A nine-person advisory jury in Oakland needed roughly two hours to end the highest-profile lawsuit in the artificial intelligence industry, and it never touched the question everyone tuned in to hear answered. After a three-week trial, six women and three men decided that Elon Musk had simply waited too long to sue OpenAI, Sam Altman, and Microsoft over the company’s 2019 shift to a for-profit structure. U.S. District Judge Yvonne Gonzalez Rogers adopted the verdict the same day, saying there was “a substantial amount of evidence” to support it.
The headline answer searchers came looking for: the case was dismissed on a three-year California statute of limitations, not on whether OpenAI breached its founding charitable purpose. Musk called it a “calendar technicality.” OpenAI’s outside spokesman, Sam Singer, called it “a tremendous victory.” Both are right, and both can be right at the same time, because the jury never made it to the merits.
Two Hours, Nine Jurors, Three Weeks of Testimony
The trial opened in late April in the Northern District of California with Musk’s team seeking the dismissal of Altman and OpenAI co-founder Greg Brockman, plus the dismantling of the for-profit entity that has since recapitalized into a public benefit corporation. By the time closing arguments wrapped on May 14, the courtroom had heard from Musk himself, from Altman, from Microsoft chief executive Satya Nadella, from Brockman, and from a string of early OpenAI researchers.
The deliberation took less time than most closing-day commutes out of Oakland.
- Two hours of jury deliberation followed three weeks of testimony.
- Nine jurors reached a unanimous decision; the panel was an advisory one, meaning Judge Gonzalez Rogers held final authority on liability.
- $38 million was the total Musk contributed to the nonprofit before he resigned from the board in February 2018.
- $150 billion was the figure Musk’s side referenced as the value at stake in the for-profit conversion, per reporting on the trial’s opening arguments.
What the jury did NOT decide is the part of the story that matters most for the next plaintiff. Whether OpenAI breached a charitable trust, whether Altman and Brockman were unjustly enriched, whether the 2019 pivot violated the original mission to develop AI for the benefit of humanity: none of that was ruled on. The panel simply concluded that the wrongdoing Musk described, if it had occurred, had occurred far enough in the past that he could no longer sue over it.
The Clock Started in 2019, Not 2024
California Probate Code Section 16460 sets a three-year window to bring a claim against a trustee from the moment a beneficiary discovers, or reasonably should have discovered, the facts giving rise to the claim. The jury’s verdict turned on when that clock started ticking for Musk, and the answer they landed on was unsparing.
What the Jury Found About Musk’s Timeline
OpenAI’s lawyers spent weeks building a record that Musk had been aware of the for-profit subsidiary’s creation since March 2019, when the company publicly announced the “capped profit” structure to raise compute money. That announcement was followed by a $1 billion Microsoft investment the same year. The jury agreed that Musk was on notice no later than 2021, well outside the three-year window when he filed in February 2024.
Why the Discovery Rule Didn’t Save Him
Musk’s team argued the clock should have started later, pointing to the dramatic ballooning of OpenAI’s commercial business after ChatGPT launched in November 2022. The jury didn’t buy it. Under California law, a beneficiary doesn’t get to wait for the harm to grow before filing; the trigger is awareness of the breach, not the magnitude of its eventual consequence. Once Musk knew the for-profit arm existed, his timer was running.
What the Continuing-Violation Argument Couldn’t Do
Musk’s lawyer Marc Toberoff tried to get a “continuing violation” theory into the jury instructions, an argument that each new step in OpenAI’s commercial expansion was a fresh breach that restarted the clock. Judge Gonzalez Rogers declined to include it. That decision is now the single most important question heading into the appeal, because if the Ninth Circuit thinks the doctrine should have been available to the jury, the case can come back to life. If it doesn’t, the verdict is effectively final.
What Musk Conceded on the Stand
The most damaging moment for Musk’s case came during his own cross-examination, when he acknowledged that no written agreement governed the terms of his donations. That admission undercut the spine of the breach-of-charitable-trust theory, because California courts generally need a written instrument or a clear oral promise to establish a charitable trust of that scope. OpenAI’s attorney William Savitt went further outside court, telling reporters that the jury found Musk had lied during his testimony about the company’s origins.
This war is not over, and I would sum it up in one word: appeal.
That was Toberoff to reporters outside the courthouse. Musk’s own response landed on X within minutes, calling the ruling “a free license to loot charities if you can keep the looting quiet for a few years” and accusing the judge of using the jury “as a fig leaf.” The post was deleted shortly after. Musk’s history of litigating his own corporate disputes in public is by now a known feature of his playbook, but the legal odds rarely follow the volume of the posts.
