Musk v. Altman: $150B OpenAI suit dies in 90 minutes

The verdict that wasn’t about the verdict

Eleven days of testimony. Three years of pretrial maneuvering. A $150 billion claim, the largest civil number ever attached to an AI company. And on Monday morning, May 18, 2026, an Oakland jury needed less than two hours to make all of it go away. The nine-member advisory panel returned a unanimous verdict that Elon Musk’s lawsuit against OpenAI and Sam Altman was barred by the statute of limitations, and Judge Yvonne Gonzalez Rogers adopted the finding from the bench the same morning. Per NPR’s verdict coverage, the case never reached the merits. The jurors did not decide whether Sam Altman and Greg Brockman “stole a charity,” whether OpenAI breached its founding covenant, or whether the world’s most consequential nonprofit pivot was an act of unjust enrichment. They decided only that Musk had filed too late.

That technicality is not a small thing. It is the legal seal that turns OpenAI’s October 2025 recapitalization from a contested experiment into a precedent, and it locks Microsoft’s 27% stake in the new OpenAI Group PBC — worth roughly $135 billion at current marks per CNBC’s coverage of the restructure — into safe legal territory. It also protects the OpenAI Foundation’s 26% equity, valued at approximately $130 billion, from a disgorgement claim that could have erased the largest philanthropic endowment in history before it ever wrote a check. Per TechCrunch’s writeup of the recapitalization close, the conversion was always contingent on regulatory tolerance in California and Delaware and on Musk’s litigation falling short. With Monday’s verdict, the last private-law barrier dropped.

The stakes were never abstract. Musk had asked the court to force OpenAI and Microsoft to “disgorge” up to $150 billion into the nonprofit foundation, a number that would have functionally re-nationalized OpenAI as a charitable institution and triggered a Microsoft writedown larger than the entire market capitalization of all but the ten biggest U.S. public companies. Per CBS News’s coverage of the dismissal, the advisory jury found Musk knew about OpenAI’s commercial drift before August 2021, the three-year cutoff that California’s limitations regime imposes. Once that finding was on the record, every other claim — breach of fiduciary duty, fraud, unjust enrichment — collapsed into a procedural footnote. The merits were never tested. The deadline was the door, and Musk arrived three years late.

What the verdict actually validates is the most aggressive corporate metamorphosis in modern technology. OpenAI began in 2015 as a 501(c)(3) charitable nonprofit committed to building artificial general intelligence “for the benefit of all humanity.” It now operates as OpenAI Group PBC, a Delaware public benefit corporation, with the nonprofit Foundation reduced to a 26% equity holder and Microsoft, SoftBank, and a constellation of private-equity vehicles holding the rest. Per NBC News’s recap of the trial’s two-hour finale, the structure that emerged is now legally protected against the only credible private claimant capable of unwinding it. Other regulators retain authority, but no private litigant with comparable standing remains. The Pandora’s box is closed. The recapitalization is durable. And the entire frontier-AI industry just received a playbook that says you can build a $500 billion company on a charitable foundation if you survive the calendar.

The mechanics of a 90-minute verdict

A statute of limitations is rarely the protagonist of a major commercial trial. It is, instead, the silent metronome behind every fiduciary claim — the rule that says you cannot wait until the value of a wrong becomes obvious before you complain about it. California’s three-year window for breach of fiduciary duty and the related charitable-trust claims began running, according to OpenAI’s defense, the moment Musk had constructive knowledge that OpenAI was abandoning its strictly nonprofit posture. The defense placed that date in 2019, when OpenAI created its first capped-profit subsidiary and Microsoft made its initial $1 billion investment. By that arithmetic, Musk’s window closed in 2022. He filed in February 2024. Per MIT Technology Review’s explainer on the verdict, the jury accepted the constructive-knowledge framing and concluded that no reasonable person in Musk’s position could have missed the signal.

The internal evidence helped OpenAI cement that timeline. Per reporting from Moneywise on the trial emails, Greg Brockman and Ilya Sutskever wrote a 2017 letter to Altman that included the phrase “we don’t understand why the CEO title is so important to you” and questioned whether AGI was Altman’s “primary motivation” or whether it connected to political ambitions Altman had not disclosed. Musk’s own response to a separate 2017 thread, sent under the subject line “honest thoughts” from worried co-founders, read “I’ve had enough” before he suggested folding OpenAI into Tesla. Those documents were meant by Musk’s team to show the conflict at the company’s core. The jury read them differently — as evidence that Musk had detailed, contemporaneous knowledge of the very governance fight he later claimed surprised him. A plaintiff who watched the deal go sideways in 2017 cannot credibly claim to have discovered the wrong in 2024.

