In a federal courtroom in Oakland, California, Ilya Sutskever took the stand this week. The former OpenAI chief scientist revealed his stake in the company now stands at roughly $7 billion. He detailed a year spent documenting Sam Altman’s alleged pattern of dishonesty. And he stood by his part in the 2023 boardroom coup that briefly ousted the CEO.
His words land at a decisive moment. Elon Musk’s lawsuit against OpenAI, Altman and Microsoft nears its close. Closing arguments come Thursday. The case, which seeks more than $150 billion in damages and governance changes, turns on whether OpenAI betrayed its nonprofit roots. Sutskever’s testimony cuts to the heart of that fight.
Back in 2015, Musk, Altman, Sutskever, Greg Brockman and others launched OpenAI to steer artificial intelligence toward broad human benefit. No one expected easy success. Compute demands loomed large. So did competition from DeepMind and others. Sutskever, fresh from Google, turned down a $6 million annual offer to join. He felt ownership from the start.
By 2017 tensions boiled over funding. Musk pushed for a for-profit arm to draw serious capital. He wanted control. Brockman and Sutskever pushed back. They met at Musk’s Hillsborough estate that August. Actor Amber Heard served whiskey. Sutskever, an amateur artist, brought a painting of a Tesla. Musk had gifted both men new Model 3s beforehand.
The conversation turned sharp. Musk demanded majority ownership. The others proposed shared control. Brockman later testified he feared Musk would strike him. “I actually thought he was going to hit me,” Brockman told the court, per WIRED. Sutskever shared those worries about absolute power over AGI.
Emails and notes from the period, revealed as trial exhibits, show the friction. Sutskever suggested Musk take three board seats while the others split the rest. A proposed cap table gave Musk 51 percent, with Brockman, Sutskever and Altman each near 11 percent. Sutskever and Brockman stressed no single person should hold unchecked sway. Musk eventually walked away in early 2018. He called the path “certain failure.”
Yet Sutskever sent a warm note on New Year’s Day 2018. He called Musk “the most overwhelmingly competent person in the world” and thanked him for pressing OpenAI toward custom hardware. The gratitude was real. So was the unease.
OpenAI moved forward without Musk. It built a for-profit subsidiary under a nonprofit umbrella. Microsoft poured in billions. ChatGPT launched in late 2022 and changed everything. Valuation soared. But cracks remained.
By late 2022 Sutskever grew troubled by Altman’s management. He spent a full year gathering examples. The result: a 52-page document. It outlined a “consistent pattern of lying” and “undermining and pitting executives against one another.” Such behavior, he testified Monday, was “not conducive to any grand goal, including the creation of safe AGI.” He discussed removal with Mira Murati for months.
In November 2023 the board acted. Sutskever joined the vote to fire Altman. He helped draft the announcement. Chaos followed. Employees revolted. Microsoft weighed options. Sutskever later regretted the rushed process. The board lacked experience, he said. Legal advice fell short. He voted to reinstate Altman. “I didn’t want it to be destroyed,” he testified, according to WIRED. “I felt like I put my life into it, and I simply cared for it.”
That stance helps OpenAI’s defense. It shows internal concern over leadership, not outright rejection of the for-profit model. Sutskever told the court the shift to for-profit made sense. Building a computer as powerful as the human brain required “a lot of dollars.” Donations helped some. The difference between nonprofit and for-profit? “The difference between an ant and a cat. If there’s no funding, there is no big computer.”
His $7 billion stake, disclosed in court Monday, underscores the point. Earlier valued near $5 billion in late 2025, it reflects OpenAI’s explosive growth. Brockman previously revealed a nearly $30 billion position. These numbers fuel Musk’s argument that the nonprofit mission has been lost. They also make Sutskever one of the largest individual shareholders. Musk’s team tried to label him a hostile witness. The judge granted extra questioning latitude given his “unique position.”
Sutskever left OpenAI in 2024. He launched Safe Superintelligence Inc., focused on safe AGI development. His old superalignment team at OpenAI, which he led, disbanded months after his exit. That work, he said, represented “the most important work at OpenAI for the long term.”
Microsoft CEO Satya Nadella testified the same day. He defended the partnership as a calculated risk with huge upside. An old email showed his frustration: Microsoft “will lose 4 bil next year!!!” on the deal. Yet the investment paid off handsomely.
Trial evidence paints a messy picture of the founding. Brockman’s old journal entries, read in court, capture debates over control and employee lists that echoed Musk’s later DOGE-style efficiency demands. Brockman called Musk’s hard-driving style ill-suited to AI research. Everyone worried talent might migrate to Tesla.
Musk, for his part, has testified that OpenAI tried to “steal” a charity. He sees the for-profit pivot as betrayal. OpenAI counters that Musk never locked in special nonprofit guarantees. Negotiations dragged. Consensus formed around the need for capital.
Sutskever’s testimony adds nuance. He backed the ouster effort yet now supports OpenAI in the suit. He confirmed post-ouster talks with rival Anthropic about a possible merger. He was “not excited” by the idea. The board considered it amid the turmoil.
Recent coverage captures the stakes. Reuters reported Sutskever’s year-long evidence collection and the safety implications. Forbes highlighted the billionaire creation inside a once-nonprofit lab. Bloomberg and others noted the $7 billion figure under oath, not estimate.
These revelations come as OpenAI eyes a potential trillion-dollar IPO. The company raises fresh billions for compute. Its models power products worldwide. Yet the trial exposes foundational compromises. Sutskever’s words remind insiders that personalities, power and money shaped every turn.
He hasn’t spoken to Altman in over a year. Estranged from Brockman too. The “6-day story” of the ouster still echoes. Sutskever prepared that detailed memo on management failures. The board cited lack of candor in its initial statement.
Whatever the jury decides, Sutskever’s account stands out. It shows one of AI’s most respected minds wrestling with loyalty, safety and ambition. He cared enough to try to remove the CEO. He cared enough to regret the execution. And he cares enough now to testify under oath about billions, betrayal and the path to AGI.
The case may reshape how the world builds superintelligence. Or it may simply confirm that even the most idealistic ventures bend under pressure. Either way, Sutskever’s testimony leaves little doubt. The founding compromises were real. The personal costs run deep. And the money at stake has grown beyond anyone’s early imagination.
WebProNews is an iEntry Publication