OpenAI's crisis shows the team, not the board, controls an AI lab

Most of OpenAI's staff threatened to follow their ousted CEO to Microsoft, and the board gave way. Talent and governance are now core diligence items.
Sam Altman is back in charge of OpenAI because most of its roughly 770 employees said they would leave without him. The nonprofit board removed him as chief executive on 17 November and dropped president Greg Brockman as its chair; Brockman quit the same day. On 20 November Microsoft, which has backed OpenAI with about $13 billion since 2019 but held no board seat, said the two would lead a new advanced AI research team inside Microsoft, and the staff letter followed, threatening to go with them unless the board resigned.
Within days the company agreed to bring Altman back under a new initial board: Bret Taylor as chair, Larry Summers and Adam D'Angelo, the only director who stayed. His return became official on 29 November, Microsoft took a non-voting observer seat and the board committed to an independent review of the events. Legal control never changed hands, yet the board lost. For capital, the real asset of a model lab is a team that can walk, and the second is governance that cannot surprise its backers.
OpenAI's capped-profit structure gives control to a nonprofit board with a duty to its mission rather than to shareholders, and its backers had accepted that. When the board used its power, people decided the outcome, not documents. A lab whose research and engineering staff could move to Microsoft in a week left the board little value to protect. Larger investors in AI labs will now ask for information rights, observer seats and clear terms on control before the next round.
Unusual structures carry a cost that shows up in a crisis, not in a pitch. A nonprofit parent, a public benefit charter or a safety board can be the right choice, provided founders write down in advance who can remove the chief executive, on what grounds, and how investors and staff are told. The week also showed how strong employees are in AI: broad equity, a clear mission and a leader the team will follow kept OpenAI together. Companies building on a single model provider learned a separate lesson. A supplier's governance can put a product at risk overnight, and a second top-tier provider is cheap insurance.
Paper may yet win back some ground. If the independent review and the names of the next directors move OpenAI toward a conventional board, and the next financing gives Microsoft or other investors more formal rights, control will sit closer to capital than it did in November. Enterprise customers hold the other half of the answer: whether they spread their workloads across several model providers after a week in which one of them nearly lost its team.
Fanaura invests its own and LP capital through a separate SPV for each deal, and for such money governance is now part of model risk. It looks for AI-native companies whose structure, board and incentives are engineered as carefully as their product, and whose teams stay for the long horizon because they own a real part of the outcome. A board can fire a chief executive. Only the people who own the outcome decide whether the company survives it.
Building an AI-native company? Apply in five minutes. →
More from Fanaura
Building an AI-native company with a team that owns the outcome and a structure investors can underwrite? Send us the project.
Fanaura invests its own capital and the capital of its limited partners (LPs), usually through a separate SPV for each deal. If we see a fit, we get in touch.
Get the news by email
One short note each weekday. No marketing, and you can unsubscribe at any time.