OpenAI and Microsoft on Monday unveiled the most significant rewrite of their seven-year partnership, and tucked inside the legal scaffolding is the quiet death of one of the strangest provisions in modern tech contracting: the so-called AGI clause. The amended agreement, announced jointly by both companies, replaces the artificial-general-intelligence trigger with a flat, non-exclusive license that runs through 2032 regardless of how far the technology advances.
The AGI clause had hung over the relationship since the original 2019 investment. It stipulated that once OpenAI's board declared the company had achieved AGI, Microsoft's commercial IP rights to OpenAI's models would be voided. An October 2025 amendment had already softened the mechanism, routing any AGI declaration through an independent expert panel rather than leaving it solely to OpenAI's directors. Monday's rewrite eliminates the trigger altogether.
A predictable runway through 2032
Under the new terms, Microsoft receives a non-exclusive license to OpenAI's models and products through 2032, with no AGI-based off-ramp. The financial architecture is similarly stripped of contingency: Microsoft will no longer pay a revenue share to OpenAI, while OpenAI's revenue share payments to Microsoft continue through 2030 at the same percentage, subject to a total cap. The Microsoft announcement specifies these payments will continue "independent of OpenAI's technology progress" — language that effectively forecloses the AGI-as-escape-hatch reading of the original deal.
Azure remains OpenAI's primary cloud partner, and new OpenAI products will ship first on Azure when Microsoft can and chooses to support the required capabilities. Outside of that carve-out, OpenAI is now free to serve customers across any cloud provider, opening the door to deployments on AWS, Google Cloud and Oracle.
Why removing the trigger matters
The AGI clause had become an awkward fixture of every subsequent OpenAI fundraise and Microsoft earnings call. Investors had to model a partnership whose financial spine could, in theory, be severed by a board vote keyed to a term with no agreed technical definition. By converting the relationship into a fixed-term license with capped payments, both companies trade contingency for certainty.
For OpenAI, the practical win is the freedom to court rival cloud providers and pursue capacity-driven deals without renegotiating the AGI mechanism each time. Wedbush analysts said the clean structure also removes friction tied to a possible OpenAI public offering by simplifying how the company's revenues, IP and cloud commitments are encumbered by the Microsoft relationship.
For Microsoft, the trade-off is losing exclusivity in exchange for a predictable, time-bounded license that no longer depends on whether or when AGI is declared. Microsoft retains its shareholder stake and continues to collaborate on data-center expansion, next-generation silicon and cybersecurity work that both companies flagged as ongoing priorities.
What the death of the clause says about the AI moment
The AGI clause was always less a technical milestone than a piece of governance theater — a way of bolting a hard limit onto a commercial relationship neither side could fully scope in 2019. Removing it is an admission that the industry has stopped trying to legislate frontier-model thresholds inside private contracts and is instead pricing the relationship like any other strategic license.
Whether or when AGI arrives now matters less to the legal architecture binding the two largest commercial players in AI. That is itself a quiet but telling shift in how the industry has come to talk about the term.
