Musk Attacks Altman's Credibility as Jury Prepares to Decide Winner

Elon Musk is asking a federal court to unwind OpenAI's 2025 restructuring, remove Sam Altman and Greg Brockman from their roles, and award as much as $134 billion in damages from OpenAI and Microsoft, with proceeds directed to OpenAI's nonprofit entity. A jury will begin deliberations Monday and deliver an advisory verdict as soon as the following week. That verdict, though non-binding, will inform US District Judge Yvonne Gonzalez Rogers, who holds final authority. If she rules for Musk, the outcome threatens to collapse OpenAI's trajectory toward an IPO at a valuation approaching $1 trillion .

Jill Horwitz, a law professor at Northwestern University who studies nonprofits, described the governance condition of OpenAI's nonprofit to MIT Technology Review in terms that should unsettle any institutional investor underwriting that IPO valuation: "We're left with this nonprofit that doesn't have any voice. It doesn't have much money, and OpenAI doesn't think it has any obligation to fund it. It barely has a staff. It's unclear how on earth the nonprofit is supposed to exercise its duties and control the entire company" .

For capital markets, the trial's third week delivered the same message as the first two, only louder. OpenAI is a $1 trillion valuation asset wrapped inside a governance structure that a Northwestern law professor says cannot actually govern it.

The $134 Billion Ask and the IPO at Risk

The damages figure Musk seeks, $134 billion from OpenAI and Microsoft combined, is not a legal formality. It is a number calibrated to the scale of Microsoft's investment relationship with OpenAI and to the capital appreciation that accrued to Altman, Brockman, and other insiders through the conversion of OpenAI's for-profit subsidiary into a public benefit corporation in 2025 .

Musk's attorney Steven Molo argued that Altman and Brockman broke their original promise to develop AI for the benefit of humanity through a nonprofit structure, and instead engineered a corporate conversion that made them "extraordinarily wealthy." OpenAI's attorney Sarah Eddy countered that no conditions were ever placed on Musk's donations and that Musk filed suit in 2024 after statutes of limitations had already expired. Eddy's core argument was that Musk's real motive is competitive: he launched his own AI company, xAI, in 2023, and a ruling that hamstrings OpenAI benefits xAI directly .

The timeline of Musk's awareness is central to the statute of limitations argument. OpenAI created a for-profit subsidiary in 2019. Musk testified that he only recognized OpenAI had abandoned its nonprofit mission in 2022, when Microsoft was preparing to invest $10 billion in OpenAI, a deal that closed in 2023. His text to Altman after reading about the $20 billion valuation read: "This is a bait and switch" .

The asymmetry matters. Whatever the verdict, Musk's xAI faces no comparable legal overhang. Its planned SpaceX-linked IPO at $1.75 trillion proceeds on its own timeline. OpenAI's path to public markets runs directly through the outcome of this case.

Altman's Credibility Problem and Its Capital Markets Consequence

The trial's third week placed Sam Altman's personal credibility under sustained assault, and the record that emerged carries disclosure implications for any prospectus OpenAI eventually files.

Molo presented testimony from four former OpenAI insiders: former chief scientist Ilya Sutskever, former CTO Mira Murati, and former board members Helen Toner and Tasha McCauley. All four testified that Altman had lied to them. The 2023 board firing of Altman, which the board subsequently reversed after investor pressure, is now part of the trial record. Altman himself described the board's reversal as "its own kind of governance failure" .

On conflicts of interest, Altman testified that he steered OpenAI toward buying power from Helion Energy, a nuclear energy company in which he personally owns roughly one-third. Last Friday, the US House oversight committee launched an investigation into those potential conflicts. Attorneys general from more than half a dozen states called on the Securities and Exchange Commission to review them as well .

For underwriters and institutional investors considering OpenAI paper, this is not peripheral color. Governance risk, CEO credibility risk, and regulatory scrutiny risk are now documented in federal court testimony and congressional record.

The Nonprofit Fiction and What It Means for AI Governance

OpenAI's restructuring argument rests on the claim that its nonprofit entity still controls the company and remains committed to safe AGI development. Eddy told the jury: "The OpenAI nonprofit is the best-resourced nonprofit in the world" .

Molo's rebuttal was structural rather than rhetorical. Seven of OpenAI's nonprofit board's eight members also sit on the for-profit board. The nonprofit hired its first employees only one month before the trial began. Its current activity is limited to grant-making, not AI research. The same people who control the for-profit entity control the nonprofit entity that is supposed to provide oversight of the for-profit entity .

Horwitz's assessment for MIT Technology Review was categorical: the nonprofit cannot exercise its statutory duties under these conditions .

This matters beyond the courtroom. A new study published in Nature on May 13, 2026 documented that state media control already shapes LLM training data and output, with models prompted in the languages of low-media-freedom countries exhibiting measurably stronger pro-government valence . The finding reinforces a broader principle that who governs AI model training, and who holds them accountable, determines what those models say and do at scale. OpenAI's governance structure, as described in trial testimony, cannot answer that question with any institutional integrity.

