OpenAI Timeline: Elon Musk's Proposed For-Profit Structure and Departure

OpenAI has released a detailed timeline and internal correspondence asserting that Elon Musk actively sought to transition the organization from a nonprofit to a for-profit entity in 2017 to secure absolute control and majority equity. This disclosure serves as a rebuttal to Musk's legal claims, illustrating that the current for-profit elements of OpenAI were a direction Musk himself once championed, provided he could lead the organization as CEO.

The 2017 Shift Toward a For-Profit Structure

By early 2017, OpenAI researchers realized that achieving Artificial General Intelligence (AGI) would require billions of dollars in compute power, far exceeding their initial nonprofit resources. This realization led to a consensus between OpenAI leadership and Elon Musk that a for-profit structure was the necessary next step to attract the required capital.

In September 2017, Elon Musk directed his wealth manager to create a public benefit corporation called ‘Open Artificial Intelligence Technologies, Inc.’. This entity was intended to be the new structure for OpenAI. According to internal messages, Musk's terms for this transition included:

  • Majority Equity: Musk sought ownership between 50% and 60%.
  • Absolute Control: Musk proposed a board structure where he would ‘unequivocally have initial control of the company.’
  • Executive Leadership: Musk demanded the role of CEO.

Rejection of Terms and the ‘AGI Dictatorship‗

OpenAI leadership rejected Musk's proposal in September 2017, concluding that giving one individual unilateral control over AGI technology was contrary to the organization's mission. In an email dated September 20, 2017, Ilya Sutskever expressed concerns that the proposed structure would create a path to an ‘AGI dictatorship,’ noting that while Musk claimed he did not want to control the final AGI, his negotiation tactics suggested that absolute control was a priority for him.

Following this rejection, Musk ended his funding of OpenAI, stating, ‘Discussions are over.’

Post-Departure Conflict and Tesla Merger Proposal

After the breakdown of negotiations, Musk proposed that OpenAI merge into Tesla to secure a billion-dollar budget. OpenAI's team declined the merger, as they did not wish to work at Tesla. By January 2018, Musk warned that OpenAI was on a ‘path of certain failure relative to Google’ unless immediate and dramatic action was taken.

Musk resigned as co-chair of OpenAI in February 2018. Even after his resignation, he continued to provide feedback on fundraising, stating in December 2018 that the organization needed ‘billions per year immediately or forget it.’

Transition to Capped-Profit and xAI

In March 2019, OpenAI announced the creation of OpenAI LP, a capped-profit entity governed by the nonprofit. Internal communications reveal that OpenAI offered Musk equity in this capped-profit arm multiple times, but he declined.

The conflict culminated in March 2023 when Musk formed xAI, a public benefit corporation and direct competitor to OpenAI. Shortly thereafter, Musk co-signed an open letter calling for a pause in training systems more powerful than GPT-4, a move OpenAI characterizes as a targeted request for only OpenAI to stop development, as they were the only lab with a GPT-4 level system at the time.

Technical Context: The Compute Constraint

Internal emails from 2017 highlight the technical drivers behind the push for a for-profit structure. Ilya Sutskever noted that AI breakthroughs are consistently tied to models that take 7–10 days to train, meaning hardware capacity defines the surface of potential breakthroughs. To remain competitive with labs like Google Brain and FAIR, OpenAI identified the need to:

  • Increase GPU clusters from 600 to 5,000 GPUs immediately.
  • Scale headcount from 55 in July 2017 to 200 by January 2020.
  • Secure a hardware advantage to ensure experiments could be run at a scale and speed that allowed for rapid iteration.

Sources