OpenAI products can now be hosted on Amazon Web Services, Google Cloud, and more. Compiled by | Yang Jingli Edited by | Yun Peng
Zhidx reported on April 28 that early this morning, Microsoft and OpenAI jointly announced a revision to their long-standing partnership agreement, ending Microsoft's exclusive distribution rights for OpenAI models. From now on, OpenAI can offer its full range of products to customers of all cloud service providers. Microsoft will no longer pay revenue share to OpenAI; instead, OpenAI's revenue share to Microsoft will continue until 2030, subject to a total cap.
Since 2019, Microsoft has cumulatively invested over $13 billion (approximately RMB 88.79 billion) in OpenAI, serving as a crucial provider of funding and computing power. In October 2025, following OpenAI's transition to a for-profit corporate structure, Microsoft acquired approximately 27% equity at a valuation of $135 billion (approximately RMB 922.05 billion), and OpenAI committed to purchasing $250 billion (approximately RMB 1.71 trillion) worth of Azure cloud services. Following the release of the revision announcement, OpenAI CEO Sam Altman posted an update stating, "We have updated our partnership with Microsoft."
Discussions on the X platform quickly diverged into various voices. Some believe this is a win-win deal, as Microsoft remains the primary partner while OpenAI gains more flexibility.
Others criticized Altman for burning bridges, commenting, "Microsoft built the infrastructure for you, and now you're just walking away. Nice job, Sam."
Additionally, some netizens expressed regret, noting that "primary cloud partner" sounds exactly like a breakup line: "We still care about each other, but we're going to date other servers now."
This revision marks a major adjustment in the relationship between the two parties, with key changes focusing on distribution exclusivity, IP licensing duration, and revenue-sharing structure.
01.End of Exclusive Distribution: OpenAI Products Available on Any Cloud, Azure Retains First-Access Priority
According to the joint statement, Microsoft remains OpenAI's primary cloud partner, and new OpenAI products will debut on Azure first, unless Microsoft is unable or chooses not to support the required capabilities. OpenAI can now offer its full range of products to customers of any cloud service provider, no longer bound by Microsoft's exclusive distribution terms. Bloomberg noted that this change paves the way for cooperation between OpenAI and cloud providers like Amazon Web Services (AWS). Amazon CEO Andy Jassy posted that AWS will offer OpenAI models to customers via the Bedrock service in the coming weeks, with more details to be announced at an event in San Francisco the following day. He stated, "This gives developers more choices to pick the right model for the right task."
Reportedly, OpenAI reached a strategic partnership with Amazon in February 2026, with Amazon committing to invest up to $50 billion (approximately RMB 341.5 billion). The two parties will expand their existing $38 billion (approximately RMB 259.54 billion) AWS agreement by $100 billion (approximately RMB 683 billion) over the next eight years; AWS will also serve as the exclusive third-party cloud distributor for OpenAI's enterprise platform Frontier (released in early April). Denise Dresser, Head of Revenue at OpenAI, admitted in an internal memo earlier this month that the exclusive relationship with Microsoft "limited our ability to meet enterprise needs."
02.IP Licensing Extended to 2032, Revenue Share Extended to 2030 but Capped, AGI Trigger Clause Removed
The statement shows that Microsoft's IP licensing for OpenAI models and products is extended to 2032, but this license is no longer exclusive. OpenAI's revenue share to Microsoft will continue until 2030, with the percentage remaining unchanged but subject to an undisclosed total cap. CNBC, citing an anonymous source, reported that the revenue share percentage is 20%, and Microsoft will continue to receive a share from every ChatGPT subscription. Bloomberg noted that the previous agreement stipulated that revenue sharing would cease once OpenAI was deemed to have achieved Artificial General Intelligence (AGI); the new agreement removes this trigger clause, ensuring Microsoft receives steady revenue shares until 2030 regardless of whether OpenAI achieves AGI. Meanwhile, Microsoft will no longer pay revenue share to OpenAI for OpenAI products it resells on the cloud.
Under the old mechanism, when users paid to access OpenAI models via Azure, Microsoft would pay a certain percentage of revenue share to OpenAI; this cash flow is now cut off. Microsoft still directly participates in OpenAI's growth with its approximately 27% equity stake. The statement emphasized that although this revision simplifies the partnership, the ambition to jointly advance their mission remains unchanged. From expanding gigawatt-level data center capacity to co-developing next-generation chips, and applying AI to advance cybersecurity and other fields, the two parties will continue to work together to advance and expand AI development for individuals and organizations worldwide.
Following the announcement, Microsoft's stock price fell 1.78% at the open before recovering above the opening price, while Amazon dropped 1.09% today. Analysts at Evercore ISI stated in a client report, "We do not believe the revised agreement is a surprise to investors at this point in time. Microsoft has increasingly signaled its interest in a multi-model strategy, and OpenAI has ample incentive to expand its distribution footprint." Notably, this week Microsoft and OpenAI will also face a lawsuit from Elon Musk. Musk accuses OpenAI of straying from its founding mission by transforming into a for-profit company backed by billions of dollars from Microsoft, and is seeking $134 billion (approximately RMB 915.22 billion) in damages.
03.Conclusion: Untying the Knot, Reshaping the Distribution Landscape for AI Giants
This revision adjusts the relationship between OpenAI and Microsoft from deep binding to primary partnership: Azure retains only first-access priority and no longer blocks the rollout of OpenAI models on Amazon Web Services and Google Cloud. For Microsoft, the cost is losing exclusive distribution rights and giving up the reverse revenue share from its own cloud resales of OpenAI products. In return, it secures the extension of IP licensing to 2032, the continuation of revenue sharing until 2030, and the removal of the AGI trigger clause.
Viewed in the broader context of computing power allocation, OpenAI requires gigawatt-level computing power from more than one cloud provider to support agent-based scenarios; Meta recently also secured $48 billion (approximately RMB 327.84 billion) in committed computing power from CoreWeave and Nebius. The diversification of computing power sources among top AI companies is becoming a trend. To a certain extent, the revision of the agreement is a form of "untying" for both parties.