The Competitive Landscape Reshaped While Lawyers Argue

While the trial consumed the courtroom, the AI infrastructure race continued at full speed outside it, and the developments matter for investors calibrating exposure across the sector.

Anthropic, whose current CEO Dario Amodei appeared in trial testimony as a former OpenAI employee who received the "jackass for safety" trophy on behalf of Joshua Achiam (OpenAI's current chief futurist), announced on May 14, 2026, a restructured pricing model for agentic AI usage. Every paid tier, starting June 15, will receive a monthly "programmatic credit pool" for agentic tool use, priced from $20 per month at the lowest tier to $200 per month for the Max 20x tier. Agentic use cases will no longer draw from standard subscription chat credits .

The move is not philanthropic. Boris Cherny, head of Claude Code at Anthropic, had already cut off third-party agent frameworks from the subscription tier the prior month, citing server strain from surging OpenClaw usage. The credit pool restructuring is Anthropic's answer to a compute cost problem created by agentic adoption growing faster than subscription economics can absorb . GitHub is making the same structural shift: it announced it is moving Copilot to an AI Credits model on June 1 .

The pattern is consistent across vendors. As agentic workloads scale, the flat-rate subscription model breaks. Infrastructure margins compress. Metered pricing becomes the only defensible unit economics model. Institutional investors underwriting AI software multiples should model this transition explicitly.

Meanwhile, ElevenLabs, founded in 2021 and valued at $11 billion as of its $500 million raise in February 2026, reported $330 million in revenue with the bulk coming from enterprise customer service deployments. The company has signed voice licensing agreements with Michael Caine and Liza Minnelli, resurrected James Earl Jones for a Fortnite partnership, and added Matthew McConaughey as an equity investor. CEO Mati Staniszewski described studio conversations as active, with talent integration as the primary unlock .

ElevenLabs and Flick (a generative AI filmmaking startup that closed a $6 million seed round in May 2026, backed by True Ventures, GV, Lightspeed, and Y Combinator) represent a different capital formation dynamic than OpenAI's trillion-dollar narrative. Smaller, capital-efficient, and built around defensible content workflows rather than AGI ambition, they face no governance litigation and no $134 billion damages exposure.

Investment Positioning

The Plocamium View

The market is pricing OpenAI's IPO as if the trial is noise. It is not.

The structural argument Horwitz articulated for MIT Technology Review is the one that survives the courtroom regardless of the jury's advisory verdict. OpenAI has built a for-profit enterprise of near-trillion-dollar scale inside a nonprofit shell that, by the testimony of its own participants, cannot govern it. That is not a legal technicality. It is a material governance deficiency that belongs in any S-1 risk factor section and in any LP due diligence checklist.

Our thesis: the trial's real output is not the verdict. It is the discovery record. Congressional investigators now have sworn testimony about Altman's conflicts of interest, four named executives who testified he misled them, and a governance structure that a law professor described as incapable of exercising its own duties. The SEC has been formally asked by multiple state attorneys general to review those conflicts. That regulatory file does not close when the jury delivers its advisory verdict.

The second-order play is Anthropic and the infrastructure tier beneath it. Amodei's company is now the beneficiary of every hour of negative OpenAI testimony. Dario Amodei appears in this trial not as a defendant but as the man who received a trophy for opposing reckless AGI development. That positioning is not accidental, and it compounds every week the trial generates headlines.

The metered-compute transition at Anthropic and GitHub signals a structural repricing of AI software multiples that has not yet fully propagated into institutional models. Flat-rate subscription economics cannot survive agentic adoption curves. The companies that solve this transition cleanly, building per-credit or per-outcome pricing with predictable gross margins, will attract the capital flows that currently chase OpenAI's narrative.

On the geopolitical dimension: the Nature study published May 13 documenting state media influence on LLM training data introduces a regulatory vector that institutional investors have not yet priced into AI software exposure. If regulators in the EU or US act on the finding that LLM output varies systematically by language in ways traceable to state media control, model audit requirements and training data disclosure mandates become live compliance costs for every major AI vendor.

The Bottom Line

The Musk v. Altman trial ends its evidentiary phase with one documented fact that no verdict can erase: the entity nominally responsible for governing OpenAI's pursuit of artificial general intelligence hired its first employees one month before the trial started. Institutions underwriting a $1 trillion IPO valuation on that governance foundation carry a risk that is now in the public record. The advisory verdict arrives as early as next week. Judge Gonzalez Rogers's ruling follows. Either outcome resets the OpenAI IPO timeline. Meanwhile, xAI targets a $1.75 trillion public market debut via SpaceX as early as June, with no comparable legal overhang and with trial testimony painting its founder as the aggrieved party in a broken promise about AI for humanity. That narrative, whatever its legal merit, is now the one the jury has heard three weeks to believe.