Microsoft caps OpenAI revenue share at $38B and removes Azure exclusivity in deal rewrite
- OpenAI and Microsoft rewrote their partnership on April 27, letting OpenAI sell across any cloud while keeping Azure as its primary launch partner. - The key money change is one-way: Microsoft stopped paying OpenAI revenue share, while OpenAI’s 20% payments now run to 2030 under a cap. - The cap was undisclosed publicly, but The Information said it is $38 billion — far below earlier internal projections.
The Microsoft-OpenAI deal just got a lot less exclusive. That matters because this was the contract that helped define the AI boom — who got OpenAI’s models first, where they ran, and how the money flowed back and forth. The old version tied OpenAI tightly to Azure and gave Microsoft unusually broad rights. The new version loosens all of that without actually ending the partnership. ### What changed in the deal? On April 27, Microsoft said OpenAI can now serve all of its products to customers across any cloud provider, while Azure stays the “primary” cloud partner and still gets first shipment of OpenAI products unless Microsoft can’t or won’t support the needed capabilities. Microsoft also said its license to OpenAI IP now runs through 2032 on a non-exclusive basis, and Microsoft will no longer pay a revenue share to OpenAI. (blogs.microsoft.com) ### So Azure lost exclusivity? Basically, yes. Azure is still first in line, but it is no longer the only place OpenAI can sell or run services. That is the real structural shift. Enterprises that already live on AWS or Google Cloud no longer have to treat Azure as the mandatory doorway for every OpenAI deployment, at least under the amended commercial framework. (blogs.microsoft.com) ### What happened to the revenue share? The money flow got simplified and tilted toward Microsoft less aggressively than before. Microsoft’s public post says OpenAI will keep paying Microsoft a revenue share through 2030 at the same percentage, but now with a total cap. CNBC added one crucial detail — the rate remains 20%, which means Microsoft still takes a cut of things like ChatGPT subscriptions, but Microsoft no longer pays a reciprocal share to OpenAI. (blogs.microsoft.com) ### Where does the $38 billion number come from? Microsoft did not publish the cap number. But Reuters, citing The Information on May 11, said the total revenue-sharing payments are capped at $38 billion. The same report said this replaces a much larger internal scenario in which OpenAI could have paid Microsoft as much as $135 billion through 2030. That is why this looks less like a tweak and more like a major reset. (blogs.microsoft.com) ### Why would OpenAI want this? Because the old setup was becoming a bottleneck. CNBC reported that OpenAI’s revenue chief told staff the Microsoft arrangement had limited OpenAI’s ability to meet enterprises where they are. If a big customer wants OpenAI services inside AWS or Google Cloud for procurement, compliance, or latency reasons, OpenAI now has a cleaner path to say yes. (money.usnews.com) ### Why would Microsoft agree? Because Microsoft still keeps the parts that matter most. It remains OpenAI’s primary cloud partner, keeps a non-exclusive IP license through 2032, and still collects revenue share through 2030. Microsoft has also invested more than $13 billion in OpenAI since 2019, so it still benefits directly as a major shareholder even if the commercial plumbing is less one-sided than before. (cnbc.com) ### What else quietly changed? The AGI trigger lost its bite. Microsoft said OpenAI’s payments continue through 2030 independent of OpenAI’s technology progress, which means the deal no longer hinges the same way on whether OpenAI declares it has reached artificial general intelligence. That removes one of the strangest pressure points in the old relationship. (blogs.microsoft.com) ### Bottom line This is not a breakup. It is a rewrite from exclusivity to managed flexibility. OpenAI gets room to go multi-cloud and keep more upside. Microsoft gives up monopoly-style control, but keeps early access, long-dated rights, and a capped stream of payments. In AI infrastructure terms, that is a big power rebalance. (blogs.microsoft.com)