Microsoft CTO's Internal Email Details OpenAI Leadership Crisis
Microsoft Chief Technology Officer Kevin Scott sent a detailed email to CEO Satya Nadella and President Brad Smith, along with other top executives, explaining the complex series of events that culminated in the dramatic firing of OpenAI CEO Sam Altman in November 2023. The email, dated November 19, 2023, reveals that Altman's sudden removal was not the result of a single ideological disagreement but rather stemmed from multiple escalating tensions within the organization.
The Two Primary Friction Points That Led to the Crisis
The email, which was later posted by the X account Internal Tech Emails, indicates that OpenAI co-founder and board member Ilya Sutskever had been growing increasingly at odds with Altman over two critical issues. According to Scott's analysis, these conflicts created an unsustainable environment that ultimately led to the board's decision to terminate Altman.
First Conflict: The Battle Over Computing Resources
The initial point of contention centered on the allocation of GPU computing power within OpenAI. As ChatGPT's popularity exploded, creating unprecedented demand for computational resources, internal competition for GPUs became what Scott described as a "zero-sum game." Altman's strategic decision to prioritize GPU allocation for consumer products and the API division angered core research teams who felt their fundamental work was being "starved" of the resources necessary for developing next-generation models.
Scott explained in his email that this tension between research and applied divisions is common in technology companies, but at OpenAI it became particularly acute. The research teams, responsible for training new AI models, viewed their computational needs as "literally insatiable" and perceived the success of the applied division as directly causing their resource constraints. According to Scott, Altman recognized this global optimization problem and had been pushing aggressively to increase overall computing capacity to make the situation non-zero-sum.
Second Conflict: The Personal Promotion That Fractured Relationships
The second conflict was deeply personal to Sutskever and involved the promotion of researcher Jakub Pachocki. According to Scott's email, Pachocki had been making significant research breakthroughs that were driving OpenAI's progress forward. In recognition of these achievements, Altman promoted Pachocki to lead the company's core model development, placing him in charge of major research directions.
While Scott characterized this move as "logical from a CEO perspective," it profoundly alienated Sutskever. The promotion meant that someone who had previously worked under Sutskever was now leading critical research initiatives, including solving problems that Sutskever himself had been attempting to address for years with limited progress. Scott noted that Pachocki's work accelerated significantly after his promotion, creating additional tension as his successes highlighted the leadership change.
The Board's Susceptibility and the Final Decision
Scott's analysis reveals why these conflicts escalated to Altman's termination. In a normal corporate structure, an employee dissatisfied with such decisions would either accept them or resign. However, Sutskever was not just an employee but also a founder and board member. The board's composition made it particularly susceptible to Sutskever's arguments against Altman's decisions.
According to Scott, two board members were "effective altruism folks" who ideally wanted unlimited resources to develop Artificial General Intelligence (AGI) for study rather than practical application. None of the board members had sufficient experience in managing such situations or understood OpenAI's internal dynamics well enough to recognize that firing Altman would exacerbate rather than solve their concerns.
The Timeline of Events Leading to the Firing
Scott provided a detailed timeline of the events surrounding Altman's ouster:
- Thursday late night: The board informed OpenAI CTO Mira Murati of their decision to remove Altman.
- Friday morning: Murati called Scott and Nadella approximately 10-15 minutes before the board notified Altman, marking the first time Microsoft executives learned about the situation.
- Friday noon: The OpenAI board notified Altman of his termination and removed Greg Brockman from the board, immediately publishing a blog post about the decision.
- Friday 2 PM: OpenAI held an all-hands meeting with visibly rattled staff.
- Friday afternoon: Brockman resigned after being blindsided by the decision, having not participated in board deliberations.
- Friday evening: Researchers including Jakub Pachocki began reaching out to Altman and Brockman, expressing loyalty and threatening resignation, with several following through that night.
The email provides unprecedented insight into the internal dynamics that led to one of the most dramatic leadership changes in the technology industry, revealing how resource allocation disputes and personal tensions can escalate to board-level crises even at the most prominent AI companies.
