Anthropic / 29 September 2026

Anthropic founders seek 50.1% voting control after IPO

Anthropic's seven co-founders would direct 50.1% of shareholder votes through a proposed Founder LLC and a single powerful Class F share after its IPO, according to prospectus details reported by Reuters on 29 September. The arrangement covers major corporate matters including some board elections. Ordinary Class A shareholders would have one vote per share but little practical ability to outvote the founder bloc. Anthropic says the structure is designed to protect its public-benefit mission as it enters markets that may press for faster model releases and greater returns on enormous compute commitments. It is an unusually concrete answer to who will decide when commercial opportunity conflicts with the lab's stated safety priorities.

The Founder LLC initially includes chief executive Dario Amodei, company president Daniela Amodei, chief compute officer Tom Brown and researcher Chris Olah among the seven co-founders, Reuters reported. A majority of that group would direct the Class F vote. The IPO document describes them as ‘distinctly equipped to be stewards of our mission.’ Some elements of the planned structure were reported earlier in September; the prospectus account supplies the voting percentage, class mechanics and fuller relationship to the board. The control is collective rather than held by Dario Amodei alone, though the founders have a history of working closely together since they left OpenAI and created Anthropic.

Anthropic also retains its Delaware public benefit corporation status, which asks directors to consider its stated mission alongside investor interests. A separate Long-Term Benefit Trust would elect four of the company's board directors, while the founder-linked voting structure and Class A holders would elect three, according to the prospectus account. Current trustees include former Federal Reserve chair Ben Bernanke and national security expert Richard Fontaine. This creates two centers of mission-related influence, one among the founders and one through the trust. It may make it harder for a future shareholder majority to force a change in model-release policy, but it also asks outside investors to accept less power over spending, capital allocation and executive accountability.

The filing says the structure could permit decisions that conflict with short- or long-term financial performance and lower the value of Class A shares. Anthropic points to past choices, such as limiting access to a powerful cybersecurity model and declining to develop image or video generation, as examples of directing resources toward its research and safety priorities. The founder voting class begins a sunset process when no more than two qualifying co-founders or successors remain. Reuters also reported that the founders pledged to devote 80% of their personal Anthropic equity to charitable causes. Those provisions give the mission an organizational design, although the results will depend on how the leaders exercise their power during hard product and financing choices.

The plan arrives alongside reports that Anthropic may seek a valuation above $2 trillion while its prospectus describes hundreds of billions of future infrastructure spending. New public shareholders would provide capital and liquidity but not proportionate control. That can be attractive to investors who believe the founding team's judgment is a durable asset; it can deter those who want stronger influence over returns, dilution or succession. The prospectus has not been publicly filed, and Reuters reviewed a confidential copy rather than a final offering document. Still, the disclosed control mechanism is specific enough to change how prospective buyers would value a Class A share: they would purchase exposure to the business under a mission-directed governance bargain.