OpenAI Co-Founders Held Undisclosed Cerebras Equity, Trial Reveals
Greg Brockman confirmed under oath on May 5 that he held personal Cerebras equity while pushing OpenAI to commit billions to the AI chip company — without ever disclosing that stake to Elon Musk. The trial admission confirms both Brockman and Sam Altman held undisclosed personal interests in Cerebras as fiduciaries of a 501(c)(3) charity that ultimately committed over $20 billion to the company.
What the Source Actually Says
The courtroom exchange is stark. Asked whether he was "actually an owner of Cerebras" during deal discussions, Brockman testified: "There was some overlap between discussions and being an investor in Cerebras. Yes." When asked to identify any email, chat, or text informing Musk of this stake, Brockman's answer was an unbroken no. When pressed on personal gain, he conceded: "I suppose so, but it wasn't something on my mind."
The financial timeline makes the conflict concrete. Brockman's Cerebras stake dates to 2017. By December 2025, OpenAI had signed a $10 billion deal with Cerebras and loaned it an additional $1 billion. By February 2026, Cerebras's valuation tripled from $8 billion to $23 billion on OpenAI commitments alone. By April 2026, OpenAI expanded its commitment to over $20 billion through 2029, and Cerebras filed for IPO at a potential $26.6 billion valuation — the cash-out event both co-founders stood to benefit from.
A parallel disclosure failure surrounds Altman: trial evidence indicates he separately invested in Cerebras, and court records show he committed $51 million of OpenAI funds to a chip startup he personally backed while telling the US Senate he had no "direct" investment in OpenAI. Gary Marcus, providing granular trial commentary, noted Altman simultaneously held indirect OpenAI equity via Y Combinator. California charitable-trust law explicitly classifies these arrangements as self-dealing.
Strategic Take
The Cerebras disclosures reframe the trial as a governance case study, not a Musk personality dispute. Organizations building on OpenAI infrastructure should track the outcome: a ruling against OpenAI leadership could constrain its procurement authority and introduce operational uncertainty for an enterprise AI stack many teams have made foundational.
