Why the People Who Started Anthropic Want to Keep Control After It Goes Public
Anthropic wants its seven founders to keep 50.1% of the vote on most company decisions, even after the company sells shares to the public. TechCrunch
The plan uses a new type of special share. It gives extra voting power but no extra money. Think of it like owning a small slice of a pizza but still getting to decide the toppings.
That control would stay in place as long as at least three of the founders keep a minimum amount of shares.
Each founder owns about 2% of the company, including CEO Dario Amodei. So the group would have majority voting power with a minority ownership stake. The co-founders have pledged to give away 80% of their wealth.
The plan would also change board and employee rules. Anthropic's Long-Term Benefit Trust would continue to choose most of the board. The founders' own board seats would rise from two to three. Employees would get their own class of stock to break ties on some issues.
The request is linked to a public listing that is already underway. Anthropic filed confidentially for an initial public offering, selling shares to the public for the first time. It announced the step in a company blog post. TechCrunch
The expected values are high. Anthropic was valued at $965 billion in May 2026. It was more recently valued at $1.5 trillion on the secondary market, where existing private shares change hands. Nvidia Corp. is considering investing as much as $10 billion in Anthropic PBC's initial public offering. Bloomberg
Economics and control would diverge. That is the point.
The broader context here is how AI labs are trying to fix their rules before public investors arrive. A 50.1% founder bloc centralizes veto power, even while the Trust still picks the board and staff hold a tie-breaker on defined questions.
In my view, the structure is best read as an attempt to separate capital scale from decision scale. Anthropic needs public-market money at very large size, but wants model and safety choices shielded from short-term shifts. The no-extra-money rule, the three-founder rule, and the pledge to give away most wealth point in that direction.
Worth flagging is what remains unspecified in the verified terms. Minimum stake is not defined. The exact scope of most corporate matters is not defined, nor is the some issues for an employee tie-break. Those definitions will decide who settles contested cases.
The optimistic reading here, and the one I lean toward, is that steady rules can help long-term technical work survive a listing. Training, testing, and release decisions play out over years, not quarters.
