Nadella fears Microsoft will be the ‘next IBM’ as OpenAI’s $92B return projections revealed at trial

Satya Nadella told a judge on Monday that he feared Microsoft would be “The next IBM” when OpenAI becomes the next Microsoft. The admission, taken from an internal email of April 2022 presented by Elon Musk’s lead lawyer, reveals the strategic concerns that have driven the largest corporate investment in the history of artificial intelligence. Microsoft did not invest 13 billion dollars in OpenAI because it believes that the non-profit goal is to develop a safe AI company. It is not important if it is not.

A January 2023 memo from Microsoft president Brad Smith to the company’s board, also presented before the judge, outlined a $92 billion return on that combined investment, with an annual escalator of 20 percent starting in 2025. The document repositions the Microsoft-OpenAI relationship from a technical collaboration into what could be the largest financial hedge to survive in the history of the world’s most valuable company. The era of AI alone.

Email

The IBM analogy is not a common one. In the 1980s, IBM developed a personal computer and released an operating system to a small software company in Redmond, Washington. That decision made Microsoft and IBM undone. Nadella was telling his team that the same thing was happening with AI. OpenAI is building an inference engine. Microsoft is building cloud infrastructure. If OpenAI becomes a platform and Microsoft becomes a commodity, the company that defined business software for forty years will fade into insignificance like the company that defined business hardware for three years.

Musk’s lawyers presented the email to suggest that Microsoft’s investment was commercially motivated from the start, undermining OpenAI’s non-profit origins. Nadella’s response was to defend the relationship as mutually beneficial. But the email speaks for itself. Microsoft’s CEO wasn’t writing about improving AI security. He was writing about survival.

Refund

Brad Smith’s prediction of 92 billion dollars sat on the desks of the Microsoft board one month before the company publicly announced that it invested 10 billion in OpenAI. The memo includes a 20 percent annual escalator starting in 2025, which means expected returns will compound as OpenAI models become more commercially valuable. At that time, ChatGPT had been public for less than two months.

Accounting was straightforward. Microsoft was the exclusive cloud provider for OpenAI models and has exclusive rights to sell them for resale through Azure. Every dollar of OpenAI’s revenue flows through Microsoft’s infrastructure. The 13 billion dollars was not a non-profit organization donation. It was a down payment on the most important technology distribution monopoly of the decade.

OpenAI is now valued at $852 billion. Microsoft owns 27 percent of the for-profit entity that appeared in the October 2025 reform. The non-profit organization that was supposed to manage the technology keeps 26 percent. The alignment between machines and money that the founders of OpenAI had promised has been replaced by a cap table.

Blind spots

Under cross-examination, Nadella admitted that he did not know of any full-time employees of the non-profit organization OpenAI before March 2026. He could not identify any grants, research, or open technology that the non-profit organization had produced. He was not informed in advance that the board planned to fire Sam Altman in November 2023. He was not given an explanation as to why Altman was removed.

The admission paints a picture of a partnership where the investor knows everything about commercial operations and knows nothing about non-profit governance. Musk’s legal team wants a judge to conclude that the nonprofit was a shell. Nadella’s testimony does not contradict that frame. It strengthens it from the point of view of a company that has had many gains on the commercial side.

Witnesses

The trial lasted three weeks accumulating evidence dispelling the alleged motives of all the participants. Greg Brockman, founder and president of OpenAI, disputed Musk’s account of the startup’s early days and testified that Musk had OpenAI employees do secret work on Tesla’s self-driving technology. Brockman’s journals, presented as evidence, contain material that the non-profit campaign called “false”, undermining Musk’s claims that the machines are sacred and OpenAI’s claim that they are preserved.

Former board members Helen Toner and Natasha McCauley testified that Altman was dishonest, withheld information from the board, and sometimes lied. McCauley told the judge that the board “buckets of worry” about Altman’s leadership, including an incident in which Altman lied that OpenAI’s legal department canceled the launch of the GPT-4 Turbo in India without a safety board review. The women who fired Altman in November 2023 told the judge why, and their reasons had nothing to do with Musk’s case.

Admissions

Musk made the decision in the first week of the trial and told the judge that OpenAI leaders tricked him into registering the company. He repeated the speech that became the speech of the case: “You can’t just steal from charity.” He stated that he was not opposed to a small for-profit arm supporting a non-profit organization but lost confidence in Altman when he heard about Microsoft’s $10 billion investment, writing to Altman in late 2022: “What is being done? This is a nickname and a change.“

Then came the question about distillation. Asked if xAI uses OpenAI models to train Grok, Musk said it’s standard industry practice. Asked if that means yes, he replied: “Half.” The admission that his AI company was copying technology he claimed was stolen from a charity caused a stir in court. Musk told the judge that the case would set a precedent for “robbing every charity in America” ​​while admitting he was using the organization’s proceeds to build a rival.

Shivon Zilis, a former OpenAI board member and the mother of Musk’s four children, testified that Musk tried to hire Altman to lead a new AI lab at Tesla. He offered Altman a seat on Tesla’s board. He asked Andrej Karpathy to send a list of top OpenAI researchers to poach. The man who sued for breach of the covenant to help the poor, according to the testimony of his witnesses, was actively trying to deprive him of leadership and talent.

Self defense

Altman assumed the position on Monday. He pointed out that Musk’s departure from the OpenAI board in 2018 was “.to encourage morale” to other employees because Musk had discredited key researchers by posting their achievements.” Altman told the judge that Musk left because he lost hope in the project and wanted long-term control that other founders would not give him.

In a heated exchange, Musk’s attorney confronted Altman about a message Musk sent on February 18, 2023: “Thank you so much for all you have done to help. I don’t think OpenAI would have happened without you.” The implication was that Altman privately accepted Musk’s offer while publicly downplaying it. The memo was sent three months after Musk heard about Microsoft’s investment and seven months before the board fired Altman.

The trial began with $150 billion at stake over whether OpenAI’s conversion from a non-profit to a for-profit organization was a breach of trust. Musk wants the court to reverse the conversion, exonerate Altman and Brockman, and direct damages to the nonprofit. OpenAI says Musk is suing because he wanted control of the world’s most valuable AI company and didn’t get it.

The fence

While the trial is being played out in Oakland, Microsoft is quietly proving that Nadella has learned IBM’s lesson. Microsoft has given up its exclusive license to OpenAI technology, keeping only a non-exclusive agreement until 2032. It did so voluntarily, which only makes sense if Microsoft no longer needs to choose because it has other options.

It happens. Microsoft has unveiled three in-house AI models that directly challenge the partner it spent $13 billion to cultivate. The company that feared IBM responded by doing what IBM never did: build its own operating system before a partner locked it out. Nadella’s April 2022 fear that Microsoft would become dependent on OpenAI appears to have been the start of an entire corporate strategy designed to ensure that it won’t.

The case is expected to continue until May 21 before Judge Yvonne Gonzalez Rogers. A judge will decide whether OpenAI’s leaders breached the charity’s trust and whether Musk owes the money back. But Nadella’s testimony already answers a different question. The strongest corporate backer of non-profit AI machines invested because he feared his company would die without it. Projecting a return of 92 billion dollars was not a co-creation. It was the point. The non-profit book that Musk claims was stolen may not have contained what any of the parties believed it did.