Microsoft allows OpenAI to use other cloud providers

Microsoft allows OpenAI to use other cloud providers

Microsoft has changed its agreement with OpenAI to allow the AI giant to use data center capacity from other cloud providers to train and run LLMs. However, Microsoft retains the right to require that this capacity be purchased from Azure first.

Microsoft recently modified its exclusive contract with OpenAI, which is valid through 2030. Previously, only Azure data centers could be deployed, but OpenAI now has the option to purchase capacity from other cloud providers as well.

Right of first refusal

In exchange for relaxing exclusivity, Microsoft has negotiated a so-called right of first refusal. This means that Microsoft gets first refusal to provide additional needed data center capacity. Only if Microsoft does not take advantage of this will OpenAI be allowed to bring in other providers.

The contract modification coincides with the announcement of the Stargate Project, an initiative of OpenAI in collaboration with Oracle and Softbank. Arm and Nvidia are also involved in the project. The Stargate Project consists of the creation of a complete AI infrastructure and data center capacity within the United States, worth as much as $500 billion.

No breach

The new terms do not indicate a possible rift between Microsoft and OpenAI. Microsoft retains strong influence within the partnership. For example, the tech giant has rights to OpenAI’s intellectual property, which it uses in its own products, such as Copilot.

In addition, Microsoft remains exclusively responsible for processing compute power requests through OpenAI’s API.

OpenAI emphasizes that the Stargate Project builds on its existing collaboration with Microsoft. The AI giant will continue to partner with the tech giant for additional computing power to train models and deliver products and services.

Read also: 500 billion Stargate AI infrastructure puts US in strong position