Unredacted Messages Expose Microsoft’s Role in OpenAI’s Board Shake-up
Newly unredacted court filings from the Musk v. Altman trial pull back the curtain on how deeply Microsoft executives were involved in reshaping the OpenAI board during the November 2023 leadership crisis. In a text thread with Sam Altman, Microsoft CEO Satya Nadella and president Brad Smith, CTO Kevin Scott sifted through potential directors for what would become a revamped OpenAI governance structure. Some names drew enthusiasm, such as former Airbnb COO Belinda Johnson and former Gates Foundation CEO Sue Desmond-Hellmann, whom Nadella personally suggested and who ultimately joined the board. Others triggered outright resistance: Scott’s proposal of former Google Cloud CEO Diane Greene prompted his “strong, strong no,” and Nadella later acknowledged he opposed Greene and veteran gaming executive Bing Gordon over their ties to Microsoft’s AI rivals. The exchanges underscore how board composition became a strategic battleground, not just a governance formality.
The IBM Ghost: Nadella’s Fear of One-Sided Dependency
Behind the scenes, Microsoft’s assertiveness over OpenAI governance was rooted in a specific anxiety: becoming “the next IBM.” In an internal email presented at trial, Nadella warned colleagues that as Microsoft weighed another USD 10 billion (approx. RM46 billion) investment in April 2022, he did not want a repeat of the historical dynamic where a hardware partner ceded the strategic high ground to a software upstart. He described the OpenAI partnership as a “one-way door,” emphasizing that Microsoft could not afford to build separate supercomputers for itself and OpenAI. Instead, it was effectively outsourcing core AI research while committing scarce compute to a single partner. Nadella testified that this heavy dependency made it essential for Microsoft to secure access to OpenAI’s intellectual property and to ensure the company continued building its own internal AI expertise alongside the partnership.
Boardroom Gatekeeping as Competitive AI Strategy
The OpenAI board debate reveals Microsoft’s broader AI strategy competition: protect its investment while limiting competitor influence. Nadella testified that he objected to Diane Greene and Bing Gordon specifically because of their past leadership roles connected to Google Cloud and Amazon, both major players in AI infrastructure and services. By contrast, Microsoft executives favored candidates with strong governance credentials but fewer direct ties to cloud or AI competitors, such as Anne Sweeney, Leslie Kilgore, and others proposed by Brad Smith and Kevin Scott. Nadella even rejected Scott’s tongue-in-cheek suggestion that he join the board himself, highlighting a preference for influence without overt control. The resulting framework, centered on Bret Taylor, Larry Summers, and Adam D’Angelo, reflected a delicate compromise: OpenAI would publicly project independence, while Microsoft quietly ensured the board would not tilt toward rival ecosystems that could jeopardize its strategic position.
Power, Independence, and the Future of OpenAI Governance
Internal emails and courtroom testimony highlight a persistent tension between Microsoft’s financial stake and OpenAI’s nonprofit-origin mission. Musk’s lawsuit argues that Microsoft’s attempts to shape OpenAI governance undermined the charitable trust meant to prioritize humanity over any single corporate partner. Nadella countered that Microsoft assumed enormous risk to back a lab few were willing to fund, helping create a nonprofit that now sits atop a valuable stake in OpenAI’s for-profit arm and enabling products such as ChatGPT and Copilot. Yet under cross-examination, he acknowledged limited knowledge of the nonprofit’s staff and outputs during this period, fueling questions about how independent OpenAI’s governance truly was. The board that ultimately emerged—with figures like Desmond-Hellmann, Bret Taylor, Adam D’Angelo, and later additions such as Zico Kolter and Paul Nakasone—embodies that unresolved balance: a foundation formally separate, but shaped by a partner determined not to lose control of the next wave of AI innovation.