The last lock comes off OpenAI

On April 27, 2026, Microsoft and OpenAI rewrote their partnership and ended the one piece of exclusivity that had outlasted all the headlines. The compute story — OpenAI renting raw capacity from Oracle, Amazon, and others — had already gone non-exclusive in stages through 2025. What survived was subtler and more valuable: Azure was the only cloud legally allowed to host and resell OpenAI's models, so any company that wanted to use GPT had to route through Microsoft's API. That is what came off.

Microsoft's rights no longer end the day OpenAI declares AGI — a sci-fi tripwire traded for a license that simply expires in 2032.

The day after the announcement, OpenAI's flagship models, including the just-shipped GPT-5.5, went live on Amazon's Bedrock service for the first time. The practical effect is that a business already running on Amazon or, soon, Google and Oracle can now consume OpenAI inside the cloud it already pays for, rather than being pulled onto Azure to do it. For four hyperscalers fighting over the same enterprise wallets, the most sought-after models just stopped being a Microsoft-only feature.

The same amendment quietly retired a famous clause: the provision that would have severed Microsoft's rights to OpenAI's technology the instant OpenAI declared it had reached artificial general intelligence. That science-fiction tripwire is gone, swapped for a plain license that runs to 2032 and a capped revenue share. Microsoft keeps a consolation prize — new OpenAI products still ship on Azure first — but the era where a single contract bound the most valuable models in AI to one cloud is over.