Unredacted Messages Expose Microsoft’s Shortlist for the OpenAI Board
Newly unredacted court filings from the Musk v. Altman trial shed light on how deeply Microsoft executives were involved in reshaping the Microsoft OpenAI board during the November OpenAI governance crisis. In a group text with Sam Altman, Satya Nadella, Brad Smith, and Kevin Scott, potential directors were rapidly evaluated like candidates in a high‑stakes draft. Former Google Cloud CEO Diane Greene drew what Scott called a “strong, strong no,” a reaction Nadella later linked to her ties with a major AI rival. Another name, veteran gaming executive and long‑time Amazon board member William “Bing” Gordon, was also rejected for competitive‑alignment reasons. By contrast, Belinda Johnson, the former Airbnb COO, was greeted as “great,” and Nadella signaled approval. The thread reveals not only who Microsoft vetoed and endorsed, but also how quickly a tentative three‑person framework—Bret Taylor, Larry Summers, and Adam D’Angelo—emerged around Altman’s potential return as CEO.
Satya Nadella’s Strategic Fear: Becoming ‘the Next IBM’
Behind Microsoft’s assertive posture in OpenAI’s governance crisis was a deeper strategic anxiety Satya Nadella had already articulated. In an internal email presented at trial, he warned against repeating history: Microsoft, he wrote, must avoid becoming the “IBM” of this era while OpenAI became the new Microsoft. Nadella described the company’s multibillion‑dollar OpenAI investment as a “one‑way door,” stressing that Microsoft could not realistically build two full‑scale AI supercomputers—one for itself and one for OpenAI. That meant accepting a massive dependency on a partner for core AI research and intellectual property. In his Satya Nadella testimony, he emphasized the need to secure access to OpenAI’s IP while growing internal expertise, underscoring how corporate AI partnerships can blur the line between collaboration and strategic vulnerability. This concern helps explain why Microsoft pushed hard for board candidates unlikely to dilute its influence or strengthen direct competitors.
Vetoes, Endorsements, and a Carefully Curated ‘Independent’ Board
The unsealed message thread reads like a crash course in how a major partner can shape another company’s governance without formally holding the gavel. Nadella and his lieutenants filtered candidates through a competitive‑risk lens: Diane Greene and Bing Gordon were out because of ties to companies vying with Microsoft in AI. Other names triggered more enthusiasm. Sue Desmond‑Hellmann, the former Gates Foundation CEO, was Nadella’s own suggestion and later joined the OpenAI board. Brad Smith floated media and tech veterans such as Anne Sweeney and Leslie Kilgore, describing them as calm, practical, and deeply experienced. Kevin Scott compiled a longer bench of operators and investors, even jokingly proposing himself before Nadella pushed back. By the evening, Altman proposed Bret Taylor, Larry Summers, and Adam D’Angelo as a slimmed‑down board, with Altman off the board but back as CEO. Microsoft did not formally appoint directors, yet its preferences clearly bounded the field.
Governance Tension: Partnership Versus Control in Corporate AI
The OpenAI governance crisis highlighted a broader dilemma in corporate AI partnerships: how to balance independent oversight with the demands of a dominant strategic investor. Microsoft insists its suggestions were advisory, and Nadella testified that OpenAI’s board was free to ignore them. Yet the trial record shows Microsoft diligently screening candidates, calling Larry Summers directly, and opposing directors tied to rivals, all while its cloud and funding commitments underpinned OpenAI’s survival. Elon Musk argues this influence diverted OpenAI’s nonprofit mission, claiming Microsoft aided a breach of the charitable trust that originally governed the lab. Nadella counters that Microsoft took enormous risk to help create one of the world’s largest nonprofits and products like ChatGPT and Copilot. Under cross‑examination, he acknowledged gaps in the nonprofit’s activity, exposing unresolved questions about who truly steers OpenAI’s trajectory—and whether nominal independence can coexist with such concentrated commercial power.