The other quiet weapon was Altman himself. Per CNN’s coverage of Altman’s cross-examination, Altman spent most of his testimony deflecting — naming Musk as the “real control freak,” acknowledging a brief flirtation with the California gubernatorial race, and casting the for-profit pivot as a survival decision forced by the cost of training frontier models. That deflection did not need to be persuasive on the merits. It needed only to fragment the narrative enough that the jury would seize on the cleaner question first. The cleaner question was the calendar. Once the panel concluded Musk had three years of warning signs to act on and waited five, every other contested fact became unnecessary. The deliberation took roughly 90 minutes because the calendar question is the kind a juror can answer without a whiteboard.

The financial architecture the verdict protects is worth specifying. Per OpenAI’s $122 billion December funding announcement, the company closed its latest round at a post-money valuation of $852 billion, with $122 billion in committed capital and an enterprise revenue base of roughly $24 billion annualized as of Q1 2026. The OpenAI Foundation’s 26% equity stake, per the recapitalization announcement on the company’s own site, is currently marked at approximately $130 billion. The Foundation also holds a warrant that pays out additional equity if OpenAI Group’s share price rises tenfold over fifteen years — a structure that, if successful, would make the Foundation one of the largest philanthropic vehicles in the history of capital. Musk’s complaint, had it survived to merits, would have argued that every dollar of that equity sat on a foundation of breached charitable trust and should be clawed back into the nonprofit’s direct control. The verdict means none of it will be.

Stakeholder	Stake / Position
Microsoft equity	27% (~$135B mark)
OpenAI Foundation	26% (~$130B mark)
Musk’s contribution	$44M (2016–2020)
Damages sought	Up to $150B disgorgement

What the table makes obvious is the asymmetry. Musk’s original $44 million in personal contributions to OpenAI between 2016 and 2020, per the CNBC trial summary, was a meaningful early gift that bought him a board seat, a co-founder credential, and significant influence over the company’s first three years. It was not a controlling investment, and the trial record makes clear it never came with the kind of written governance lock-in that would have made the later for-profit pivot a breach. The legal theory of the case was that the breach lived in OpenAI’s public charitable representations rather than in Musk’s private contributions — a charitable-trust claim, not a contract claim. That distinction is what made the statute of limitations so devastating. A contract claim can sometimes accrue at the moment of damage. A trust-based claim runs from the moment of notice. Musk had notice in 2017. He sued in 2024. The math is unforgiving.

The appeal Musk needs and the appeal he probably won’t get

The first counterpoint to the verdict’s apparent finality is Marc Toberoff’s promise to file. Per Fox Business’s coverage of the verdict and post-trial press conference, Musk’s lead attorney told reporters “this one is not over” and indicated the Ninth Circuit will receive an appeal arguing that the statute-of-limitations ruling sets a destructive precedent for charitable enforcement. Musk himself posted on X that creating precedent for “looting charities” damages American philanthropy at large. The argument has rhetorical force. It also has substantial procedural weakness. Statute-of-limitations rulings, especially those grounded in constructive-knowledge findings supported by extensive trial evidence, are notoriously difficult to overturn on appeal because they hinge on factual determinations rather than legal interpretation. The Ninth Circuit reviews factual findings under the deferential “clearly erroneous” standard. The jury and the district judge agreed unanimously. The bar for reversal is high enough to be a ceiling.

The second counterpoint is that even if the appeal fails, the case has surfaced a discovery record that will haunt OpenAI for years. Internal 2017 emails, candid testimony about Altman’s political ambitions, depositions from senior figures questioning AGI commitments — these documents are now exhibits in a public docket. Per NPR’s coverage of Altman’s testimony day, the entire AI industry’s image suffered in the trial’s exposure of how frontier labs make strategic and governance decisions behind closed doors. The verdict closes the legal exposure but not the reputational one. Regulators in Brussels, Sacramento, and Wilmington now have a primary-source archive of internal communications. The California Attorney General, who approved the recapitalization in October 2025, has independent enforcement authority that does not depend on Musk’s standing. Per SFist’s coverage of the AG’s restructure approval, that approval was conditional and revocable. A motivated state actor reading the trial transcripts could decide the conditions have changed.

