Key Revelations Emerge from Musk-Altman Trial Over OpenAI's For-Profit Conversion

A nine-person federal jury in Oakland began deliberating on May 19 whether OpenAI's leaders broke promises to Elon Musk when they converted the AI nonprofit he helped found into one of the most valuable companies on Earth. The three-week trial produced damaging testimony from both sides, personal diary entries, internal emails, and a central question: can a $44 million donor unwind an $852 billion corporation? [1][2]

The Money at the Center

Musk donated approximately $44 million to OpenAI between 2016 and 2020 — far short of the $1 billion he initially pledged [3]. During testimony, Musk claimed his total contributions, including "intangibles" like recruiting talent and lending his name, exceeded $100 million [4]. He received no equity in the original nonprofit structure, nor was any promised — OpenAI was incorporated as a 501(c)(3) where no individual could hold ownership stakes.

The disparity between Musk's contribution and the current valuation of stakes held by others became a central trial theme. Greg Brockman, OpenAI's president, holds equity now worth approximately $30 billion despite never contributing the $100,000 he initially pledged to the nonprofit [5]. Sam Altman, who testified he took no salary or equity during the nonprofit years, now leads an entity valued at $852 billion [6].

Source: CNBC, BeBeez

Data as of Apr 7, 2026CSV

By comparison, Mozilla's spinoff of the Mozilla Corporation from the Mozilla Foundation in 2005 kept all commercial revenue flowing back to the nonprofit parent. The Wikimedia Foundation has never spun off a for-profit entity at all. No precedent exists for a nonprofit-to-for-profit conversion at anything close to OpenAI's scale.

What the Internal Documents Revealed

The trial's most damaging exhibit came from Brockman's personal journal. In a November 2017 entry, written after a meeting with co-founder Ilya Sutskever, Brockman wrote: "Cannot say we are committed to the non-profit..." [1]. Musk's attorneys argued this proved the founders always intended to go commercial. OpenAI's defense countered that the entry reflected an ongoing strategic discussion, not a fixed plan.

Former CTO Mira Murati testified via video deposition that Altman had a pattern of "saying one thing to one person and completely the opposite to another person" [7]. Former board member Helen Toner, who helped engineer Altman's brief November 2023 ouster, told the court there was a "pattern of behavior related to his honesty and candor" [7].

Altman fired back during his own testimony, painting Musk as someone who "wanted control" over AGI development rather than a principled governance advocate. He testified he "never promised Musk to keep the company a nonprofit" [8]. Microsoft CEO Satya Nadella also took the stand, with internal Microsoft emails revealing he compared the OpenAI partnership to IBM's early relationship with Microsoft — writing that he didn't want Microsoft to "become the next IBM" while OpenAI became the next Microsoft [9].

The Competitive Harassment Question

OpenAI's defense rests partly on the argument that Musk's lawsuit is competitive sabotage rather than principled objection. The timeline supports this reading to a degree: Musk founded xAI in July 2023, filed the first version of this lawsuit in early 2024, and escalated claims as xAI sought to compete directly with OpenAI's ChatGPT through its Grok chatbot [10].

During Week 1 testimony, Musk admitted under cross-examination that xAI uses OpenAI's models to train its own through a process called distillation [1]. OpenAI filed a separate request asking the California and Delaware attorneys general to investigate what it called Musk's "anti-competitive behavior" [11].

OpenAI has publicly characterized the suit as driven by "ego, jealousy and harassment" [12]. However, the competitive harassment framing has limits: Musk's concerns about the for-profit conversion predate xAI's founding, and the core legal question — whether charitable assets were improperly diverted — exists independently of who brings it.

Musk's attorneys pledged during trial that any damages awarded would be returned to OpenAI's foundation rather than enriching Musk personally [2]. Whether jurors find this credible given his competitive position remains to be seen.

The Stakes: $134 Billion and Leadership Removal

Musk's legal team asked the jury to recommend up to $134 billion in damages from OpenAI and Microsoft, characterizing the for-profit conversion's proceeds as "wrongful gains" [2]. Beyond monetary damages, Musk seeks the removal of both Altman and Brockman from their positions and a full reversal of OpenAI's 2025 restructuring [13].

The jury's verdict is advisory — Judge Yvonne Gonzalez Rogers will make the final determination on liability and remedies [14]. A second phase of the trial addressing specific remedies runs concurrently with deliberations.

Source: CNBC, OpenAI

Data as of Apr 7, 2026CSV

The Governance Structure Under Scrutiny

OpenAI completed its conversion to a Public Benefit Corporation (PBC) in October 2025, after negotiations with California Attorney General Rob Bonta [15]. Under the new structure, the renamed OpenAI Foundation holds approximately 26% of the for-profit entity — a stake worth roughly $130 billion at the time, making it one of the world's largest charitable endowments on paper [15].

The Foundation retains board appointment power and, through a special committee, can intervene on AI safety concerns. Microsoft holds 27% and employees plus other investors hold the remaining 47% [6].

