Summary: The Financial Times reported on March 18, 2026 that Microsoft was weighing legal action over Amazon’s $50 billion OpenAI deal, under which AWS becomes the sole third-party cloud for OpenAI Frontier models. Microsoft argues the arrangement crosses Azure exclusivity clauses; OpenAI says no breach occurred. The article notes Microsoft’s leverage but also antitrust scrutiny, OpenAI IPO plans, and talks toward a negotiated outcome.
The news landed on March 18, 2026: the Financial Times reported that Microsoft is weighing whether to take OpenAI and Amazon to court, and Reuters carried the story the same day. At issue is the partnership announced on February 27, in which Amazon commits $50 billion and AWS becomes the sole third-party cloud for OpenAI’s Frontier models. Microsoft’s position: the arrangement crosses the Azure exclusivity clauses written into its agreement with OpenAI.
This is not your standard commercial quarrel. The FT’s numbers show why: Microsoft’s stake in OpenAI stands at roughly 27%, valued near $135 billion, and on top of that OpenAI has pledged to purchase $250 billion worth of Azure cloud services. Investor, largest supplier of compute, and the exclusive distributor that just got routed around: three identities on one balance sheet, and that stack is exactly what makes this prospective lawsuit the clearest window into how the AI alliances are being redrawn.
Inside the $50 Billion Package
Three components made up the February 27 Amazon–OpenAI package: $50 billion of investment, OpenAI’s Frontier models landing on AWS, and Alexa+ bringing OpenAI models into its service. The dollar figure is not what irritates Microsoft. It is the phrase “exclusive third-party cloud”: by the FT’s account, beyond Microsoft’s own perimeter, AWS will be the only cloud serving OpenAI’s Frontier product. Microsoft and OpenAI issued a joint statement that same day to calm things down, restating that the existing agreement stands — and three weeks later, that very statement is the text each side now reads to its own advantage.
How the Contract Reads
The final agreement Microsoft and OpenAI signed at the close of 2025 sets out the split: API products developed jointly with third parties sit exclusively on Azure, while non-API products can be served by any cloud provider. The joint statement of February 27 went further: Azure continues as the exclusive cloud provider for OpenAI’s stateless APIs; stateless API calls that grow out of OpenAI’s third-party collaborations — Amazon among them — are hosted on Azure; and OpenAI’s own products, Frontier included, keep running on Azure. The statement also reiterated that Microsoft holds its exclusive license to the IP across OpenAI’s models, that the revenue-share arrangement extends to OpenAI’s deals with other cloud providers, and that the contractual definition of AGI remains untouched. Taken together, Microsoft expected exactly this kind of deal — it just counted on those deals flowing through Azure.
That leaves one question at the center of everything: in the Amazon deal, is the Frontier supply arrangement an “API product” or a “non-API product”? OpenAI’s view is that no breach occurred. Microsoft’s lawyers have evidently arrived somewhere else.
Microsoft’s Strong Hand — and Its Reasons to Hold Back
The strong hand first. Azure expanded 39% year over year in FY26 Q2, its backlog climbed to $625 billion, a gain of 110%, and about 45% of that figure traces back to OpenAI’s purchase commitments. The hesitations are equally concrete. Reporting across the board treats an actual lawsuit as highly unlikely: the two sides are searching for a negotiated way out; Microsoft’s Azure licensing practices already draw antitrust examination in the US, the UK, and the EU, and going to court would pull in still more regulatory attention; and OpenAI is readying an IPO at a valuation near $1 trillion, with a filing possibly arriving in the second half of 2026. As one source close to Microsoft put it, “the last thing OpenAI needs is another court case right now.”
To the FT, someone with knowledge of the matter was even blunter: “If Amazon and OpenAI want to take a bet on the creativity of their contractual lawyers, I would back us, not them.”
Developers and Enterprise Buyers: What Changes
One: the ways to access OpenAI models keep splitting apart. Azure OpenAI Service, direct purchase from OpenAI, and Frontier on AWS now sit alongside one another, each carrying its own terms, pricing structure, and compliance obligations — procurement has to line the contracts up side by side rather than assume OpenAI defaults to Azure. Two: when even Microsoft must defend an exclusive position through contract language, vendor lock-in has moved from an architecture question into litigation territory, and the case for designing swappable model interfaces and cloud-agnostic abstraction layers only strengthens. Three: the dispute broke in the same week as NVIDIA GTC, where enterprise AI infrastructure was the center of attention; Dell’s AI Factory updates there and this cloud confrontation frame each other. Who supplies the compute and where models are permitted to run now ranks among the headline entries on every enterprise AI bill.