Musk Sued OpenAI for Becoming the Company He Tried to Build Himself

Sam Altman ran OpenAI the way most corporate CEOs run theirs: pitting lieutenants against each other, saying different things to different people, creating the kind of internal chaos that makes outside observers wonder what the organization actually stands for.

That is the picture that emerged from sworn testimony in federal court this week. Ilya Sutskever, OpenAI's former chief scientist, confirmed under oath that Altman exhibited a consistent pattern of lying and pitting executives against each other. Mira Murati, the former technology chief, testified that Altman routinely told different people contradictory things. Satya Nadella, the Microsoft CEO whose company has $13 billion invested in OpenAI, called the board's brief firing of Altman in November 2023 amateur city.

The jury in Elon Musk's lawsuit against OpenAI found 9 to 0 that Musk had waited too long to sue — deliberations lasted under two hours. The verdict was a statute-of-limitations dismissal, not an adjudication of whether OpenAI broke its founding mission. Musk was asking for $150 billion; he got a procedural exit.

But the trial record is now a permanent public document. And what it shows about how Altman actually governed the organization — regardless of what its nonprofit status implied — is more consequential than the verdict.

The question the court never reached is whether OpenAI's founding documents actually constrained Altman's management, or whether they gave him the same latitude the law grants any corporate board. American corporate law, following a 1919 Michigan Supreme Court case called Dodge v. Ford, holds that company directors have broad discretion to prioritize stakeholders, community, or mission over pure shareholder enrichment — as long as they don't entirely abandon the financial interests of owners. Ford v. Dodge is often cited as the doctrine that lets corporate boards act like stakeholder-first entities. What it actually decided was narrower: that directors can't systematically sacrifice shareholder returns for other purposes without a business justification. The latitude runs both ways.

OpenAI's conversion in October 2025 — its nonprofit retaining 26 percent, Microsoft taking 27 percent of the new for-profit subsidiary, OpenAI Group PBC — was the mechanism that put the question of what the mission actually required front and center. If the mission constrained how Altman could govern, that structural conversion should have required structural accountability. The trial testimony from Sutskever and Murati describes a governance pattern that looks like standard corporate management, not mission-constrained stewardship. Whether that gap — between what OpenAI's structure implied and how it actually operated — constitutes a breach of the founding agreement is exactly what the court never adjudicated.

Musk's own internal emails, published by OpenAI in December 2024, added a structural irony the courtroom could not weigh. In September 2017, Musk incorporated his own public benefit corporation called Open Artificial Intelligence Technologies Inc. — designed as the proposed future structure of OpenAI. He demanded 90 percent equity, absolute control, and the CEO title, according to reporting by CNBC. When the co-founders rejected those terms, he wrote in an email that OpenAI was on a path to certain failure unless it merged into Tesla. He left the board in February 2018. Altman sent him a term sheet outlining the for-profit plan later that year. Musk says he never read it.

Musk had donated roughly $38 million to $44 million to OpenAI, not the $1 billion he publicly claimed.

OpenAI closed a funding round in March 2026 at an $852 billion post-money valuation, with a possible $1 trillion IPO later this year, The Guardian reported. OpenAI's for-profit subsidiary, OpenAI Group PBC, was formed in October 2025; the original nonprofit retains 26 percent, Microsoft holds 27 percent. OpenAI's own CFO has suggested publicly the company should not have to go public.

Musk, for his part, had already moved past this fight before the verdict came down. In late January or early February 2026, SpaceX acquired the rest of xAI in an all-stock reverse triangular merger. The combined entity was valued at roughly $1.25 trillion, according to CNBC. The IPO pipeline — SpaceX targeting July with a potential $75 billion raise — was already in motion.

OpenAI has since disbanded its alignment team, according to Verfassungsblog, despite repeated lawsuits over releasing models too quickly. Stanford's 2025 AI transparency index placed OpenAI second-to-last among major AI companies, down from second place previously.

Musk plans to appeal to the Ninth Circuit. Neither verdict nor appeal will answer whether OpenAI's founding promise — that the company would develop AI safely and for the benefit of humanity, not for profit — meant anything enforceable at all. The trial record shows how each of them actually governed the organizations they built. What it doesn't show is whether the mission was ever more than a legal form with no binding force.

What the verdict decided: Musk didn't need the courtroom. He needed the merger. He got both — one on his terms, one on a technicality.

What it did not decide: whether the mission was ever fulfilled. Whether it was ever enforceable. Whether the promise mattered.