The Most Insane Lawsuit in the Last 100 Years: Why OpenAI's Nonprofit Stake Is the Real Fight

Everyone keeps calling this an AI trial. That framing misses the point. The case that opened April 27, 2026 in the U.S. District Court for the Northern District of California is Elon Musk versus Sam Altman and OpenAI, and the AI product story is the smallest piece on the table. What is actually being fought over is roughly $130 billion of charitable assets donated under 501(c)(3) terms that now sit inside a for-profit structure. Nine jurors, no alternates. Their verdict is advisory. Judge Yvette Gonzalez Rogers makes the final call on liability and remedies.

The core question is blunt: can a public charity, after taking tax-deductible donations under an explicit promise that it will remain a nonprofit, convert into a for-profit where insiders, employees, and outside investors hold equity worth tens of billions? Answer that wrong and you rewrite the rulebook for every nonprofit hospital, university endowment, research foundation, conservation group, religious institution, and family foundation in America for the next hundred years.

How OpenAI started — and why the money mattered

In December 2015, Elon Musk, Sam Altman, Greg Brockman, Ilya Sutskever, Reid Hoffman, Peter Thiel, and Y Combinator Research announced an AI research lab. The pitch was specific. AI would be the most consequential technology of the century. They did not want it locked inside one corporation. So they built it as a 501(c)(3) — same legal status as the Red Cross or a hospital foundation. Donations tax-deductible. Research published openly. The founding charter said the lab would freely collaborate, oppose corporate concentration of power, and operate for the benefit of humanity as a whole.

That charter justified the tax deductions and brought the money in. Musk gave somewhere between about $38 million and $45 million between 2015 and 2017. His filings lean closer to $45 million; OpenAI's court disclosures put it nearer $38 million. Either way, tens of millions were structured as charitable contributions. Under U.S. law, those assets belong to the public-benefit purpose in the charter, not the founders.

Musk left the board in 2018. OpenAI says he proposed taking over leadership, the board said no, and he resigned. Musk emphasizes discomfort with the trajectory and a growing conflict with Tesla's AI work.

The conversions that landed in court

In 2019 came the first big structural shift: a capped-profit subsidiary, OpenAI LP, with Microsoft anchoring a $1 billion round. The nonprofit stayed in control. Investor returns were capped. Excess was supposed to flow back to the nonprofit. Board-approved, counsel-vetted, disclosed to the IRS. Musk had already been off the board for over a year.

November 2023: the board fires Sam Altman. Within five days he is back, the board is rebuilt, Microsoft signals it will follow him, employees threaten mass resignation, investors pile on. That week is widely read as the moment commercial pressure overwhelmed nonprofit governance at OpenAI.

Then in 2024 and 2025 came the full restructure. The for-profit arm became a Delaware public benefit corporation — OpenAI Group PBC. The capped-profit ceiling came off. The nonprofit, renamed the OpenAI Foundation, kept roughly 26% of the for-profit, valued around $130 billion at conversion. Microsoft held about 27%. Outside investors and employees held the rest. Board approved. Fairness opinion. Disclosed to the California AG, the Delaware AG, and the IRS.

That conversion is what is on trial. Pre-trial, Lanham Act counts dropped. The Friday before trial, Musk dismissed his own fraud and constructive-fraud claims to keep the jury on the mission-betrayal theory. What remains: breach of charitable trust and unjust enrichment — equity claims, which is why the jury is advisory. Damages sought sit in the $134–$150 billion range. If awarded, those dollars go to OpenAI's charitable arm, not to Musk personally. He says he is funding the case to enforce the original donation purpose.

Musk's case, steel-manned

First: he donated under explicit 501(c)(3) terms. Assets in public trust. No private distribution. Century-old rule.

Second: control on paper is not control in substance. Before conversion, the nonprofit ran the for-profit arm. After, the Foundation is a shareholder with board appointment rights next to Microsoft at ~27% and roughly half the cap table with employees and investors. That is not the same as a 501(c)(3) directing its own subsidiary.

Third — and this outruns Musk: if courts bless this, every nonprofit founder just learned a tax arbitrage play. Build as a charity. Take deductible donations. Grow value into the tens of billions. Convert. Insiders pocket equity. During jury selection Musk posted on X: "Scam Altman and Greg Stockman stole a charity. Do you want to set legal precedent in the United States that it is okay to loot a charity?" Historical conversions required AG approval, fair-market value, donor restitution where required, and hard fiduciary review. Musk says OpenAI checked boxes without satisfying the spirit.

