Artificial Intelligence & Machine Learning
,
Next-Generation Technologies & Secure Development
Renegotiated Pact With Microsoft Clears OpenAI Path to Enterprise Clouds
Rashmi Ramesh (rashmiramesh_) •
April 29, 2026
Image: Shutterstock
OpenAI customers don’t also have to be Azure customers following the artificial intelligence company’s renegotiation of a pact that required Microsoft exclusivity.
See Also: The Context Crisis: Cloud Security in the Age of AI
The AI firm on Tuesday launched on Amazon Web Services paid access to its models, Codex coding tool and a new agent deployment service.
Microsoft, an early investor in OpenAI that pumped $13 billion into the company, previously was the sole distributor of OpenAI products through a contractual clause that guaranteed exclusivity until AI achieves consciousness on par with a human. The rider created a conflict in February when OpenAI signed a $50 billion deal with Amazon that committed AWS to exclusive distribution rights for Frontier, OpenAI’s enterprise agent platform. On the same day, Microsoft reportedly considered suing OpenAI, saying that Azure was the exclusive cloud provider of stateless OpenAI APIs.
The two companies appear to have resolved the dispute on Monday. Microsoft’s license to OpenAI’s intellectual property is now non-exclusive and runs through 2032, meaning that the AI firm can serve all of its products across any cloud provider. Microsoft no longer pays a revenue share to OpenAI, while OpenAI continues paying 20% to Microsoft through 2030, an amount now subject to a cap. Microsoft retains its designation as OpenAI’s “primary cloud partner,” meaning OpenAI products are expected to ship first on Azure unless Microsoft cannot or chooses not to support them.
In an internal memo reported by CNBC, OpenAI revenue chief Denise Dresser told staff that the Microsoft partnership had “been foundational to our success” but had “also limited our ability to meet enterprises where they are.” Demand for OpenAI on AWS was “frankly staggering,” she wrote.
The market OpenAI is entering on AWS is one where Anthropic’s Claude models have been available for years and have built a significant lead. A Menlo Ventures report shows that Anthropic holds 32% of enterprise AI model usage while OpenAI holds 25% – a reversal from 2023, when OpenAI commanded half of the enterprise market and Anthropic held just 12%.
GPT-5.4 is available on Bedrock now, with GPT-5.5 arriving within the next couple of weeks, WS CEO Matt Garman reportedly said at a San Francisco event. Codex, which OpenAI says has more than four million weekly users, can be configured to run through Bedrock and count against existing AWS cloud commitments. Bedrock Managed Agents handles the underlying infrastructure for AI agents, allowing enterprises to deploy automated workflows without assembling the supporting architecture themselves.
To claim all $35 billion from Amazon, OpenAI will need to spin up two gigawatts of Amazon’s Trainium accelerators – Amazon’s custom chips, which compete with Nvidia’s hardware for AI workloads.
The launch is part of a broader pattern of cloud providers making large, parallel commitments to competing AI labs. Google this week announced plans to invest up to $40 billion in Anthropic, with a key component being five gigawatts of Google Cloud compute capacity over five years. Amazon has made a separate commitment of up to $25 billion to Anthropic. Microsoft, despite losing exclusivity, retains roughly 27% ownership of OpenAI’s for-profit entity and continues to benefit financially from its growth regardless of which cloud serves it.
Analyst Ben Thompson said Azure’s exclusivity had been constraining OpenAI’s enterprise reach. Enterprises wanting OpenAI models had to use Azure even when their workloads were elsewhere, a friction that worked to Anthropic’s advantage. With that constraint removed, OpenAI can now compete on the infrastructure where a significant share of enterprise AI adoption has already taken place.