Second Week of Musk vs. OpenAI Case: Brockman Reveals Details of 2017 Control Negotiations
OpenAI President Greg Brockman testified on May 5, publicly disclosing the specific process behind the breakdown of negotiations over company control in 2017, as well as the surge in OpenAI's computing costs from about $30 million that year to an expected $50 billion bill by 2026.
The early team had solicited donations one by one using the Forbes billionaire list, and Musk himself acknowledged that a purely charitable model could not sustain the organization. There was no disagreement on shifting to a profit-making model. The core disagreement was over equity: Musk demanded a 51% controlling stake and the CEO position, stating that $80 billion was needed to build a city on Mars, and that the appreciation of OpenAI's equity was a crucial fundraising avenue for him.
At a meeting in August 2017, Altman proposed that the four founders split the equity equally, which Musk immediately rejected, saying, "I could start another AI company with a tweet tomorrow." When the team suggested allowing him to increase his stake at market price, Musk became furious, took a Tesla painting drawn by Ilya Sutskever, stormed out, and immediately halted his previously promised quarterly donations. Brockman recalled being worried that Musk might become violent.
Source: Public Information
ABAB AI Insight
Greg Brockman, as co-founder and president of OpenAI, has long served as a buffer between Musk and Altman. His testimony focuses on the critical meeting details from August 2017. Previously, OpenAI had disclosed several early emails from Musk supporting the shift to profitability. This additional information about the explosive growth in computing power and the financing needs for Mars shifts the conflict from ideological differences to a concrete struggle for control.
In terms of capital strategy, Musk attempted to bind OpenAI's AI achievements to the capital needs of Tesla and SpaceX by controlling OpenAI, creating a cross-company collaborative financing model. Meanwhile, the OpenAI team insisted on a decentralized equity structure to maintain independent development, ultimately leading to Musk's exit and the founding of xAI. This case essentially settles the early resource commitments against later expectations of control.
Similar to the early equity disputes between Jobs and Wozniak at Apple, as well as other founding teams that split over control issues, OpenAI is currently in a phase of legally severing ties with its early founders as it transitions from a non-profit ideal to a trillion-dollar commercial empire.
At its core, this is about capital concentration: the negotiations in 2017 exposed the zero-sum game of founders over control amid extreme capital scarcity in early AI, reconstructing the subsequent massive investments in computing power and valuation growth from a decentralized charitable model to a concentrated path led by a single controlling party. The mechanism determines who can bind the AI technology dividends to their grand plans (such as Mars colonization) through equity structure.
ABAB News · Cognitive Law
Control has never been about who is more correct, but about who can bind future cash flows to their grand narrative. The early confidence of "I could start over with a tweet" ultimately turns into litigation evidence in the face of a $50 billion computing bill from the opponent. What the founders took when they stormed out was not just a painting, but a trillion-dollar future that they could have shared.