Fourth: the 2015 charter — anti-monopoly, open collaboration, open research — justified the IRS exemption. Today's OpenAI runs closed-source models, aggressive commercial deployment, insider equity, and a partnership with the largest software company on the planet. On his view, that gap is abandonment of the representation that procured the donations.

Fifth: 1990s hospital conversions had a rulebook — AG review, fair value, proceeds to a successor charity — and insiders generally did not walk away with direct equity. Musk's framing is that OpenAI's insiders did. The board that approved the deal had just been reshuffled after the Altman firing and rehiring. The fairness opinion came from lawyers paid by OpenAI.

OpenAI's defense, steel-manned

Both conversions were board-approved, reviewed by outside counsel, and visible to the IRS plus the California and Delaware AGs. None of those AGs sued. That is not proof of lawfulness, but it is a real signal.

The PBC is not a plain for-profit. Delaware directors must balance shareholder returns against the public-benefit purpose in the charter. The Foundation's 26% stake, $130 billion at conversion, would be the largest charitable endowment in modern philanthropy — bigger on paper than Ford ($16B) and Gates ($67B) combined. OpenAI's line: the mission did not die; it got endowed at a scale no other charity has hit.

Capital reality: frontier training costs tens of billions. No 501(c)(3) can raise that. Without conversion, DeepMind, Meta, or Chinese state-backed labs win the frontier with no public-benefit mandate. On this view, conversion preserved the mission by giving the lab a structure that can compete.

OpenAI said in October 2025 that Sam does not receive direct equity in the for-profit successor. Musk's filings anticipate otherwise. The full cap table is private. OpenAI denied the equity claim on the record. And on Musk himself: he left voluntarily, his takeover proposal was rejected, then he built a competing lab with tens of billions. OpenAI will argue this lawsuit is at least partly competitive, not purely principled.

One-sentence defense: where training runs cost tens of billions and the alternative was American AI controlled by Google, Meta, Microsoft, or Beijing, the PBC plus the Foundation stake is a reasonable adaptation of nonprofit governance to frontier capital reality.

Why hospitals, universities, and every donor should care

The U.S. charitable sector moves over $1 trillion a year — more than 3% of GDP. Nonprofit hospitals. Research foundations. University endowments (Harvard 50+, Yale 40+, Stanford 35+). Religious institutions. Land trusts. Museums. Tens of thousands of family foundations. Same nonprofit law being interpreted in that courtroom.

The black-letter rule: 501(c)(3) assets are in public trust and cannot be distributed to private individuals. The closest analogues are the 1990s nonprofit hospital and Blue Cross of California fights — successor foundations, AG review locked into statute, value to charity, not to insiders' equity. If this conversion is upheld, the new practical rule becomes: enough board approval, a fairness opinion, counsel review, and regulator visibility, and a 501(c)(3) can convert with insider equity. That rule travels everywhere. If Musk wins, the rulebook stays locked. New conversions become near-impossible without explicit donor consent. Every American who ever took a charitable deduction has skin in this, because the deduction was the bargain for the public-trust commitment.

Three ways this can land

OpenAI wins outright. Conversion stands. Expect a wave of nonprofit-to-for-profit moves over 5–10 years in tech, biotech, and academic spinouts. Donors demand no-conversion-without-consent clauses. The IRS likely writes procedural floors within a few years.

Musk wins outright. Conversion unwound or heavily constrained. Equity claims contested or clawed back toward the foundation. Microsoft's position gets revisited. Nonprofit-then-convert as tax arbitrage dies. Capital flows into PBCs and B Corps from day one.

Mixed verdict. Conversion partially upheld with constraints — equity capped or clawed back, mandatory donor notice, AG findings, fair-market value paid by insiders for any equity, heavier disclosure. IRS and state AGs add process within 12–24 months. Lawyers cite this case for half a century.

The 100-year stakes

Walz v. Tax Commission (1970) set who gets nonprofit status. Bob Jones University v. United States (1983) made clear that status is conditional. The 1990s hospital and Blue Cross fights wrote the conversion rulebook. This case is bigger for two reasons. First, the equity-pocketing question: does the century-old ban on distributing charity assets to private individuals still have teeth in 2026? Second, the AI-era stakes. A hospital conversion reshaped a regional market. An AI lab conversion reshapes who controls the most consequential technology of the next century — under what structure, with what guardrails, and with what accountability to the public that supplied the deductions, exemptions, talent, and data.

The structures set in 2026 lock in for a long time. We are deciding them now, whether we are watching or not.

Check the video here.