Nine jurors in California are now deliberating over one of the messiest disputes in AI history. On the surface, Elon Musk’s case against OpenAI’s other cofounders and Microsoft sounds straightforward: did the company betray its nonprofit mission by pivoting to for-profit motives? The jury will answer some narrow questions. But the implications are anything but narrow.
If Musk prevails, OpenAI as a for-profit operation could cease to exist. That’s the headline. What actually happens next is so unclear that the judge has already scheduled hearings for next week where both sides will argue about consequences. A verdict against Musk could render all that moot anyway.
The Betrayal Narrative
Musk’s team builds their case on a simple premise: the defendants knew exactly what he wanted. He wanted to support a nonprofit that would keep AI benefits accessible to everyone, free from corporate control. When Microsoft invested $10 billion into OpenAI’s for-profit affiliate in 2023, Musk’s attorneys say that crossed a line.
This wasn’t just another funding round. Previous investments had come earlier, before the statute of limitations. This one was different. It was the moment, they argue, when Musk’s worry calcified into certainty: his cofounders had sold out.
The Microsoft deal created something tangible to point at. Suddenly there were multibillion-dollar valuations. Founders like Greg Brockman and Ilya Sutskever held stakes worth fortunes. Microsoft itself stood to gain enormously. If Musk’s money had been meant for a charitable mission of AI safety, how did it end up enriching these particular people?
Musk didn’t sue until mid-2024, though. He’d known about OpenAI’s direction for years. He’d tweeted criticisms. He’d apparently only decided he’d been betrayed in fall 2022, when he learned about the Microsoft investment plans. That gap between knowing and suing matters, and OpenAI’s lawyers aren’t shy about pointing it out.
The Defense: It Was Always Going to Be This Way
OpenAI’s attorneys have spent the trial asking a recurring question: where are the restrictions? They asked Musk’s own financial adviser. They asked his chief of staff. They asked his special adviser, Shivon Zilis. None could point to specific conditions placed on his donations that would have prevented a for-profit structure.
Everyone involved, they argue, understood that private fundraising would be necessary. Musk himself tried to launch a for-profit that he would personally control. He later suggested merging OpenAI into Tesla. These weren’t the moves of someone who believed nonprofits should stay pure and separate from commercial activity.
A forensic accountant testified that Musk’s donations had already been fully spent by August 5, 2021. That’s significant. It means any charitable trust tied to those funds was already exhausted well before Musk filed his lawsuit. You can’t claim breach of a charitable mission if the money was already used up years earlier.
OpenAI’s team also emphasizes the work itself. Yes, there’s a for-profit doing the heavy lifting. Yes, it’s generating nearly $200 billion in equity value. But that value supports the nonprofit foundation. And the nonprofit board, they maintain, still controls the for-profit. ChatGPT is free. That distributes AI benefits to the world, which was supposedly the mission all along.
The evidence also cuts another way. Zilis, who is the mother of three of Musk’s children, voted to approve these transactions as an OpenAI board member while not disclosing her personal relationship to other board members. Musk, meanwhile, was building competing AI efforts at Tesla while still chairing OpenAI. He even hired OpenAI employees to work on AI projects elsewhere. OpenAI’s lead attorney, Bill Savitt, told the jury bluntly: “Mr. Musk abandoned OpenAI for dead in 2018.”
The Microsoft Question
Musk’s case zeroed in on “the blip” in 2023, when Sam Altman was fired by OpenAI’s nonprofit board and then rehired days later. Microsoft CEO Satya Nadella was personally involved in bringing Altman back. That involvement matters to Musk’s team. They see it as the moment when a commercial interest superseded nonprofit governance.
Microsoft’s agreement includes a clause giving the company veto rights over major corporate decisions. Musk’s attorneys point to internal wondering among Microsoft executives about whether their commercial agreement might conflict with OpenAI’s nonprofit goals. Did that conflict actually exist? Did Microsoft’s commercial priorities corrupt OpenAI’s mission?
Microsoft’s witnesses say no. They insist company executives didn’t know about any specific conditions on Musk’s donations despite doing extensive due diligence. They never actually used their veto rights, they claim. And Microsoft’s investments and compute power enabled OpenAI’s actual breakthroughs.
That’s a convenient claim, but it’s not obviously false either. The trial hasn’t produced smoking-gun evidence of Microsoft explicitly telling OpenAI to abandon safety work or compromise its mission. What it has produced is the architecture of incentives: everyone benefits from a successful for-profit.
Why This Matters Beyond the Courtroom
The Business stakes here involve more than just OpenAI’s corporate structure. A verdict for Musk could force a restructuring of the world’s most consequential AI lab eight years after he last formally participated in it. That seems almost absurdly backward until you remember the counterargument: if a nonprofit’s mission was genuinely betrayed, the timeline shouldn’t matter.
But there’s another layer. Technology companies routinely blend nonprofit and for-profit structures. Universities do the same. The question the jury is actually answering is whether that blending itself constitutes betrayal when the original founders had different intentions.
OpenAI’s attorneys argue that their current structure has existed since the first Microsoft investment in 2018. Forcing a restructuring now would be unreasonable. Musk’s team counters that the 2023 investment was categorically different because it came after the statute of limitations on earlier ones, changing the character of the enterprise.
The jurors have to decide if Musk’s donors’ intent was ever actually restricted by the organization’s behavior, or if everyone involved simply understood that you need money to build something real. They have to weigh whether a nonprofit board that lost control of its for-profit subsidiary mid-crisis still meaningfully “controls” it. They have to assess whether ChatGPT being free fulfills an AI safety mission or whether safety work itself was compromised.
The narrow questions the jury will answer don’t feel narrow at all once you start thinking about what they mean.
