Musk vs. Altman: The Trial That Will Decide Who Controls the Future of AI — Day-by-Day Report

The story of OpenAI is, at its core, a story about a big idea, an enormous bet, and a friendship that curdled into one of the most consequential feuds in the history of technology. It begins in 2015, at the dinner tables and backrooms of Silicon Valley, where a small group of technologists had become convinced that artificial intelligence was the most important thing humanity would ever build — and that it was developing in exactly the wrong hands.

That group included Elon Musk, who was already the most famous entrepreneur in the world; Sam Altman, then president of Y Combinator and one of Silicon Valley’s most respected operators; Greg Brockman, a brilliant engineer; Ilya Sutskever, one of the world’s foremost AI researchers; and Shivon Zilis, a venture capitalist who would become, over the following years, one of the most significant and most quietly controversial figures in the entire OpenAI saga.

Their shared worry: Google and its DeepMind division were consuming the world’s best AI talent, pointing it at commercial products, and operating in closed labs beyond public scrutiny. The group believed that if artificial general intelligence — AI matching or exceeding human intelligence — was going to be developed, it needed to be developed in public, with its research freely shared, governed by people who prioritized human benefit over shareholder returns.

And so in December 2015, OpenAI was born — as a nonprofit, with Musk and Altman as co-chairs, and a founding charter that committed to developing AI “for the long-term benefit of humanity.” Musk pledged up to $1 billion in donations. He ultimately contributed $38 million before leaving the board in 2018. What happened between that founding moment and the Oakland courtroom in 2026 is the subject of the most important technology lawsuit in history.

Musk’s lawsuit, filed in 2024 and finally reaching trial in April 2026, makes three core claims that his legal team must prove to a jury:

The exchange about Microsoft’s investment of $10 billion was particularly revealing. Musk testified that when he learned about the deal, he “reacted quite negatively” and texted Altman something like “What the hell is going on?” He said the scale of Microsoft’s involvement — and the implied expected return — convinced him that OpenAI had ceased to function as a charity. “With all due respect to Microsoft, do you really want Microsoft controlling artificial general intelligence?” Musk told the jury.

Throughout Day 2, Judge Gonzalez Rogers repeatedly intervened, urging Musk to answer simple questions directly. When Savitt pointed out to the judge that he was having difficulty getting direct answers, Rogers responded with dry humor: “That is the challenge of all litigants.”

Both sides have introduced documents that they claim are decisive. Here are the most significant pieces of evidence and what each side says they prove:

In any other trial, the key witness might be a CFO, a board member, or a whistleblower. In this trial, the most consequential potential witness is a woman who shares four children with the plaintiff, has worked at companies he controls, and sent emails to his employees while serving on the board of the defendant organization. Shivon Zilis may be the most complicated witness in the history of technology litigation.

Here is what we know for certain about Zilis’s role: She was a venture capitalist who joined OpenAI’s board. She sent emails discussing potential corporate restructuring to an employee of Elon Musk. She now works at Neuralink — one of Musk’s companies. She lives with Musk. She is the mother of four of his children. And she may be called to testify in a trial where both sides believe her testimony supports their version of events.

OpenAI’s lawyers want the jury to understand Zilis as a conduit between Musk and OpenAI’s leadership — someone whose emails prove Musk was kept fully informed about the for-profit discussions and, crucially, did not object to them as long as he was in control. Musk’s lawyers want the jury to understand Zilis as an OpenAI insider whose emails reflect her own communications, not instructions or knowledge that Musk acted upon.

The truth is almost certainly somewhere in the complicated middle of a personal and professional relationship that is genuinely unlike anything a court has previously been asked to parse. The emails exist. The relationship exists. What the jury makes of both will be central to its verdict.

OpenAI’s defense strategy, articulated by attorney William Savitt in his opening statement and cross-examination, rests on a single devastating argument: Musk’s objection to OpenAI’s for-profit structure was never really about protecting a charitable mission. It was about protecting his own control over the organization — and when he lost that control, he left, and is now trying to use the courts to get it back.

The evidence OpenAI has introduced to support this narrative is substantial. Emails from the early for-profit discussions show Musk explicitly seeking “initial control” and the majority of the capitalization table. The Zilis emails show that restructuring plans — specifically, the hybrid nonprofit/C-Corp model — were discussed with people in Musk’s direct orbit. And Musk’s own 2020 X post shows he publicly acknowledged Microsoft’s relationship with OpenAI three years before filing his lawsuit.

OpenAI also targets the statute of limitations directly. If Musk knew about the Microsoft relationship — and his own post suggests he did — then his 2024 lawsuit may have been filed too late under California law. Microsoft’s lawyers pressed this point in their opening statement, noting the September 2020 post as evidence that Musk’s claims had already expired before he filed them.

Microsoft’s presence in this trial is significant and, for a company whose CEO Satya Nadella is scheduled to testify, uncomfortable. Musk accuses Microsoft of aiding and abetting OpenAI’s alleged breach of charitable trust through its investments — effectively arguing that Microsoft’s money enabled the abandonment of OpenAI’s nonprofit mission.

Microsoft’s defense lawyer, Russell Cohen, offered a straightforward counter in his opening statement: Microsoft did not and could not have aided a breach it didn’t know was occurring. Cohen also pointed to the statute of limitations argument, noting that Musk’s own 2020 X post — in which he described OpenAI as “essentially captured by Microsoft” — suggested Musk had knowledge of the relationship years before filing suit.

Separately, the week of the trial also saw Microsoft and OpenAI announce a revamped version of their partnership — with the explicit dissolution of their exclusivity agreement. The timing was awkward for both companies but legally was presented as business-as-usual evolution of a partnership. Microsoft’s investment in OpenAI, in its original form, exceeded $13 billion across multiple tranches since 2019.

The verdict will have consequences far beyond the parties in the courtroom. Here is the realistic impact of each outcome:

Judge Yvonne Gonzalez Rogers of the U.S. District Court for the Northern District of California has presided over the case with evident control of an exceptionally difficult room. She has managed Musk’s combativeness with dry humor and judicial firmness, intervening during cross-examination to urge direct answers and handling the social media situation with measured authority.

The trial proceeds in two phases. Phase 1 — currently underway — asks the jury to determine whether the defendants are liable for the claims Musk has made. Phase 2, which would only occur if Musk prevails in Phase 1, would determine the specific remedies. Additional witnesses expected to testify include Sam Altman, Satya Nadella, and potentially Shivon Zilis.

This trial is, at its most fundamental level, a story about what happens when the promises we make before we understand the stakes meet the reality of what we built. Musk promised money he didn’t fully deliver. Altman promised openness he didn’t fully maintain. And somewhere in the gap between those two failures, an $800 billion company was born — and a federal jury in Oakland, California, is now being asked to decide who was right.