Who chooses the board now?
According to Reuters’ account of Anthropic’s October 26 IPO filing, its Long-Term Benefit Trust holds Class T shares that let it elect a majority of directors. Separately, the seven co-founders’ Founder LLC holds Class F supervoting rights worth 50.1% of voting power on certain matters. Those include elections for directors outside the trust-selected majority and shareholder proposals generally. The trust and founders therefore have distinct roles; neither figure means ordinary investors have no vote at all.
What happens after the founders leave?
The filing describes a conditional future change. Once most co-founders have departed and Class F supervoting rights are gone, the board would be split into three similarly sized classes with staggered three-year terms. Only some directors would stand for election each year. That can make a sudden change in board control harder, but it also limits shareholders’ ability to replace all directors at once. Anthropic has not already completed this transition.
The safety mission and the accountability trade-off
Anthropic is a Delaware public benefit corporation, which places a stated public-benefit purpose alongside investor interests. Reuters says the trust’s power makes Anthropic a “controlled company” under Nasdaq rules, allowing some governance exemptions. The filing names Daniela Amodei to its nominating committee, Reed Hastings as lead independent director, and Vas Narasimhan and Chris Liddell to a safety and risk committee. These arrangements may help sustain a long-term mission, but they do not guarantee AI safety or remove potential conflicts. Outside shareholders will have less ability to change leadership quickly.
What remains a proposal or estimate?
The reported $1.5 trillion valuation is an expectation, not a completed IPO valuation or current public market capitalization. The later staggered board depends on the specified founder departures and Class F rights ending. Anthropic’s dramatic scenarios about AI’s potential benefits and harms are warnings about possible futures, not observed outcomes. Watch the Daily AI Chat Short or listen to the full podcast episode. Source: Jeffrey Dastin and Ross Kerber, Reuters, October 27, 2026.
