TheChek News Flow

New agreement is set to remove one of several barriers for the startup, which is seeking to restructure and become a for-profit company

Microsoft MSFT 0.13%increase; green up pointing triangle and OpenAI said they have reached a deal to extend their partnership, an apparent detente that could ease the startup’s path toward changing its structure to include a for-profit corporation.

The agreement, which has yet to be completed, comes after tensions boiled over between the companies. Microsoft was one of OpenAI’s earliest partners and is among its largest investors, integrating the company’s technology into many of its products. But it also has a stranglehold over the startup, given that Microsoft is OpenAI’s exclusive cloud provider and has access to its latest technology.

The companies didn’t disclose the terms of the new contract and said the current agreement was nonbinding. It caps a summer of difficult negotiations between the two sides, who helped fuel one another’s rise during the early years of the AI boom but have since grown into competitors.

The relationship at one point grew so fractious that OpenAI considered going to antitrust regulators to break out of the contract, The Wall Street Journal reported.

OpenAI is also putting final touches on the ownership structure behind the new for-profit company it is trying to create. The company said it plans to keep the nonprofit’s control over the new for-profit and endow it with a stake valued at more than $100 billion. That would make the nonprofit one of the largest philanthropies in the world on paper, although it is unclear how long it would take for funds to be available for distribution.

The announcement of a large endowment for the nonprofit was disclosed just days after the Journal reported that OpenAI’s executives were rattled by mounting pushback to the restructuring. The attorneys general in California and Delaware are investigating the startup to make sure the proposed plan doesn’t violate charitable law.

Microsoft and the OpenAI nonprofit are each set to initially receive a roughly 30% stake in the proposed new company, according to people familiar with the matter. The rest will go to employees and investors, the people said.

Microsoft shares rose by nearly 2% in after-hours trading.

OpenAI is facing an array of opponents that are actively seeking to block its restructuring effort or intervening with regulators to ensure its nonprofit mission remains intact. They include billionaire Elon Musk, Meta Platforms and an array of advocacy organizations. Musk has sued OpenAI.

The agreement is a tentative win for OpenAI, which needed Microsoft to approve its for-profit plan before formally sharing it with state regulators. OpenAI told its venture-capital investors it would complete the restructure by the end of the year, or risk losing out on $19 billion in funding.

The startup is currently run as a subsidiary that issues profit-sharing units instead of standard stock—a structure that is unpopular among investors and makes it virtually impossible to be listed on a public exchange.

OpenAI initially wanted the new for-profit company to be separate from the nonprofit, but it pulled this plan in May after drawing criticism that it was illegally diverting charitable assets to private investors.

Because the new terms of the Microsoft-OpenAI partnership weren’t disclosed, it was difficult to parse who prevailed in negotiations on a number of disputed issues.

During the talks, OpenAI wanted Microsoft to allow the startup to sell its AI products through other cloud providers. Microsoft was fighting hard to keep its exclusive access to OpenAI’s intellectual property, even if the startup were to declare that its systems had reached a high level of capability known as “artificial general intelligence.”

Microsoft proposed replacing AGI with an even higher threshold, dubbed “artificial superintelligence,” a person familiar with the matter said.

Microsoft was once key to OpenAI’s success, building giant data centers to train the startup’s AI models and showering it with billions of dollars in funding, long before the release of ChatGPT. In exchange, it got early access to OpenAI’s technology, which it integrated into its coding and office-productivity products.

But the tech giant was caught off guard by ChatGPT’s overnight success and began launching competing products. It also grew nervous about relying too much on OpenAI for its AI strategy after the startup’s chief executive, Sam Altman, was briefly fired in November 2023.

Afterward, Microsoft CEO Satya Nadella launched an in-house competitor to OpenAI, picking one of Altman’s rivals to run the lab, and backed away from a giant data-center project called Stargate that Altman wanted. Over time, OpenAI executives grew more distrustful of Microsoft, and were more hesitant to share the startup’s latest technology with the company, leading the two sides to drift apart.

Microsoft recently disclosed that it has started to publicly test a homegrown AI model that could be used for its Copilot chatbot, which has historically relied on OpenAI’s models.

The tech giant has looked to other partners outside of OpenAI to supply its AI needs. The company recently struck a deal with OpenAI rival Anthropic to use the startup’s models to power AI tools for its 365 products, people familiar with the matter said. Microsoft, which will access the models through Amazon Web Services, already uses Anthropic for AI tools inside its GitHub coding platform. Microsoft Windows also uses non-OpenAI models. Microsoft’s cloud-computing unit also began hosting models from Elon Musk’s xAI earlier this year.

Microsoft still plans to use OpenAI as its primary option for AI models, but the company has found that Anthropic performs better for certain functions in 365, one of the people said. The Information earlier reported on the Anthropic deal.

News Corp, owner of The Wall Street Journal and Dow Jones Newswires, has a content-licensing partnership with OpenAI.