How Elon Musk left OpenAI, according to Greg Brockman

In late August 2017, OpenAI’s co-founders gathered to discuss creating a for-profit arm to fund compute-heavy AGI research. Elon Musk, who had just given each co-founder a Tesla Model 3, demanded “unequivocal” control of the new entity. CTO Greg Brockman saw the cars as buttering-up. Head of research Ilya Sutskever even commissioned a Tesla painting as a friendly gesture. But when the others proposed equal shares, Musk reacted angrily. According to Brockman’s journal, Musk sat silently for several minutes, then said “I decline,” grabbed the painting, and stormed around the table. Brockman said, “I thought he was going to hit me.” Musk then asked, “When will you be departing OpenAI?” None of the co-founders left. Musk stopped his regular donations and left the board within six months, though he paid for shared office space with Neuralink until 2020.

The inciting incident for the for-profit pivot was OpenAI’s DOTA II victory, which convinced the team that compute scale was critical. Over 20 variations of the for-profit structure were discussed, including linking to Tesla’s AI work. But Musk’s demand for sole control ultimately fractured the partnership. Brockman’s deeply personal journal, now public in the ongoing trial, offers a rare window into the negotiation. Two days before the trial, Musk texted Brockman: “By the end of this week, you and Sam will be the most hated men in America. If you insist, so it will be.” While the jury won’t see that text, Musk’s lawyers are portraying Altman and Brockman as having “stole a charity,” while OpenAI’s team argues Musk had the same for-profit plans himself.