Microsoft and OpenAI Walk Away With the Structure Intact
The biggest practical consequence of the verdict is not what happens to Musk; it’s what does not happen to the corporate structure that has taken five years and three rounds of regulatory negotiation to build. OpenAI completed its for-profit recapitalization in October 2025, creating OpenAI Group PBC, a public benefit corporation in which the nonprofit OpenAI Foundation holds a 26% stake. Microsoft holds roughly 27%, a position valued at about $135 billion, with intellectual property rights to OpenAI models running through 2032. Microsoft’s blog post announcing the next phase framed the deal as the resolution of nearly a year of talks with OpenAI and state regulators.
None of that was on the table when the jury walked out. Had Musk prevailed, the remedies he sought included unwinding the for-profit entity, removing Altman and Brockman, and reclaiming the assets of the nonprofit. Here is how the relief sought compared with what the verdict produced.
| Party | What Was at Stake | Outcome |
|---|---|---|
| Elon Musk | Restitution of donated funds; dismantling of the for-profit entity; removal of Altman and Brockman | All claims dismissed on timing; appeal filed |
| OpenAI / Altman | The for-profit conversion and the PBC structure completed in October 2025 | Structure unchallenged; merits never ruled on |
| Microsoft | 27% stake worth roughly $135 billion; IP rights through 2032 | Aiding-and-abetting claims dismissed as a matter of law |
| OpenAI Foundation | 26% controlling stake in PBC | Governance unchanged |
For Microsoft, the practical effect is that its largest single investment in the AI era is no longer carrying a live courtroom risk. OpenAI’s own published account of the dispute reproduces internal emails from 2015 to 2018 in which Musk, the filings indicate, supported a for-profit conversion and proposed Tesla as the acquiring entity. That history, paired with the timing finding, is the foundation OpenAI’s lawyers will lean on if the Ninth Circuit takes up the appeal.
The Appeal: Continuing Violation Doctrine and Long Odds
Toberoff has signaled that the appeal will rest primarily on the continuing-violation doctrine, the theory that a pattern of repeated breaches resets the limitations clock with each new act. Two legal experts who weighed in publicly after the verdict were both skeptical.
Carl Tobias, a law professor at the University of Richmond, told reporters that the jury’s decision was “very fact-based” and that appellate courts rarely overturn jury verdicts grounded in factual findings. Raffi Melkonian, an appellate lawyer, said winning an appeal of a jury verdict is generally “very difficult.” Both observations track the broader statistics: the Ninth Circuit reverses civil jury verdicts in roughly 12% to 15% of cases that reach merits review, and reversal rates drop further when the trial judge’s findings of fact align with the jury’s.
Three things on the appeal calendar will shape whether this is a real fight or a footnote:
- The jury-instruction objection. Toberoff will argue that Judge Gonzalez Rogers’ refusal to instruct on the continuing-violation doctrine was reversible error. The Ninth Circuit reviews jury-instruction questions de novo, which gives Musk his best opening.
- The discovery-rule cross-appeal opportunity. Musk’s side can argue that the discovery rule should have extended his clock based on facts that emerged only after the 2022 ChatGPT release.
- The credibility findings. The trial judge’s adoption of the jury’s verdict, paired with Savitt’s accusation that Musk lied on the stand, will weigh heavily on appellate review.
The Underlying Question Nobody Answered
For the broader AI sector, the verdict’s lasting value lies in what it leaves open. The transition of a research nonprofit into a $500 billion-valued public benefit corporation, with private investors holding majority economic interest, has now survived its most prominent legal challenge without ever being tested on the substance. That is a partial win for OpenAI and a green light for the next mission-driven AI lab considering a similar conversion. It is also, in the strict legal sense, no precedent at all on the merits.
If the Ninth Circuit upholds the verdict, the practical floor under future nonprofit-to-PBC conversions becomes the three-year clock. Any donor who watches a mission drift and waits more than three years from the public moment of drift to file a claim will face the same wall Musk just hit. If the appeal succeeds on the continuing-violation theory and the case comes back for a merits trial, the second act will be the one that actually decides what a charitable promise is worth when the assets are AI compute and the money is venture capital. Until that question gets a real answer, the score is one calendar dismissal, no merits ruling, and a six-year-old corporate decision that survives by clock, not by argument.
