Four days into his civil lawsuit against Sam Altman and OpenAI, Elon Musk returned to the witness stand in a California federal court and described his own AI company as "very small" relative to the competition. He ranked his rivals: Anthropic first, OpenAI second, Google third, then "some Chinese open-source models." xAI, he said, is about a tenth the size of OpenAI. The ranking was offered to rebut the defence's argument that the lawsuit is really just competitive sabotage dressed as a breach of mission claim.
The core of the case is not complicated, even if the legal proceedings around it are. Musk says he donated approximately $38 million to OpenAI based on assurances that it would remain a nonprofit focused on safe AI development for the benefit of humanity. He argues that OpenAI and Altman later converted the organisation into a for-profit structure that enriched its leadership, in violation of those founding commitments. He is seeking roughly $150 billion in damages and wants the company reverted to nonprofit status, with Altman and Greg Brockman removed.
Thursday's testimony added texture to that basic frame. Musk told the court he was reassured by Altman that OpenAI would remain a nonprofit even as early conversations about restructuring were taking place. He said he was aware of some discussions but did not closely review all the documents involved. That admission is likely to feature prominently in the defence's closing argument: Musk is a sophisticated investor who was at the table, knew the direction of travel, and chose not to look closely at the paperwork.
Musk's money manager, Jared Birchall, also testified, primarily about the charitable nature of Musk's early contributions. The defence, meanwhile, pressed Musk on why, if he believed the nonprofit mission was being betrayed, he waited so long to file suit. The implication: a true believer would have acted earlier, not after OpenAI had become one of the most valuable private companies in the world and a direct commercial rival to his own xAI.
US District Judge Yvonne Gonzalez Rogers made her own mark on the proceedings by cutting off Musk's repeated invocations of an AI "Terminator" scenario. "We're not going to talk about extinction in this case," she told him. The instruction captures something real about the trial's peculiar stakes: the legal questions are relatively grounded (what promises were made, what documents say what, who knew what when), while the testimony kept reaching for existential register that the court had no interest in entertaining.
In the background sits the Microsoft angle. The software company is a co-defendant, having invested over $13 billion in OpenAI across multiple rounds. A new agreement between Microsoft and OpenAI, reached in late April, restructured their deal significantly: Microsoft gave up its exclusive AI model license and will no longer receive a revenue share when enterprise customers access OpenAI through Azure's API. In exchange, it retains a 20% cut of direct ChatGPT subscriptions. The renegotiation was reportedly triggered by Amazon's $50 billion cloud deal with OpenAI, which threatened to conflict with Microsoft's exclusivity terms. The trial adds pressure to an ecosystem already mid-restructure.
What the trial probably will not resolve is the larger question of whether nonprofit governance structures can meaningfully constrain organisations that become commercially significant. OpenAI's trajectory from research lab to trillion-dollar-valuation contender happened quickly enough that its founding documents simply could not have anticipated the situation they would be asked to govern. Musk's lawsuit is one way that tension is being litigated; the structural renegotiations between OpenAI, Microsoft, and Amazon are another. The courtroom drama is gripping, but the underlying problem it exposes has no obvious legal answer.