The third counterpoint is the one OpenAI cannot litigate away. xAI exists. Per Al Jazeera’s verdict coverage, Musk founded the competing lab in March 2023, and OpenAI’s defense framed the entire lawsuit as a competitive sabotage operation dressed up in charitable-trust language. The jury bought the framing. The market should not. Even a losing plaintiff with a $200 billion private company at his back and an X distribution platform with 600 million monthly users can reshape the public conversation around a defendant. The trial produced narrative ammunition — Altman’s gubernatorial flirtation, the “we don’t understand why the CEO title is so important” emails, the discovery that the original founders had material doubts about Altman’s motives — that Musk will rehearse in public for years. OpenAI won the verdict. It did not win the story.

The fourth counterpoint is structural. The verdict establishes that a nonprofit can convert to a for-profit at enormous valuation increments without triggering a successful private breach-of-trust claim, provided the conversion is disclosed early enough that the limitations clock runs out before a plaintiff sues. That is, in effect, a procedural roadmap for any frontier AI lab considering a similar restructuring. Anthropic has not converted — it began as a public benefit corporation — but other labs structured as nonprofits or hybrids now have a much clearer view of how to manage the legal risk of conversion. The risk is no longer existential. It is calendrical. Per CNN Business’s coverage of the verdict, William Savitt, OpenAI’s lead trial lawyer, characterized the suit as “a hypocritical attempt to sabotage a competitor.” That framing now becomes legal canon. Every future challenger to a nonprofit-to-PBC conversion will face the precedent that delay equals defeat and competitive motive equals dismissal.

The final counterpoint is the one that should give Altman the least comfort. The verdict shields OpenAI from Musk. It does not shield OpenAI from the underlying questions the case raised. Was $44 million in early founder capital adequately compensated when the original donor was excluded from the upside? Should a charitable foundation’s mission be legally redefinable by the very executives whose ambitions the original mission was meant to constrain? Per PBS’s coverage of the verdict’s larger implications, legal scholars are already debating whether the limitations doctrine, as applied here, leaves charitable beneficiaries underprotected against governance shifts that happen slowly enough to evade the clock. That debate will reach Congress eventually. The verdict bought OpenAI time. It did not buy a permanent answer.

What this verdict really priced

The clearest signal from Monday’s two-hour jury room is that OpenAI’s IPO path is now substantively clearer than it has been since the recapitalization closed. Per the company’s own statement on its corporate structure, the OpenAI Group PBC is structured to permit a future public offering, with the nonprofit Foundation retaining controlling oversight via its equity position and warrant structure. The most credible threat to that path was always private litigation alleging that the underlying recapitalization was fraudulent or breached charitable trust. Monday’s verdict removes that threat for the largest possible private plaintiff. Smaller plaintiffs may surface, but none has the $44 million in original contributions, the founder credentials, or the legal-budget arsenal that Musk brought. The IPO window — already being modeled by underwriters for 2027 or 2028 — just became materially more bankable.

It also reinforces the strategic logic behind moves like the OpenAI Deployment Company launched on May 12, which would have been functionally impossible to defend in court had Monday’s verdict gone the other way. The verdict also reprices Microsoft’s $135 billion equity stake. Per the CNBC restructure coverage, Microsoft’s effective entry basis in OpenAI Group is roughly $50 billion against a current mark of $135 billion, an implied 170% paper gain. That gain was contingent, until Monday, on the for-profit conversion surviving private litigation. It is now substantially de-risked. Microsoft’s CFO will not announce a writeup, but the company’s risk disclosures in upcoming SEC filings should reflect a meaningful reduction in the legal-contingency category. Other shareholders — SoftBank, Thrive Capital, MGX, T. Rowe Price — see the same de-risking, though without the public-disclosure obligation. The total beneficiary class of the verdict, measured in private equity holdings of OpenAI Group, runs into the hundreds of billions of dollars.

The competitive implications are sharper than the headlines suggest. Anthropic, the only frontier lab currently positioned as a peer rival, now has a clearer view of the legal landscape it operates in but no comparable structural challenge — it was always a PBC. xAI, Musk’s lab, just lost the legal lever that might have slowed OpenAI’s commercialization. Per the trial-day-two CNBC live blog, OpenAI’s defense argued throughout that Musk’s true motivation was competitive interference. The jury agreed by implication. Whatever moral force Musk’s anti-Altman campaign carried in the press, the legal system has now spoken: he will need to compete on product and capital, not on courtroom theater. Grok, xAI’s flagship model, must close the gap with GPT-5 on its own merits. The legal handicap is gone.

Where this leads operationally is a tougher calculus. The verdict tells founders, investors, and corporate counsel that the nonprofit-to-PBC pathway is viable when timed correctly, but it also lays out the precise documentation that successful execution requires. Operators thinking about adjacent structures should treat the trial transcript as the most expensive case study available on how to — and how not to — manage a charitable-to-commercial transition. The operator checklist below distills the lessons; they are not novel as governance principles, but the verdict has given each of them a price tag.

• Document the commercial trajectory early and publicly. Per the verdict, constructive knowledge of a structural change starts the limitations clock. If your conversion is going to happen, announce its direction long before you execute it. The 2019 capped-profit subsidiary was the act that, in retrospect, gave OpenAI five years of statute-of-limitations runway.

• Treat founder departures as material events. Musk left OpenAI’s board in 2018. The departure narrative the company controlled at the time established Musk’s distance from the firm and started him on a path where every governance decision after that point was visibly not his to make. Future labs should plan founder exits with the same care as IPO disclosures.

• Negotiate the charitable trust language in the original incorporation documents with future flexibility in mind. The case turned partly on the lack of binding contractual language that would have made the charitable mission legally enforceable against the founders’ later choices. Charitable mission statements are aspirational unless drafted otherwise. Founders who want their original mission protected past their tenure must write it into governing documents that survive board reconstitution.

• Anticipate the state Attorney General’s continuing oversight. The California AG’s conditional approval of the recapitalization, per SFist’s coverage, is the live wire that remains exposed after Monday’s verdict. Operators should staff a permanent regulatory-affairs function whose only job is keeping the AG’s office informed of mission-relevant decisions. Private litigation is bounded by statute of limitations. State enforcement is not.

• Build a public-benefit narrative that survives discovery. The 2017 emails Musk’s team surfaced were embarrassing because they suggested the founding mission was always contested internally. A PBC’s public-benefit purpose statement must be a position the entire executive team can endorse under deposition five years later. Internal dissent that contradicts the external mission is now discoverable evidence.

• Price competitor-driven litigation into long-term planning. Musk’s lawsuit, regardless of merit, cost OpenAI an enormous amount of legal time, discovery preparation, and executive attention across three years. The cost of defending was real even though the cost of losing would have been catastrophic. Frontier labs should budget seven to nine figures of annual legal reserve for competitor-driven nuisance and structural litigation. Per Tech Times’s pre-deliberation analysis, Musk’s appeal — even one that fails — will continue this cost for additional years.

The two-hour deliberation was the punctuation mark, not the sentence. The sentence is that OpenAI has now successfully privatized what began as one of the most ambitious charitable experiments in technology history, with the nominal blessing of a federal court. The market will reward the certainty. Regulators will sharpen their pencils. Competitors will copy the structure. And Elon Musk will spend the next eighteen months in the Ninth Circuit telling a story whose ending is already written.

In other news

Karpathy joins Anthropic’s pre-training team. Andrej Karpathy, an OpenAI co-founder and former Tesla AI lead, announced on May 19 that he has joined Anthropic to work under pre-training team lead Nick Joseph. Per TechCrunch’s reporting on the move, Karpathy will help launch a new effort using Claude itself to accelerate pre-training research, marking his return to frontier model development after founding Eureka Labs in 2024.

Pope Leo XIV’s AI encyclical will launch with Anthropic’s co-founder. The Vatican confirmed that Pope Leo XIV’s first encyclical, titled Magnifica Humanitas, will be presented on May 25 at the Synod Hall alongside Anthropic interpretability research head Christopher Olah. Per PBS NewsHour’s coverage of the announcement, the document focuses on “the protection of the human person in the age of artificial intelligence” and represents the first papal teaching document dedicated entirely to AI.

Cursor Composer 2.5 matches Claude Opus 4.7 at a tenth of the cost. Cursor released its in-house Composer 2.5 coding agent on May 18, scoring 79.8% on SWE-Bench Multilingual — within one point of Claude Opus 4.7’s 80.5% — while pricing at $0.50 per million input tokens. Per Tech Times’s launch coverage, the model is built on Moonshot’s open-source Kimi K2.5 base with Cursor’s post-training, signaling a step-change in the cost structure of frontier coding agents.

Google and Apple confirm Gemini-powered Siri for 2026. Google Cloud CEO Thomas Kurian used the Cloud Next stage in Las Vegas to confirm that Gemini will power the next generation of Apple Foundation Models, with the new personalized Siri shipping later in 2026 inside iOS 27. Per the joint Google-Apple statement, Apple will name Google as its preferred cloud provider for the project, with the first features expected to debut at WWDC on June 8.