reflects a changed balance of power. When ChatGPT was just taking off, Microsoft was OpenAI's primary source of capital, computing, and enterprise sales channels. In 2026, the situation is already different: OpenAI is building a multi-partner infrastructure, while Microsoft itself is simultaneously investing in its own models and products to avoid depending solely on a single external supplier.

The waiver of the obligation to pay OpenAI a share of its revenue and clearer fixing of reciprocal payments until 2030 make the alliance less financially entangled and likely more manageable for both sides.

For OpenAI, the new structure removes an important barrier in enterprise sales. Large customers dislike switching their base cloud provider just for one AI service, and the previous Azure exclusivity made it difficult to reach AWS and Google Cloud customers. Now OpenAI will be able to sell its models and tools where these companies are already deploying the rest of their infrastructure.

For Microsoft, the decision looks more mixed, but not necessarily disadvantageous. The company loses some of its exclusive advantage, but retains the status of primary partner, access to OpenAI's technology, and a position as a major shareholder. Moreover, loosening the tie may reduce regulatory pressure: in the United States, the UK, and Europe, the Microsoft-OpenAI alliance has long been viewed as a potential source of excessive influence on the cloud and enterprise AI markets.

The main conclusion is simple: the era when OpenAI was almost a built-in extension of Azure ended on April 27, 2026. Now Microsoft and OpenAI's relationship looks less like a monopolistic alliance and more like a strategic partnership with clear boundaries of interests.

OpenAI gains freedom to scale its business and prepare the ground for further growth, possibly including a public offering. Microsoft, in turn, reduces its dependence on a single model supplier and can more precisely allocate capital between Azure, Copilot, and its own AI development.

For the entire market, this means intensified competition between clouds and a more open struggle for the enterprise customer.