OpenAI and Microsoft amend partnership to allow OpenAI products to run on non‑Azure clouds

Cloud infrastructure is the real story here. For years, the practical deal was simple: if you wanted OpenAI’s newest products at scale, Microsoft Azure sat in the middle. That changed on April 27, when OpenAI and Microsoft rewrote the commercial terms of their partnership. OpenAI can now serve all of its products across any cloud provider, even though Microsoft still stays in the picture as the primary cloud partner and first launch venue for OpenAI products on Azure. ### What actually changed? The big change is that Azure is no longer the only route. The amended agreement says OpenAI products will ship first on Azure, but only as long as Microsoft can and wants to provide the needed capabilities. If Microsoft cannot — or chooses not to — OpenAI can go elsewhere. More broadly, OpenAI the old setup. ### Did Microsoft really give up exclusivity? Basically, yes — but not the relationship. Microsoft still has a license to OpenAI intellectual property for models and products through 2032. The catch is that the license is now non-exclusive. Microsoft also remains a major shareholder and OpenAI’s primary cloud partner, so this is not a breakup. It is a shift from privileged gatekeeper to preferred partner. ### What changed on the money side? One payment stream stopped, and another kept going. Microsoft will no longer pay a revenue share to OpenAI under the amended deal. But OpenAI will keep making revenue-share payments to Microsoft through 2030, at the same percentage as before, now with a total cap. That matters because ### Why does this matter beyond corporate paperwork? Because cloud distribution is power. If a model company is tied to one cloud, that cloud gets the traffic, the enterprise contracts, the telemetry, and a lot of the surrounding tooling business. Once OpenAI can run across clouds, AWS and Google Cloud can compete not just where enterprises buy AI. ### Did the market move that fast? Turns out, yes. Amazon moved almost immediately. CNBC reported on April 28 — one day after the amended deal — that OpenAI models were becoming available on Amazon’s cloud. That speed matters because it suggests other cloud providers were waiting on contract terms, not on years of technical integration. Once the gate opened, product listings followed. ### What does this mean for customers? Multi-cloud sounds liberating, but it adds operational mess. A company using OpenAI services through Azure, AWS, and maybe Google Cloud now has to think about identity, logging, compliance boundaries, data residency, network egress, and cost control across several platforms. In plain terms and more complicated. ### Why keep Azure first, then? Because Microsoft still brings massive datacenter capacity, enterprise distribution, and a deep product stack around OpenAI services. Azure-first preserves a commercial edge without absolute exclusivity. It is a softer advantage — more “preferred launch partner” than “sole authorized home.” That gives OpenAI room to grow while letting Microsoft keep a front-row seat. ### Bottom line This is less a divorce than a rewiring. OpenAI just turned its cloud strategy from single-home to multi-home, and Microsoft accepted a smaller kind of advantage in exchange for longer-term clarity. For customers, the headline is simple — OpenAI is no longer effectively Azure-only. The harder part starts now: figuring out which cloud, for which workload,