Nadella to testify in federal case on OpenAI's restructuring
The last person to build OpenAI was Satya Nadella.
That is not a joke. It is, apparently, the legal position Microsoft will stake out in a federal courtroom in Oakland on Monday, when the company's CEO takes the witness stand in Elon Musk's lawsuit against the AI lab he helped found and the man — Sam Altman — who now runs it.
The wire called it congressional testimony. It is not. Nadella is testifying in federal court, as a witness in a case that will determine whether OpenAI broke its founding promise when it restructured as a for-profit company worth upward of $850 billion (France24 live coverage). The distinction matters: congressional hearings are theater. Courtrooms produce evidence.
And the evidence, so far, has been damning in ways that don't neatly break toward either side.
The trial that began two weeks ago has pulled back the curtain on the strangest and most consequential partnership in AI history. We have learned that Musk personally recruited Nadella in 2017, texting him that he was the only person who could "call Satya" (Reuters key takeaways). We have learned that Microsoft feared OpenAI would defect. In a January 2018 email, Kevin Scott, Microsoft's chief technology officer, warned his Microsoft counterparts that OpenAI might "storm off to Amazon in a huff and shit-talk us and Azure" — reflecting Microsoft's anxiety that its crucial AI partner might bolt for a competitor (GeekWire courthouse reporting). We have learned — crucially — that Nadella himself wrote an email in January 2018 expressing skepticism about OpenAI's research value while simultaneously noting that Musk was telling everyone OpenAI was on the verge of AGI breakthroughs (France24 live coverage).
That email is the exhibit Musk's legal team will use on Monday.
The question it poses is deceptively simple: when a skeptic writes a check anyway, are they a founder or an acquirer?
OpenAI was founded in 2015 by Musk, Altman, Greg Brockman, Ilya Sutskever, and a small group of researchers who believed artificial general intelligence was close enough to matter and fragile enough to need protecting. Microsoft was not in the room. The company that now claims a "founding partner" role invested nothing until 2019, when it began funneling $13 billion into a partnership structured specifically to give Azure exclusive access to OpenAI's models — a deal that effectively locked much of the AI industry into a single cloud provider (GeekWire).
The mythology says OpenAI built something transformative. The contracts say Microsoft built the infrastructure layer that made it impossible to use without them.
This is the tension Nadella will have to navigate on the witness stand. He cannot simply deny Microsoft's role — the emails and investments are in the record. But he also cannot claim the ideological motivation that the original nonprofit structure implied. Musk's lawyers want the jury to see a cynical pivot: Microsoft saw AGI as a revenue line, not a mission, and wrote the check anyway because the business case eventually closed.
Nadella's defenders will argue that every serious tech company was skeptical of AGI timelines in 2018 and that the subsequent investment was precisely what made the technology broadly accessible. The Azure partnership was, by this reading, not a colonization but a delivery mechanism — the unglamorous infrastructure work that turned a research lab's output into a product the world could use. Microsoft didn't co-opt OpenAI; it enabled it.
But there is a version of this story that neither side wants to tell. It is the version where both are right. Where OpenAI genuinely believed it was building something that would change the world, and Microsoft genuinely believed it was making a smart investment, and the collision of those two beliefs produced something neither fully intended: a for-profit company with a nonprofit's rhetoric, an investor with a founder's mythology, and an AI infrastructure that half the industry depends on and half the industry fears.
The jury is advisory. Judge Yvonne Gonzalez Rogers will make the final ruling, and she has indicated she will likely follow the jury's recommendation (France24). A verdict against OpenAI could jeopardize its planned IPO — the restructuring that converted it from nonprofit to for-profit Delaware Public Benefit Corporation was already a fragile compromise, and a finding that it breached its founding mission would create significant legal exposure. A verdict for OpenAI gives Altman and Microsoft legal cover for the transition and likely accelerates the public offering.
Nadella is expected to testify for roughly half a day, following testimony from Shivon Zilis, a Musk ally who sat on OpenAI's board, and preceding Sam Altman's likely appearance on Tuesday or Wednesday. The verdict, whatever it is, will not resolve the deeper question: who built the most important AI company in the world, the researchers who gave it its ideology or the investor who gave it its infrastructure?
That question doesn't have a legal answer. It has a political one, and we are living in it.
Key sources: