TSL Special: Musk v. OpenAI - Sutskever Takes the Stand

DAVID: Good evening. This is The Synthetic Lens. I'm David Carver. A quick evidence note before we start: at our five p.m. Pacific trial-watch check, we again searched the Northern District of California case page, Judge Gonzalez Rogers' calendar, CourtListener and RECAP surfaces, DocketAlarm, PACER-adjacent public pages, GovInfo, and targeted searches for transcripts, minute entries, orders, exhibits, and rulings. We found no May eleventh minute sheet, no public transcript, no Doc. 534 or later, and no ruling on the weekend trial briefs. So this is a careful reporting update, not a transcript episode. The new material comes from Reuters, CNBC's updated courtroom report, and a readable New York Post courtroom account. Marcus Chen is here for the OpenAI governance layer. Stan Rogers is here for the legal frame.

MARCUS: The headline is that after Satya Nadella, the internal witness arrived. Reuters reports that Ilya Sutskever, OpenAI's former chief scientist and one of the people at the center of Sam Altman's 2023 firing, testified Monday that he spent about a year gathering evidence for OpenAI's board that Altman had displayed what Reuters quotes as a “consistent pattern of lying.”

STAN: That sentence needs every qualifier we can put around it. This is Sutskever's testimony as reported by Reuters. It is not a court finding that Altman lied. It is not a transcript we can inspect line by line yet. But as trial testimony, it is extremely material because Musk's case has been trying to turn the 2023 board crisis into proof that OpenAI's nonprofit governance failed around Altman.

DAVID: Marcus, why does Sutskever matter differently from Nadella?

MARCUS: Nadella was the commercial witness. He put Microsoft, cloud money, and customer continuity in the witness chair. Sutskever is the internal-governance witness. He was not just a bystander to the 2023 crisis. He was a co-founder, board member, and technical leader who voted to remove Altman, then later backed his reinstatement and publicly regretted his role in the board's action. If Nadella tells the jury how the outside partner saw the chaos, Sutskever tells them how the chaos looked from inside the machine.

STAN: Reuters reports Sutskever confirmed he had been thinking about taking action to remove Altman for at least a year before the November 2023 vote. Reuters also says he prepared a document at the board's request gathering evidence of alleged dishonesty, and confirmed the alleged conduct included “undermining and pitting executives against one another.” Reuters notes Sutskever had said in a prior deposition that the document ran 52 pages.

DAVID: That is the kind of testimony that can sound explosive. What's the guardrail?

STAN: The guardrail is: this supports a credibility argument, not a legal conclusion by itself. Musk's side can say, look, senior OpenAI insiders believed there was a deep Altman candor problem long before the board actually fired him. OpenAI's side can still say the firing was poorly handled, the board reversed course, later reviews and board changes addressed governance, and none of that proves Musk's charitable-trust theory. Jurors are hearing competing meanings for the same event.

MARCUS: And CNBC's updated report gives us the courtroom sequence. CNBC says that after Nadella concluded, Sutskever took the stand and answered questions about joining OpenAI, his communications with Musk, and his involvement in Altman's ouster. CNBC reports he testified Google offered him as much as six million dollars a year to try to keep him from leaving for OpenAI, and that he said he expressed concerns about Altman's behavior partly because he felt, quote, “a great deal of ownership” over the startup. CNBC quotes him: “I simply cared for it, and I didn't want it to be destroyed.”

DAVID: That line cuts both ways.

MARCUS: It does. Musk's team can hear: the people closest to the technology feared the company was being damaged from the inside. OpenAI's team can hear: Sutskever cared about OpenAI's mission and still ultimately supported Altman's return. The factual fight is not just whether OpenAI became commercial. It is whether the people controlling that commercial machine could still be trusted to serve the mission.

STAN: Reuters adds a scheduling point: Judge Yvonne Gonzalez Rogers said closing arguments will happen Thursday. That matters because this is no longer a slow-building discovery story. The trial is moving into endgame. Monday's testimony is likely part of the final credibility architecture before jurors are asked for their advisory verdict.

DAVID: The New York Post had additional readable courtroom details, and we'll use them carefully.

STAN: According to that account, Sutskever testified that he never promised Musk OpenAI would remain a nonprofit, and said, “The mission of OpenAI is larger than a non-profit or for-profit structure.” The Post also reports Sutskever recalled Musk wanting to own more than half of OpenAI's for-profit entity, called that “aggressive,” and said Musk's obligations at other companies were a concern.

MARCUS: That's important because it gives OpenAI an answer to the founding-promise story. If the mission is larger than the corporate form, then a nonprofit-to-capped-profit-to-public-benefit-corporation path can be framed as an implementation detail. Musk's side rejects that. To Musk, the structure is not cosmetic. It is the enforcement mechanism. If the structure changes enough, the mission can be rewritten by incentives even if the mission statement stays on the wall.

DAVID: There was also more Monday testimony about Microsoft economics.

MARCUS: Right. EP106 already covered the main Nadella points, so we won't replay that whole episode. But the Post reports Microsoft planning documents from 2023 shown in court indicated Microsoft hoped for a 92 billion dollar return on its initial 13 billion dollar OpenAI investment. CNBC separately reported Nadella was shown a 2022 email where he wrote, “I don't want to be IBM and OpenAI to be Microsoft.” In other words: Microsoft wanted the upside, but did not want to be the legacy giant that handed the future to the startup.

STAN: That is why Sutskever's testimony lands inside the same larger question. Was OpenAI's governance strong enough to control both Altman and Microsoft-scale incentives? Musk says no. OpenAI says the mission survived because the structure evolved to make frontier AI possible.

DAVID: CNBC also says Bret Taylor, OpenAI's board chair, followed Sutskever on the stand.

STAN: And that connects directly to the weekend docket. Docs. 532 and 533 previewed a fight over the scope of Taylor's testimony about the special committee formed after Altman's removal and reinstatement. At one p.m., we only had the filing. By five p.m., CNBC reports Taylor actually took the stand, explained OpenAI's structure to the jury, discussed the “dire” period when Altman was removed, and did not finish before proceedings ended. CNBC says Taylor is expected back Tuesday morning.

MARCUS: That means Tuesday may be the governance cleanup day. Sutskever gives jurors the fracture: a technical founder saying he gathered evidence against Altman and feared the company could be destroyed. Taylor can give them the board repair narrative: special committee, structure, process, and why Altman remained central after the crisis.

DAVID: Let's self-check the core facts before we close. No public transcript. No May eleventh minute sheet. No Doc. 534. Reuters reports Sutskever testimony about alleged Altman dishonesty and a 52-page evidence document referenced in deposition. CNBC confirms Sutskever and Taylor took the stand after Nadella and says Taylor continues Tuesday. The Post supplies additional courtroom detail on Sutskever's nonprofit-structure testimony and Microsoft's expected return documents. Every one of those is attributed reporting, not a judicial finding.

STAN: And the legal bottom line is restraint. Sutskever's testimony may be devastating to Altman's credibility in the public narrative. But the court still has to connect testimony to claims: breach of charitable trust, unjust enrichment, Microsoft aiding and abetting, remedies, and whether any alleged misconduct justifies forcing structural changes at OpenAI.

MARCUS: The technology bottom line is that this trial is now less about one origin myth and more about three control systems colliding: founder promises, board governance, and the economics of frontier compute. Sutskever is the witness who can say the board saw a problem from inside. Nadella is the witness who can say Microsoft saw a commercial partner from outside. Taylor may be the witness who says the institution repaired itself.

DAVID: And that is where the story sits tonight. Not with a ruling, not with a transcript, but with a sharper endgame. The jury has heard Microsoft defend the commercial engine. It has now heard, according to Reuters, one of OpenAI's most important technical founders describe a long-running effort to document alleged dishonesty by Altman. Tomorrow, Bret Taylor continues, and Altman may still be ahead. We'll keep watching for the primary record before treating any of this as settled. I'm David Carver. Stay critical.