The New York Times: Elon Musk says he ‘was a fool’ to provide OpenAI’s early funding

It was 2018, and Elon Musk was having a very busy year.

His electric car company, Tesla, was struggling with manufacturing problems. He had a plan to take it private, but he kept angering regulators. SpaceX, his rocket company, was just starting to show momentum. And OpenAI, the non-profit artificial intelligence lab he was supposed to be involved in, wasn’t getting much of his attention.

“I did not even have the time to attend board meetings,” Musk said Wednesday, US time, the second day of his testimony in a trial that pits him against OpenAI, two of its co-founders and the company’s giant partner, Microsoft.

That year, the seeds of today’s court fight were sown. Executives at OpenAI, which was founded as a non-profit, were considering attaching a for-profit company to its structure. Musk did not have a problem with that, he said, but he wanted to make sure the non-profit was still in charge.

Now he regrets that he did not play a more active role before he left the lab in early 2018 and that he gave it money to get off the ground in the first place.

“I was a fool who provided them free funding to create a startup,” Musk said in response to a question from his lawyer, Steven Molo. “I gave them $38 million of essentially free funding to create what would become an $800 billion company.”

In what is expected to be a month-long trial in federal court in Oakland, California, Musk is arguing that OpenAI breached its founding contract when it took on Microsoft as an investor and started creating commercial products. He is asking for $US150b in damages and for OpenAI to unwind the for-profit company it created last year.

Today, OpenAI is one of the tech industry’s most influential companies. Musk, of course, emerged from that rough year to become the wealthiest man in the world. But the relationships he had at the time with Sam Altman, OpenAI’s CEO, and Greg Brockman, the company’s president, have long since been severed.

The hard feelings between the tech tycoons have been difficult to miss in the trial. Musk and Altman have known each other for years and hobnobbed in the same elite Silicon Valley circles. But mutual admiration — which led to their creation of OpenAI in 2015 — has soured to unapologetic animosity.

As Altman and Brockman watched from the courtroom gallery, just behind their lawyers, Musk, the first witness in the trial, continually described his former co-founders as a deceitful pair who duped him and steered OpenAI away from its altruistic roots.

Musk described how OpenAI evolved after he left the lab in early 2018, though he continued to donate money to the lab and received updates on its progress. About a year later, Microsoft invested $1 billion in OpenAI — a hint that it was starting to take its commercial potential seriously.

Musk said he was aware of Altman’s efforts to create a for-profit company and to raise money from Microsoft. But the company had put a cap on the profits that would flow to investors, so he said he did not think it was a big problem.

OpenAI and Microsoft also had a plan to dissolve their partnership if OpenAI created artificial general intelligence, essentially a machine that can do anything the human brain can. That provided additional reassurance, Musk added.

But he said he became very concerned when Microsoft said in early 2023 that it had invested $10 billion in OpenAI. That was after the release of ChatGPT, which quickly became an industry sensation. Musk said he sent a text to Altman asking, “What the hell is going on?” and calling the investment a “bait and switch.”

“They had essentially turned the nonprofit into a company with a $20 billion valuation,” he said.

Microsoft “is not going to put $10 billion into something unless they think they are going to get a very big return,” he added. “There is no other way they are going to think about it.” Altman and OpenAI offered him equity in the company after Microsoft’s investment, Musk said. But he did not take it.

“Frankly, it felt like a bribe,” Musk said.

But William Savitt, OpenAI’s lead counsel, implied that Musk’s testimony contradicted what he had said in depositions before the trial. Did he donate $38 million to OpenAI — as he said in his testimony — or $100 million? In a deposition, he said it was $100 million.

Their exchange grew combative. “Your questions are not simple,” Musk said. “They are designed to trick me, essentially.”

Savitt also has a complicated relationship with Musk. He once represented Musk and Tesla in a securities fraud lawsuit over the carmaker’s acquisition of the solar company SolarCity. Then he switched sides and represented Twitter against Musk, after the billionaire tried to back out of his 2022 acquisition offer for the social media company.

After Musk completed his purchase of Twitter, now X, the company sued Savitt’s law firm, Wachtell, Lipton, Rosen Katz, claiming that the firm’s earnings of $90 million for its representation of Twitter under previous management was “unjust enrichment.” The social media company dismissed its lawsuit last year.

Savitt returned to a number of emails Musk had discussed in his earlier testimony and showed how they could be interpreted very differently. One email, Savitt said, demonstrated that as OpenAI explored a shift to a for-profit model in late 2017, Musk wanted a large stake in the organisation and full control. That included the ability to choose a majority of the board members.

Savitt displayed other old emails where Musk promised the other OpenAI founders that he would create a for-profit company with him in control.

As his testimony stretched into the afternoon, Musk became visibly frustrated with Savitt, calling his questions “definitionally complex.”

Musk also lashed out at Altman. Savitt displayed a 2018 email in which Shivon Zilis — a longtime employee of Musk’s and the mother of four of his children who was also a former OpenAI board member — asked Musk if she should stay close to OpenAI to keep feeding him information on the company.

Musk responded that Altman “did not keep the board informed, which was why he was fired from the company.” OpenAI’s board briefly forced out Altman in late 2023, but Zilis was not on the board at the time. He returned after five days of negotiations.

Musk now has his own AI company, xAI, which has become part of SpaceX. Savitt implied that Musk’s suit was just an attempt to delay OpenAI’s progress as his engineers improved xAI’s chatbot, Grok.

The trial and Musk’s testimony are expected to continue Thursday.