Critics have identified structural weaknesses. A CalMatters investigation found "numerous ways the for-profit company could end up calling the shots" despite the nonprofit's nominal control, with "important, unanswered questions about the safeguards" [16]. The Foundation's board appointment power means little if the for-profit's operational decisions aren't subject to meaningful override.

How Does This Compare?

Anthropic, founded by former OpenAI employees Dario and Daniela Amodei, operates as a PBC but with different investor constraints. Amazon and Google each hold less than 15% of Anthropic and own no voting shares, meaning they cannot elect board members [17]. This contrasts sharply with Microsoft's 27% voting stake in OpenAI.

Google's DeepMind operates within Alphabet's corporate governance, with no independent nonprofit oversight mechanism. OpenAI's structure theoretically provides more external accountability than DeepMind's but less investor independence than Anthropic's.

The original "capped-profit" model that OpenAI introduced in 2019 limited investor returns to 100x their investment [18]. That cap was eliminated in the 2025 PBC conversion — a change worth tens of billions to early investors like Microsoft.

California's Role and the AG's Intervention

California Attorney General Rob Bonta investigated OpenAI's conversion starting in January 2025 [19]. A coalition of approximately 60 organizations, led by the San Francisco Foundation and LatinoProsperity, petitioned Bonta to block the conversion, arguing it could "jeopardize nonprofit assets — with value estimates at as much as $157 billion" [20].

Bonta ultimately approved the restructuring in October 2025, signing off on the arrangement where the Foundation retains its 26% stake [15]. Delaware's attorney general also approved the conversion, as OpenAI's for-profit arm was incorporated there.

The legal framework for challenging such conversions is thin. Under California law, the attorney general has standing to enforce charitable trust obligations, but once Bonta approved the restructuring, the primary remaining avenue for challenge became private litigation — exactly what Musk's suit represents.

Legal Precedent for Donor Claims

The enforceability of charitable pledges varies dramatically by state. Under general contract law principles, a pledge is enforceable if there is an agreement between donor and charity with consideration given in exchange [21]. Some states enforce pledges even without consideration as a matter of public policy.

However, Musk's claim is unusual. He is not arguing that OpenAI failed to spend his money as promised — he is arguing that the entire organizational structure shifted away from its charitable purpose. The closest precedent may be the Robertson family's suit against Princeton University, where donors alleged the university spent endowment funds inconsistently with donor intent. That case settled for nearly $100 million [21].

Legal doctrines of "deviation" and "cy pres" permit courts to change donor restrictions without donor approval when original purposes become impractical [21]. OpenAI could argue that remaining a pure nonprofit became impractical given the capital requirements of frontier AI development — though this argument sits uncomfortably next to the company's $852 billion valuation.

4,000 Employees and $1.5 Million Average Equity

The trial's outcome has direct financial consequences for OpenAI's approximately 4,000 employees. As of early 2026, OpenAI's average stock-based compensation per employee reached $1.5 million — the highest of any tech startup in history [22]. Equity is granted as Profit Participation Units (PPUs), which vest over four years with no cliff period.

Total employee stock compensation consumes roughly 46% of OpenAI's annual revenue, compared to 15% at Google and 6% at Meta [22]. OpenAI told investors that stock-based compensation expenses are projected to increase by an additional $3 billion by 2030 [22].

If the conversion were reversed or the restructuring unwound, the value of these PPUs would become uncertain. Early employees with large grants stand to lose the most in absolute terms, while the structural question — whether PPUs issued by a for-profit subsidiary of a nonprofit retain value if that subsidiary's legal basis is voided — has no clear answer.

Who Captures the Value?

The valuation gap between OpenAI-as-nonprofit and OpenAI-as-for-profit represents the core economic stakes. As a nonprofit, OpenAI's assets would remain charitable — unusable for private enrichment. As a for-profit valued at $852 billion, that value is distributed among identifiable parties:

• OpenAI Foundation: ~26% ($221 billion at current valuation)
• Microsoft: ~27% ($230 billion)
• SoftBank: significant minority (invested $162 billion across two rounds) [6][23]
• Employees and other investors: ~47% combined
• Sam Altman: holds approximately 3.8% of OpenAI equity [24]
• Greg Brockman: stake valued at approximately $30 billion [5]

The Foundation's $221 billion stake is designated for charitable purposes, but the remaining $631 billion in value has been privatized — a transfer from charitable to private hands that Musk's attorneys characterized as the largest misappropriation of nonprofit assets in history [2].

What Comes Next

The jury's advisory verdict could come as early as the week of May 19. Judge Gonzalez Rogers is not bound by their recommendation but would face political and practical pressure to follow it. Even a partial finding for Musk — say, on the breach of fiduciary duty claims but not the full damages — could force renegotiation of the conversion terms.

OpenAI's planned IPO, potentially the largest in history at a valuation near $1 trillion, hangs in the balance [22]. Microsoft's $13 billion cumulative investment and SoftBank's $162 billion would face uncertainty under any ruling that questions the conversion's legitimacy [14].

Whatever the jury decides, the trial has already produced a public record that regulators, future litigants, and Congress can use. The question of who controls — and profits from — the development of artificial general intelligence will not be resolved in a single verdict.