How Elon Musk and Sam Altman went from best friends to adversaries

In December 2015, Elon Musk and Sam Altman sat down for an interview at the Vanity Fair New Establishment Summit in San Francisco, publicly touting their new partnership as co-chairs of the fledgling Artificial Intelligence Research Institute.

Musk became a billionaire thanks to his shares. teslawent public five years ago, and Altman ran Y Combinator, a well-known startup incubator. The two had worked closely together to thwart AI efforts that year. google Preventing the establishment of monopolistic control over powerful technologies. Their nonprofit project was called OpenAI.

Over the past three weeks, the collapse of the once-strong bond between two of the AI ​​world’s most prominent figures has been the subject of a high-profile trial in Oakland, California, after Musk sued Altman and OpenAI for violating a promise to keep OpenAI as a nonprofit organization in 2024. OpenAI is now valued at more than $850 billion, while Musk’s SpaceX is valued at $1.25 trillion after merging with his AI lab xAI in February.

Both companies are competing for the public market, with SpaceX expected to release its prospectus as early as this week ahead of a record-breaking public offering next month. Before addressing eager investors, Musk had to testify to a jury in downtown Oakland to prove his point, but if successful, it could add a major twist to OpenAI’s ambitious plans.

“What you shouldn’t do is have your cake and eat it too,” Musk said on April 29 in response to questions from OpenAI’s lawyers. He accused Altman and Greg Brockman, OpenAI’s president and another co-founder, of using philanthropy to line their own pockets while also reaping the positive connections that come from running a nonprofit organization.

Musk used his time on the witness stand to emphasize a message he has been shouting for years on his social media app X, also now owned by SpaceX: Without him, there is no OpenAI.

“I came up with the idea and the name, hired the key people, taught them everything I know, and provided all the initial funding,” Musk said.

Mr. Altman testified last week that he and his co-founders had not made any promises to Mr. Musk about the company’s corporate structure.

He argued that a big problem from the beginning was that Musk felt very strongly about having complete control over OpenAI, at least initially, in part because he didn’t trust others to make decisions.

Altman testified that he was “very uncomfortable” with Musk’s pursuit of power.

Lawyers for Musk and OpenAI concluded their closing arguments on Thursday after a three-week trial. The jury will begin deliberations on Monday to determine the merits of Musk’s claims and whether OpenAI, Altman and Brockman should be held liable for breach of charitable trust and unjust enrichment.

Stavros Gadinis, a professor at the University of California, Berkeley School of Law, said that regardless of the final outcome, neither tech giant is likely to prevail in the court of public opinion.

“After weeks of damaging testimony, the public is left to choose between two dueling billionaires, each convinced that he is the rightful custodian of transformative technology,” Gadinis said in an email. “The answer most people arrive at is neither.”

The partnership began 11 years ago, in May 2015, when Altman emailed Musk to ask if he thought it was a good idea for Y Combinator to launch a “Manhattan Project for AI.” Musk said the idea is “probably worth talking about.”

OpenAI was launched in December 2015, with Musk pledging up to $1 billion in funding for philanthropy.

“I’m very impressed with everyone so far,” Musk wrote to Altman in November 2015, according to an email released as part of the scandal. “This is a great team.”

By 2017, cracks began to appear.

As OpenAI progressed in research and development, Musk required Altman and other co-founders, including Brockman and Ilya Sutskeva, to compile a list of employees and their key contributors and immediately fire anyone who did not meet their ratings, according to the filing.

OpenAI was burning cash and needed more computing resources. Leaders discussed converting the lab into a commercial facility. The question of who should be CEO and control the stock was especially troubling for Musk, who wanted as much as 90% ownership of the commercial company.

Altman and his other co-founders rejected this, arguing that no one person or group should have unilateral control over “artificial general intelligence” technology that could prove to be smarter than humans.

A key point of tension in the relationship became apparent in June 2017, when Tesla poached AI researcher Andrei Karpathy from OpenAI. Text messages between Musk and employees, including OpenAI board member Siobhan Gillis and project director Sam Teller, show that Musk’s team was rooting for the hire, according to communications released in discovery.

Gillis, who has four children with Musk, took the stand earlier this month and was questioned by lawyers on both sides about conversations she had about OpenAI’s corporate structure around 2017 and 2018 and whether Tesla was trying to poach OpenAI employees while she was on the board.

