Microsoft CTO's Candid Email Exposes OpenAI's Internal Turmoil
In a revealing internal communication that sheds new light on one of tech's most dramatic leadership crises, Microsoft's Chief Technology Officer Kevin Scott provided a blunt assessment of why OpenAI's board abruptly fired CEO Sam Altman in November 2023. The email, sent to Microsoft CEO Satya Nadella on November 19, 2023, exposes how internal power struggles over computing resources and personal rivalries triggered the shocking dismissal.
The GPU Resource Battle That Divided OpenAI
Scott's detailed account reveals that the first major conflict centered around computing resources, specifically Graphics Processing Units (GPUs). With ChatGPT's explosive success creating unprecedented demand, OpenAI faced difficult allocation decisions between research teams and applied AI products.
Research teams felt increasingly starved as GPUs were diverted to support the API and consumer-facing products. As Scott explained in his email, researchers could "always use more GPUs because what they're doing is literally insatiable." This created a fundamental tension within the organization, with research teams blaming Applied AI for hogging resources while Applied AI believed they deserved priority for product development.
The situation highlighted a critical organizational failure: neither side fully appreciated the bigger picture of resource allocation in a rapidly scaling AI company. This resource battle created simmering tensions that would eventually contribute to the board's decision-making.
The Promotion That Changed Everything
The second, more personal conflict involved Jakub Pachocki, a researcher who had previously worked under OpenAI co-founder and board member Ilya Sutskever. Pachocki's significant research breakthroughs caught Altman's attention, leading to his promotion to lead OpenAI's core model development.
While this move made logical sense from a CEO's perspective for advancing the company's technical capabilities, it created deep personal friction. Sutskever reportedly struggled significantly with watching his former subordinate leapfrog into a leadership position tackling problems that he himself had been attempting to solve for years with limited progress.
Scott noted in his email that "Ilya has had a very, very hard time with this," describing how the rapid shift in organizational hierarchy created personal tensions that would eventually influence board dynamics.
Board Inexperience and One-Sided Narratives
Scott's assessment of OpenAI's board was particularly critical. He described some members as "effective altruism folks" who wanted unlimited resources to build Artificial General Intelligence (AGI) primarily for study and contemplation rather than practical application.
The board lacked sufficient business experience to understand that firing Altman wouldn't address their underlying concerns about resource allocation and organizational direction. Instead, Scott argued, the dismissal would exacerbate existing problems. The board also failed to seek appropriate external counsel about managing a CEO transition at what was arguably "the hottest company in the world" at that time.
The Chaotic Timeline of Events
Scott's email provides a detailed chronology of the unfolding crisis:
- Thursday night: The board alerted interim CEO Mira Murati about their decision
- Friday noon: Altman was officially fired, Greg Brockman was removed from the board, and a public blog post announced the changes
- Late Friday: Researchers began flooding Altman with messages pledging loyalty and threatening to resign
- Within five days: Nearly every OpenAI employee signed a letter demanding Altman's return
Pachocki, whose promotion had contributed to the initial tensions, was among the first to reach out to Altman following the dismissal. Microsoft quickly intervened with job offers to OpenAI staff, and the board eventually capitulated to employee pressure.
Aftermath and Reflection
In the aftermath of the crisis, Altman later told Microsoft co-founder Bill Gates that the entire ordeal was "extremely painful" but ultimately strengthened the company. Scott's email captures the fundamental reasons behind the drama: inexperienced board members fell for an incomplete narrative about resource battles and bruised egos, triggering a crisis that nearly destroyed one of AI's most promising companies.
The incident serves as a cautionary tale about governance in fast-growing technology companies, highlighting how internal conflicts over resources and organizational hierarchy can escalate into full-blown leadership crises when handled by inexperienced directors.
