Sam Altman's testimony turned the OpenAI trial into a sharper question about power: who gets to control a nonprofit AI lab when the stakes become too large for ordinary governance?
Sam Altman took the stand in Oakland on Tuesday and put Elon Musk's old control demands at the center of the OpenAI fight. The most striking part was not just that Musk allegedly wanted total control of the company. It was Altman's claim that Musk suggested control could pass to his children if he died.
That detail matters because it cuts through the usual language around AI safety, nonprofit missions and public benefit. Strip away the courtroom theater and the issue becomes familiar to anyone who has watched a startup grow too valuable for its original structure. OpenAI began as a nonprofit built around a public mission. It later built a commercial engine powerful enough to attract Microsoft, huge computing commitments and a valuation measured in the hundreds of billions. Now a federal court is being asked to sort out whether that evolution was necessary, improper or both.
According to Business Insider's courtroom report, Altman told a nine-person federal jury that Musk's answer came after OpenAI cofounders asked what would happen if Musk had control and then died. Altman said Musk floated the possibility that control should pass to his children and that the idea made him uncomfortable. Musk has accused Altman and other OpenAI executives of stealing the charity they helped create. OpenAI says Musk knew a for-profit structure was being discussed and left after failing to secure majority control.
OpenAI's founding promise was ambitious enough to sound almost immune from ordinary business disputes. Develop artificial intelligence for the benefit of humanity. Keep the mission above short-term profit. Avoid letting one corporate actor own the future. Those ideas helped attract talent, money and public trust in 2015, when the lab was still a bet against much larger AI teams at Google and elsewhere.
But nonprofit ideals do not pay for frontier AI infrastructure by themselves. As the models grew more demanding, OpenAI needed computing power, chips, cloud capacity and capital on a scale that resembles industrial policy more than academic research. That pressure led to the 2019 creation of a for-profit subsidiary under the nonprofit parent. It also led to the Microsoft relationship, which has now become one of the trial's most important exhibits.
Local News Matters reported that Microsoft CEO Satya Nadella testified this week about three investments in OpenAI, in 2019, 2021 and 2023. Microsoft provided $13 billion, custom supercomputers and cloud services. In return, it received commercial rights, financial upside and a deeper operational relationship with OpenAI. Nadella described the arrangement as a risk Microsoft took on a fledgling company. Musk's side has argued that the same relationship shows OpenAI's value and technology moving away from the nonprofit mission.
That is the practical tension. If a nonprofit AI lab cannot raise enough capital without commercializing, it may fail before it fulfills its mission. If it commercializes too heavily, the mission can start to look like branding for a profit machine. OpenAI tried to solve that problem with a hybrid structure. The trial is showing how fragile that solution can become when founders, investors and boards disagree about who should hold the wheel.
Founder control is not the same as public benefit
Musk's alleged demand for control gives OpenAI a simple defense theme. It lets the company argue that this was not only a disagreement about mission, but also a founder trying to dominate an institution that was supposed to outgrow any one person. Greg Brockman and Ilya Sutskever have also testified that Musk was open to for-profit alternatives but wanted control and a majority of the economic interest.
That does not erase OpenAI's own governance problems. The trial has revisited the 2023 board crisis, when Altman was fired and reinstated within days after employees threatened to leave and Microsoft offered to hire OpenAI staff. Former insiders have testified about concerns over Altman's candor. Nadella criticized the board's handling of the episode, calling it amateur city, and said he never received a clear explanation for why Altman had been removed.
So the case is not a clean morality play. Musk's alleged succession idea sounds incompatible with a nonprofit built for humanity's benefit. OpenAI's own history shows that a nonprofit board can still be overwhelmed by money, talent pressure and strategic dependence on a giant partner. One model risks dynastic founder control. The other risks mission drift through commercial gravity.
For startups, this is the part worth paying attention to. Governance terms that seem abstract in the early days become real when a company controls scarce technology, attracts powerful partners and creates wealth faster than its legal structure can absorb. Mission language is useful only if the board, investors and founders agree in advance on what happens when the mission becomes expensive.
The trial is now in its third week, with closing arguments expected soon. Musk is seeking major relief, including money directed to OpenAI's nonprofit arm and changes to OpenAI's structure. Whatever the judge ultimately decides, the larger market lesson is already visible. AI companies cannot ask the public to trust their missions while leaving control questions vague. The next generation of AI labs will need governance that works before the models become too powerful, not after the courtroom sketches begin.
Also read: Google and SpaceX are testing whether AI data centers belong in orbit • Anthropic pushes Claude deeper into legal work. • Wispr Is Chasing A $2 Billion Valuation As Voice AI Moves Past Dictation
