OpenAI and Microsoft reach tentative deal to reshape corporate structure; nonprofit to receive $100 billion stake
Agreement is nonbinding and draws scrutiny from state attorneys general and others concerned about governance, safety and rights tied to future AGI

OpenAI said Thursday it has reached a tentative, nonbinding agreement with Microsoft that would give OpenAI’s nonprofit parent a $100 billion equity stake in the for-profit company that develops and sells its AI products. The announcement outlined a new phase of the longtime partnership between the two companies but left key contractual terms unresolved.
The companies said they are still "actively working to finalize contractual terms in a definitive agreement," and both declined further comment. Regulators, competitors and advocates have pressed OpenAI over the past year about the proposed restructuring, and state attorneys general in California and Delaware have signaled inquiries into whether the changes could undermine the nonprofit’s stated safety mission.
OpenAI was founded as a nonprofit in 2015 and has maintained a governance structure in which a nonprofit board oversees a for-profit subsidiary. The company says the new arrangement would transfer a large equity stake to that nonprofit. It is not clear from the companies’ statements whether the $100 billion equity stake would constitute a controlling interest in the for-profit business.
California Attorney General Rob Bonta announced last week that his office is investigating OpenAI’s proposed restructuring of its finances and governance. Bonta and Delaware Attorney General Kathy Jennings wrote to OpenAI expressing concerns about the safety of ChatGPT following a meeting with the company’s legal team. "Together, we are particularly concerned with ensuring that the stated safety mission of OpenAI as a non-profit remains front and center," Bonta said in a statement.
Microsoft first invested $1 billion in OpenAI in 2019 and later became the exclusive provider of the large-scale cloud computing capacity OpenAI used to train its models. Microsoft in turn integrated OpenAI’s technology into its own suite of products. Earlier this year, on Jan. 21, the two companies announced changes allowing OpenAI to build its own computing capacity "primarily for research and training of models," a move that coincided with OpenAI’s separate deal with Oracle to construct a large data center in Abilene, Texas.
Those shifts illustrated a loosening of the tighter interdependence that had characterized the partnership, though some elements of the relationship remained unsettled. The newly disclosed $100 billion equity arrangement is described as part of the next phase of the partnership, but the statement from the companies provided only partial details and called the agreement nonbinding.
OpenAI’s governance has been a focal point of debate because its nonprofit board was given the authority in earlier agreements to decide when AI systems have reached the level of "artificial general intelligence," or AGI — broadly defined as technology that can perform most economically valuable tasks as well as or better than humans. Under prior terms between OpenAI and Microsoft, certain commercialization rights for Microsoft would not extend to systems that cross that AGI threshold, with the earlier contractual language applying only to pre-AGI technology.
The proposed restructuring has also prompted litigation. Elon Musk, who was an early backer and co-founder of OpenAI, filed a lawsuit seeking to prevent the company from taking control away from the nonprofit board, alleging the organization betrayed a promise to develop AI for the benefit of humanity. The lawsuit argues the changes undermine the nonprofit’s governance role, allegations OpenAI disputes.
Industry and policy observers have expressed concerns that altering the balance of control and financial incentives could affect how safety and public-interest considerations are prioritized as the company develops more advanced systems. Advocates and some competitors have asked regulators to scrutinize whether the structure could create conflicts between profit motives and the safety mission that OpenAI’s nonprofit founders articulated.
OpenAI’s board has added high-profile figures since the company’s founding, including a former U.S. Treasury secretary, and the governance makeup has been central to discussions about accountability, transparency and oversight. State regulators have authority to review nonprofit reorganizations in ways that could affect whether the proposed equity transfer proceeds as described.
For now, the announcement leaves substantial questions about the final terms and implications of the deal. The companies said they were continuing to work through contractual details and did not provide a timetable for finalizing a definitive agreement. Meanwhile, state investigations and ongoing litigation are likely to shape how — and whether — the proposed restructuring moves forward.