OpenAI said on Tuesday that it had adopted a new for-profit structure, a long-sought change that could allow the business to operate like a more traditional company while it raises the billions of dollars it needs to develop artificial intelligence.
The widely anticipated move was considered one of the keys to the future of OpenAI. It will allow the maker of the popular ChatGPT chatbot to raise new funds like any other company and potentially sets the stage for a blockbuster initial public offering on Wall Street.
The shift also firmly establishes OpenAI as one of the tech industry’s standard-bearers in the A.I. boom, allowing the San Francisco company to compete on more solid footing with giants like Google, Amazon and Meta.
For more than 18 months, Sam Altman, the high-profile chief executive of OpenAI, and his colleagues have been trying to rework OpenAI’s unorthodox structure, which blends a nonprofit with a for-profit company. The announcement on Tuesday offers some clarity about how the influential A.I. company, which was founded 10 years ago, will operate.
OpenAI said in a blog post that it had become a public benefit corporation, or P.B.C., which is a for-profit corporation designed to create public and social good. OpenAI rivals like Anthropic and xAI, which is owned by Elon Musk, use a similar structure.
OpenAI said it had completed the restructuring after negotiations with the attorneys general in Delaware, where, like many other companies, it was incorporated, and California, where it has its headquarters.
The attorneys general, who were being lobbied by a number of OpenAI’s critics, could have blocked the company’s plans with lawsuits. Rob Bonta, the attorney general of California, expressed particular concern in a recent interview with The New York Times about how ChatGPT interacted with young people.
“OpenAI is a registered charitable organization operating in California, and that means we have a duty to ensure they are fulfilling their mission, which includes benefiting all of humanity,” Mr. Bonta said.
Now that Mr. Bonta has signed off on the deal, the nonprofit that has controlled OpenAI since it was founded has a roughly $130 billion stake in the company. That makes the nonprofit, now called the OpenAI Foundation, the most valuable foundation in the country, easily topping the $86 billion held by the Gates Foundation, which was created by the billionaire Microsoft co-founder Bill Gates and his former wife, Melinda.
Mr. Altman does not have a significant stake in the new for-profit company, according to a person familiar with the new corporate structure who spoke on the condition of anonymity because the person was not authorized to speak about the arrangement. He did not have a major stake under the previous structure, either, other than a tiny slice of an old investment in the company from the start-up incubator Y Combinator, where he was once president.
OpenAI said the nonprofit would continue to control the for-profit company, which is called OpenAI Group PBC, through its board. Microsoft, OpenAI’s largest investor, will maintain a stake in the new company worth roughly $135 billion. The nonprofit holds a 26 percent stake in the new company, Microsoft owns a roughly 27 percent stake and the remaining 47 percent is held by current and former employees and other investors.
The nonprofit’s board of directors will appoint all members of the for-profit board and can replace them at any time. The members of the nonprofit board will also serve on the for-profit board, with the exception of Zico Kolter, an academic who specializes in A.I. safety. He will be a nonvoting observer on the for-profit board and will oversee the nonprofit’s Safety and Security Committee, which monitors the safety of the organization’s technologies.
Some OpenAI critics were unhappy that the company moved ahead with its restructuring.
“Today’s announcement from OpenAI is an attempt to entrench the status quo, in which OpenAI Nonprofit serves at the beck and call of OpenAI For-profit, even though the nonprofit is supposed to exert operational control over the for-profit,” said Robert Weissman, the co-founder of Public Citizen, a nonprofit consumer group, in a statement.
In March, the Japanese conglomerate SoftBank and other investors agreed to invest $40 billion in OpenAI. If OpenAI had not adopted its new corporate structure by the end of the year, SoftBank had the right to cut its total contribution to $20 billion from $30 billion.
SoftBank has agreed to make its full investment now that OpenAI has restructured, according to the same person familiar with the arrangement.
OpenAI also renegotiated its agreement with Microsoft. The deal was hashed out much like any major corporate deal, bringing in advisers from some of the largest investment banks. The nonprofit board was advised by Goldman Sachs and Michael Klein, the former Citigroup banker who has since formed his own bank. Microsoft was advised by Morgan Stanley.
The previous contract between the two companies gave Microsoft unfettered and exclusive access to all OpenAI technology to sell to customers and use in its own products until 2030. According to that contract, Microsoft would lose access to OpenAI’s most powerful technology if OpenAI’s board declared it had reached artificial general intelligence, or A.G.I., which is, roughly speaking, a machine that can do anything the human brain can do.
Microsoft and OpenAI have extended that agreement to 2032. The new contract provides Microsoft with access to technologies after OpenAI has reached A.G.I., provided that the technology is equipped with “appropriate safety guardrails.” An independent panel will judge whether OpenAI has achieved A.G.I.
Microsoft will continue to have access to OpenAI’s research technologies, before they become public products, until 2030 or until the expert panel declares that the start-up has reached A.G.I.
But the tech giant will not have rights to hardware developed by OpenAI, which has been developing various hardware prototypes after acquiring the firm founded by Jony Ive, the former Apple executive who designed the iPhone.
Under the new contract, OpenAI has the right to develop products alongside other partners, and Microsoft has the right to pursue A.G.I. on its own or alongside partners.
OpenAI has also committed to buying $250 billion in computing power from Microsoft, but is free to sign deals with other cloud computing providers. Previously, OpenAI needed Microsoft’s approval to sign other deals.
Now OpenAI can move into a courtship with Wall Street, though company executives have avoided saying much about that possibility.
Large public listings — in the vein of Saudi Aramco’s $25.6 billion I.P.O. — have not been common over the past several years, as investors have zeroed in on profitability, and large companies have found ample backing in the private market. But public investors’ support of cryptocurrency and A.I. could now open the doors to OpenAI going public.
Last summer, Mr. Musk sued OpenAI to prevent its change to a for-profit company. The suit is approaching a trial in Northern California, but no date has been set.
Near the end of 2024, Meta sent a letter to Mr. Bonta. The letter urged him to block the company’s restructuring, saying it would set a dangerous precedent by allowing start-ups to enjoy the benefits of nonprofit status before abandoning the arrangement once they find ways of becoming profitable.
OpenAI’s critics received new ammunition in late August when a couple in Southern California sued Mr. Altman and OpenAI, blaming ChatGPT for the death of their son.
Mr. Bonta and the Delaware attorney general, Kathy Jennings, discussed the teenager’s death with Mr. Altman and other executives. Mr. Bonta told The Times that he was encouraged by the changes OpenAI planned to make to ChatGPT, but he was still waiting to see how these plans would play out.
OpenAI has since added parental controls to ChatGPT that allow parents to oversee how their teenage children use the chatbot and are designed to notify them if their children show signs of self-harm.
(The New York Times has sued OpenAI and Microsoft, claiming copyright infringement of news content related to A.I. systems. The two companies have denied the suit’s claims.
Karen Weise contributed reporting from Seattle.
Cade Metz is a Times reporter who writes about artificial intelligence, driverless cars, robotics, virtual reality and other emerging areas of technology.
Lauren Hirsch is a Times reporter who covers deals and dealmakers in Wall Street and Washington.
The post OpenAI Restructures to Become a More Traditional For-Profit Company appeared first on New York Times.
