Anthropic has asked shareholders to approve a governance structure that would hand its seven co-founders a combined 50.1 percent of the vote on most company matters, according to a report in The Information on September 24. The request came with a short deadline: shareholders were asked to decide within days.
The arrangement would rest on a single condition. At least three of the seven founders must keep a minimum shareholding for the structure to remain in force, a clause designed to keep the control block from outliving the group’s involvement in the company.
Dual-class share structures are familiar in technology, though rarely in this form. Palantir Technologies uses a similar model, and the structure is the explicit reference point for the Anthropic proposal. Mark Zuckerberg holds supervoting shares at Meta Platforms, and Evan Spiegel does at Snap. What makes this plan unusual is that seven people would share the control bloc rather than one founder holding it alone.
None of the founders holds much equity. Each owns roughly 2 percent of the company, according to the report, and all seven have pledged to give away 80 percent of their wealth. The new shares would not carry additional economic rights, meaning the founders would gain influence over decisions without gaining a larger claim on profits.
Supporting provisions fill out the design. A long-term benefit trust would continue to determine a majority of board seats, the founders’ representation on the board would rise from two seats to three, and employee shares could be used to break deadlocks on some questions. Anthropic created the benefit trust as part of its unusual corporate design, giving the body the power to appoint directors rather than leaving that choice entirely to shareholders.
The company’s valuation gives those provisions unusual weight. Anthropic raised money in May at a valuation of $965 billion, and secondary market trades have since priced it at $1.5 trillion. It filed confidentially for an initial public offering in June and selected the Nasdaq for its listing.
Control questions have followed Anthropic since its founding, when the company was set up as a public benefit corporation with a mission centered on safety. The benefit trust was part of that design. The new voting structure extends the same instinct, keeping decisions out of the hands of public shareholders even as the company takes their money.
The founders have described the proposal as a way to protect the company’s mission through the demands of quarterly reporting. A chief executive who can be removed by a shareholder vote faces pressure to prioritize short-term results, and the argument for supervoting shares is that it removes that pressure during the years when an AI developer’s choices about safety and deployment carry the most consequence.
Critics of the structure make a narrower point. Index funds and large institutional investors have campaigned against dual-class listings for years, on the ground that they give public shareholders capital at risk without a corresponding voice. Providers of proxy advice generally recommend against such structures, though their objections rarely stop a listing when the company is large enough to be added to major benchmarks.
Investors in an IPO face a narrow choice. They can buy into one of the largest private companies in the world and accept that their votes will not decide much, or they can decline. The second option is harder for funds that track broad market indexes, which will have to hold the stock if it is included.
The proposal also raises a practical question about accountability. A control group of seven founders with small personal stakes and no added economic rights has less financial exposure to the decisions it makes than an individual founder with a large position. Whether the benefit trust can offset that imbalance is untested, and the trust’s own members are not elected by shareholders.
The founders have also argued that the structure is a defense against takeover pressure. A company whose shares are valued at $1.5 trillion on the private market will attract interest from acquirers and activist investors once it trades publicly, and a 50.1 percent block is the most direct way to make a hostile approach impossible.
Governance specialists have argued for years that dual-class structures tend to persist longer than their designers promise. Founders say the protections are temporary; courts and boards have generally allowed them to run indefinitely. Anthropic’s version at least includes a sunset trigger, though it is tied to the founders’ holdings rather than to a calendar.
For employees holding shares that will convert at the listing, the change in voting power is unlikely to affect the value of the stock. For anyone expecting the discipline of a shareholder vote to shape how Anthropic builds and deploys its models, the proposal answers the question before it is asked.
Shareholders have days to respond. What happens if they decline has not been described publicly.