When an OpenAI lawyer showed Mr. Gillis a text message from Mr. Karpathy congratulating him on accepting Mr. Musk’s job offer, Mr. Gillis acknowledged that Mr. Musk had approached Mr. Karpathy first.

Brockman later recalled during his testimony that Musk offered an “apology and confession” to OpenAI’s co-founders.

While drama erupted within the company, OpenAI’s technology continued to advance. By August 2017, the system was able to beat the world’s top players of Dota 2, a multiplayer action game. Musk touted his accomplishments on Twitter.

“This is the first time OpenAI has defeated the world’s best players in competitive esports,” Musk wrote. “It’s much more complex than traditional board games like chess or Go.”

A month later, he told Altman and other OpenAI leaders in an email that “enough is enough.” Even if he couldn’t control OpenAI, he was ready to walk away.

“Either do something on your own or continue with OpenAI as a nonprofit,” Musk wrote in an email revealed in a court filing. “I will no longer fund OpenAI unless you commit to staying. Otherwise, I’m just a fool giving you free money to build a startup.”

Musk has cut off his monthly contributions to the company. Instead of $1 billion, his donations totaled about $38 million.

With OpenAI desperate for help, Musk, Gillis and Teller made a final push to bring the lab under Musk’s control, suggesting it should be integrated into Tesla. To sway Altman, Musk’s team invited him on a tour of the Tesla factory and promised him a seat on Tesla’s board of directors.

In his testimony, Altman said he didn’t think that was appropriate and worried that the nonprofit would effectively be destroyed if it became part of Tesla.

“Tesla is a car company and does not have an OpenAI mission,” Altman said on stage. “I don’t think we had the ability to ensure that the mission was accomplished.”

After the merger effort was rejected, Musk wrote in a December 2018 email to Altman and OpenAI executives that “absent dramatic changes in execution and resources, my assessment of the likelihood of OpenAI being associated with DeepMind/Google is 0%, not 1%.”

Mr. Musk testified that the for-profit affiliates OpenAI has created have become “the tail wagging the dog,” violating the founding charity’s mission and the promises its founders allegedly made to Mr. Musk.

Musk left OpenAI’s board in 2018, and OpenAI said in a blog post that the move was necessary to “eliminate a potential future conflict with Elon” as Tesla began to focus on AI.

For the next five years, Musk rarely mentioned OpenAI in public. And the rift in her relationship with Altman was barely visible on social media.

Altman has regularly praised Musk on Twitter, writing in 2019 that “betting on Elon is a historic mistake” and posting in October 2022 that Musk serves as a “reminder of how much one person can do.”

The latter post was posted a month before OpenAI introduced ChatGPT. That’s when everything changed, as the generative AI boom led to a surge of investment in the field. In January 2023, microsoft Pours $10 billion into OpenAI, making it clear that the race to commercialization is underway. OpenAI had already established a commercial subsidiary.

Musk began bashing Altman and OpenAI online. In a post on Twitter, he went on a rant about the company’s funding and partnership with Microsoft, which had by then been acquired by Musk and later renamed X.

“OpenAI was founded as an open source (hence the name ‘Open’ AI), non-profit company to compete with Google, and now it has become a closed source, most profitable company effectively controlled by Microsoft,” Musk said in a February 2023 post, “which is not at all what I intended.”

Altman responded in a statement released in a court filing.

“Thank you so much for everything you’ve done to help me. OpenAi wouldn’t have happened without you, and I find it truly heartbreaking that you would attack OpenAi in public,” Altman wrote to Musk.

Mr. Musk was undaunted. We incorporated xAI in March 2023 and aggressively hired talent from the company with the aim of becoming a direct competitor to OpenAI. Gillis, who had a child with Musk at the time, resigned from OpenAI’s board.

She was heading in that direction last month, texting a friend: “If the father of your baby wants to compete and hire from OpenAI, there’s nothing you can do.”

Since filing the lawsuit in 2024, Musk has escalated a war of words with OpenAI’s co-founders, primarily on his preferred platform, X, calling the two top leaders “fraud Altman” and “Greg Stockman.”

“We could have started OpenAI as a for-profit company,” he wrote in a post to X just as the trial began. “Instead, I launched it, funded it, hired key people, and taught them everything I knew about how to run a successful startup for the public good. And they stole the philanthropy.”

—CNBC’s Kate Rooney contributed to this report.

WATCH: Musk vs. OpenAI trial week 3: Here’s what you need to know