In brief
Microsoft valued its OpenAI stake at $135 billion and extended exclusive rights to frontier models through 2032.
OpenAI can now work with outside developers and release open-weight models, while Microsoft can pursue AGI independently.
The update comes as Microsoft faces an antitrust suit alleging it used Azure’s dominance to inflate ChatGPT prices.
On Tuesday, Microsoft and OpenAI announced a restructured partnership that values the Redmond giant’s stake at $135 billion, roughly 27% of OpenAI’s new public-benefit company, even as both firms navigate antitrust scrutiny and a federal lawsuit alleging compute monopolization.
The reworked pact supports OpenAI’s conversion into OpenAI Group PBC under the nonprofit OpenAI Foundation, and positions Microsoft as the company’s “frontier model partner” through 2032, according to a Tuesday statement.
Board chair Bret Taylor and CEO Sam Altman can now control appointment and removal powers over the PBC’s board, consolidating Altman’s authority.
OpenAI will continue channeling roughly 20% of revenue to Microsoft, though both parties expect that flow to end once an independent panel certifies that artificial general intelligence has been achieved.
Microsoft retains exclusive IP licenses to OpenAI’s models and products through 2032, including post-AGI systems, but holds no rights to any consumer hardware OpenAI produces, according to the statement.
The AI giant can collaborate with third-party developers on joint products, deploy open-weight models that meet safety thresholds, serve U.S. national security agencies on any cloud infrastructure, and independently pursue its own AGI research capabilities, previously blocked by Microsoft’s exclusivity provisions.
API products developed with third parties will be exclusive to Azure, while non-API products may be served on any cloud provider.
Microsoft and OpenAI did not immediately respond to Decrypt’s request for comment.
The announcement arrives amid mounting legal pressure as a class-action suit filed two weeks earlier alleges Microsoft weaponized its 2019 Azure exclusivity arrangement to throttle computational capacity for ChatGPT, artificially maintaining subscription rates at “100 to 200 times” competitors’ levels during February’s AI pricing conflict.
“The AI we build today will shape our tomorrow. The path we are currently on, dominated by centralized AI, is fraught with peril,” Jiahao Sun, CEO of FLock.io, told Decrypt. “When a few powerful entities control AI, we risk creating systems that reflect a narrow worldview, perpetuating biases, and eroding trust.
OpenAI has also committed to purchasing $250 billion in additional Azure services, though Microsoft surrendered first-refusal rights as compute provider.
The requirement that OpenAI source all computational resources exclusively from Microsoft had become a major friction point as ChatGPT’s 800 million weekly users and research demands drove up infrastructure costs.
The partnership began in July 2019 with a $1 billion investment that made Microsoft OpenAI’s exclusive cloud provider, and deepened in January 2023 with a multi-billion-dollar expansion that positioned Microsoft as OpenAI’s primary backer.
The latest announcement also lands as OpenAI disclosed concerning mental-health signals among its user base, about 1.2 million weekly users, or roughly 0.15% of actives, showed explicit indicators of suicidal planning or intent.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
