OpenAI Ends Microsoft Exclusivity, Expands AWS Partnership — New Multi-Cloud Era Begins

On April 29, OpenAI announced a restructured partnership with Microsoft, officially ending the previous exclusive cloud service agreement. This change allows OpenAI to expand its AI models to Amazon Web Services (AWS) and Google Cloud Platform, marking a shift from a single-cloud dependency to a multi-cloud strategy. At the same time, Microsoft will no longer pay OpenAI a revenue share based on model sales through Azure, while OpenAI gains full autonomy over its own model sales — no longer requiring Microsoft's permission to deploy and generate revenue on AWS and Google Cloud.

Under the new agreement, Microsoft retains all revenue generated from selling OpenAI models via Azure, while OpenAI is freed from prior exclusivity constraints. AWS stated in a blog post: "This marks the beginning of a deeper collaboration between AWS and OpenAI." The two parties will continue integrating OpenAI's frontier models on Amazon Bedrock and jointly develop production-grade AI agent solutions based on Codex. Meanwhile, Google Cloud is actively evaluating the new agreement terms, with potential multi-cloud partnerships in the works.

Notably, in February of this year, Amazon announced a $50 billion investment in OpenAI. In exchange, OpenAI will adopt AWS's custom Trainium chips on a large scale for model training. Although OpenAI's revenue chief Denise Dresser claimed that this collaboration is "unrelated" to the Microsoft restructuring, industry analysts widely view these moves as a key step in OpenAI's accelerated pivot away from the Microsoft cloud ecosystem and toward full integration with AWS.

Looking ahead: OpenAI's "de-Microsoft" strategy is set to reshape the competitive landscape of AI cloud services. A multi-cloud architecture not only reduces single-point risk but also leverages the computing power and ecosystem advantages of different cloud platforms to accelerate commercialization. For enterprise customers, this means more flexibility to call models like GPT across multiple clouds, while competition between AWS and Google Cloud is expected to intensify further.