Why OpenAI Expands Beyond Microsoft with a US$38bn AWS Deal

The continuous advancement of AI technology has created immense demand for computing power. 

As frontier model providers push their models to new heights of intelligence, they’re increasingly turning to diversified infrastructure strategies – and OpenAI’s latest move exemplifies this change.

The AI giant and AWS have now sealed a US$38bn deal, giving the ChatGPT maker immediate access to AWS cloud infrastructure at scale. 

Amazon EC2 UltraServers powered by hundreds of thousands of Nvidia GPUs will now run Open AI’s core AI workloads with the capacity to scale to tens of millions of CPUs. 

Since AWS has experience running large-scale AI infrastructure clusters topping 500K chips, the clusters support workloads from ChatGPT inference to training next-generation models, with low-latency performance across interconnected systems.

As a result of its historic restructuring, Open AI, is no longer bound to Microsoft as its sole cloud provider, making this 7 year deal with AWS strategically possible. 

“Scaling frontier AI requires massive, reliable compute," says OpenAI Co-founder and CEO Sam Altman. 

“Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”

Why OpenAI turns to AWS for massive compute muscle

The OpenAI partnership with AWS positions the two companies closer together and ahead of Anthropic, which has a US$8bn investment from AWS.

Earlier this year, OpenAI’s open-weight models became available on Amazon Bedrock, bringing thousands of customers who now use them for coding, scientific analysis and mathematical problem solving into AWS’s ecosystem.

“As OpenAI continues to push the boundaries of what's possible, AWS's best-in-class infrastructure will serve as a backbone for their AI ambitions,” says Matt Garman, CEO of AWS. 

“The breadth and immediate availability of optimised compute demonstrates why AWS is uniquely positioned to support OpenAI's vast AI workloads.”

An age of multi-cloud partnerships

The new deal with AWS follows a broader pattern of infrastructure diversification across frontier AI labs. 

OpenAI had earlier contracted to incrementally purchase US$250bn in Microsoft Azure services.

The company has also leaned on CoreWeave, a specialised cloud provider, with around US$22.4bn in total contract value to train its AI models.

The AWS partnership continues this move toward a multi-cloud, multi-supplier compute strategy designed to prevent over-reliance on a single vendor, a shift made possible after OpenAI’s restructuring into a PBC.

As the race to train larger and more capable models intensifies, access to compute has become the defining competitive advantage.

OpenAI’s new deal with AWS marks not just another supplier partnership, but a structural shift in how frontier AI is built and distributed.

This shows that the era of one-cloud dominance is giving way to a more reliable, collaborative scaling.