Amazon and OpenAI expand their cooperation: models are added to Bedrock, and Microsoft’s exclusive deal ends

OpenAI announced on May 3 that it will expand its partnership with Amazon Web Services (AWS), and that OpenAI models and Codex coding agents will be made available to AWS customers via Amazon Bedrock. According to a report by CNBC, this expansion is a key next step after OpenAI and Microsoft ended their cloud exclusive deal in late April—moving OpenAI from “Microsoft cloud only” to multi-cloud deployment. AWS has also completed initial integration of OpenAI products on the Bedrock platform during the same period.

Background: OpenAI–Microsoft exclusive deal ends; IP licensing extended to 2032

In late April, OpenAI and Microsoft reached a new agreement: the original structure of “Microsoft’s exclusive access to OpenAI products and IP” ended, and OpenAI gained the right to operate all its products on any cloud. Microsoft will still hold IP licensing rights for OpenAI models and products through 2032, but it becomes non-exclusive. Microsoft no longer needs to pay OpenAI revenue-share, but OpenAI will continue to pay Microsoft revenue-share through 2030 (with a cap in place).

This adjustment effectively unties OpenAI from a “single dependence on Microsoft” structure and sets the stage for deeper cooperation with other cloud providers. AWS is the first competing cloud to benefit, and the Bedrock platform also becomes the concrete entry point for OpenAI model multi-cloud distribution.

AWS integration: OpenAI models + Codex coding agents added to Bedrock

AWS customers can now access OpenAI models and Codex coding agents (agentic coding tools) directly through Amazon Bedrock. Bedrock is AWS’s AI model aggregation platform, which already integrates multiple foundation models such as Anthropic Claude and Meta Llama; with OpenAI added, Bedrock fully covers major leading model lines online.

Another key to this collaboration is the $5 billion investment agreement reached in February between OpenAI and Amazon—OpenAI will use AWS’s in-house Trainium chips with 2 GW of computing power to train AI models, and both sides also agreed to jointly develop customized models for Amazon engineering teams to be used in Amazon’s consumer-facing products. OpenAI’s spending commitment to AWS increases by $10 billion.

What to watch next: progress toward multi-cloud, and shifts in Microsoft business ties

The next point to watch is the specific breakdown of OpenAI’s multi-cloud deployment—how much share AWS and Microsoft Azure each take, and whether it further adds Google Cloud or Oracle. Another point to watch is the long-term positioning of OpenAI deployments on Microsoft Azure under the new agreement—if Microsoft’s revenue-sharing drops significantly, Azure’s priority for OpenAI services may change.

This article “Amazon and OpenAI expand cooperation: models added to Bedrock, Microsoft exclusivity ends” first appeared on ChainNews ABMedia.