Microsoft keeps exclusive rights to key OpenAI technology, despite new flexibility

In short: OpenAI has more freedom to use other cloud providers, but Microsoft still holds exclusive rights to license key OpenAI technology.

What happened

The New York Times reported that Microsoft will keep licensing OpenAI’s technology, but will no longer be OpenAI’s exclusive licensee. At the same time, Microsoft’s own partnership updates from late 2025 and early 2026 say Microsoft remains the exclusive licensee for several core pieces.

Those exclusive rights include OpenAI’s most advanced “frontier” models, plus OpenAI’s intellectual property, meaning the underlying designs and key components that make the models work (like a recipe and the secret steps). They also include OpenAI’s stateless APIs, which are the standard ways other companies connect software to OpenAI models (like a plug that lets one app talk to another), hosted on Microsoft Azure.

What did change is where OpenAI can get computing power. Before, Microsoft was the only cloud provider for training and running OpenAI models. Now OpenAI can use multiple clouds for some research, training, and non-API products. But if OpenAI sells API access through third parties, those API services still have to run on Azure, and Microsoft keeps revenue shares in some third-party deals.

Why it matters

This keeps Microsoft in a strong position for selling OpenAI-based services to businesses, even as OpenAI looks for more places to rent the expensive computing power needed to build and run new models. For customers, it means OpenAI’s behind-the-scenes infrastructure may spread out, but the main “gateway” that many companies use to access OpenAI models stays closely tied to Microsoft.

Source: NYTimes