Microsoft and OpenAI have amended their partnership to reduce their dependence on each other amid an escalating global artificial intelligence race, the companies said on Monday.
Microsoft, OpenAI’s biggest financial backer, holds a stake in the start-up valued at more than $135 billion. The tech giant has long held exclusive rights to offer OpenAI’s technology to other customers on its cloud computing service, a major draw for its business.
But the companies’ relationship became strained as OpenAI grew from a small nonprofit research lab to a start-up expecting a huge initial public offering as soon as this year. OpenAI has needed more computing power than Microsoft could readily or wanted to provide, and it has sought to build partnerships with other cloud computing providers as it races against the rival A.I. start-up Anthropic and others.
Under the new agreement announced on Monday, Microsoft maintains access to license OpenAI’s A.I. technology through 2032, but it will no longer have exclusive rights to those licenses. That will allow OpenAI to partner with other companies more freely and seek business from more sources.
OpenAI will lose some of its financial security because it will no longer receive a share of revenue from Microsoft when the tech giant uses the start-up’s technologies. It agreed to release its technology on Microsoft’s cloud services first, the companies said, unless “Microsoft cannot and chooses not to support the necessary capabilities.” The start-up then has the right to release products on other cloud services.
The ability to partner with other tech companies is potentially valuable for OpenAI. A flurry of deal making in recent months has drawn ties between A.I. labs and the biggest tech companies, each looking to expand where they can reach customers and get the computing power necessary to develop and run A.I. systems. Just last week, Anthropic secured up to $65 billion in additional investment, plus enormous computing deals, with Amazon and Google.
OpenAI will pay a share of revenue to Microsoft, up to a cap, when the tech giant serves OpenAI’s technologies to businesses and consumers via Microsoft’s cloud computing services.
The new agreement creates more certainty for Microsoft, giving it that share of revenue through 2030 regardless of how A.I. advances. It removes complex language that could have ended those payments if OpenAI were to declare that it had reached artificial general intelligence, or A.G.I. That level of technology would mean, in theory, that the technology is as capable as a human brain.
Though the two companies have altered their arrangement before, the new terms seek to resolve the outstanding tensions that hung over the relationship. Microsoft has also faced investor concern that it was too dependent on OpenAI, which is still young and has had unstable governance.
On Monday, jury selection is set to begin in a blockbuster trial that cuts to the heart of the relationship between OpenAI and Microsoft. Elon Musk, an early funder of OpenAI when it was a nonprofit, sued the start-up and Microsoft in federal court, saying they breached the lab’s founding agreement by putting commercial interests over the public good. Mr. Musk is asking for more than $150 billion in damages from OpenAI and Microsoft.
Microsoft is also scheduled to report its quarterly earnings on Wednesday.
(The New York Times sued OpenAI and Microsoft in 2023 for copyright infringement of news content related to A.I. systems. The two companies have denied those claims.)
Mr. Musk founded OpenAI as a nonprofit in 2015 alongside Sam Altman and several A.I. researchers. But Mr. Musk parted ways with the company less than three years later after a power struggle. Mr. Altman then attached a for-profit company onto the original nonprofit so that it could raise the vast amounts of money needed to build its technologies.
In 2019, Mr. Altman inked a partnership with Microsoft, which agreed to invest an initial $1 billion in the start-up, to provide the computing power needed to build A.I. technologies and license them for its own use. Microsoft eventually invested another $12 billion in OpenAI.
But the partnership began to fray after OpenAI’s board of directors temporarily ousted Mr. Altman in late 2023, about a year after ChatGPT was released. Microsoft allowed OpenAI to secure computing power from other cloud providers, including Oracle and the start-up CoreWeave. OpenAI also began to raise funds from other sources both in the United States and abroad.
At the end of last year, the two companies amended their partnership so that OpenAI could restructure itself as a more traditional for-profit company. This would allow OpenAI to go public on Wall Street, which it hopes to do in the coming months.
OpenAI has since signed a major cloud computing agreement with Amazon, one of Microsoft’s primary competitors. Under its new partnership with Microsoft, OpenAI has more freedom to pursue similar deals.
