Anthropic's IPO plan would give seven founders 50.1% of the vote on about 14% of the company

A platform lead with Claude behind a customer-facing feature will eventually be asked who Anthropic answers to once it lists. The usual assumption is that a listing hands a vendor's priorities to the market. According to The Information, Anthropic's plan is built to stop that on most corporate matters. It would give the seven co-founders 50.1% of the vote through a special share class ^[1]. The shares would carry no extra economic value. Their only purpose, TNW reported, is to keep the group in charge once the stock trades ^[5].

The founders would hold that vote on a small slice of the company. Each owns about 2%, Amodei included, TechCrunch reported ^[2], so the seven together hold about 14% ^[3]. Their share of the vote would be about 3.6 times their share of the ownership ^[18].

The Information's headline calls the structure "Palantir-style" ^[11]. Palantir's founders kept control through Class F shares that also depend on a minimum stake ^[12]. The size is different. Palantir's prospectus says those shares "will give these Founders the ability to control up to 49.999999% of the total voting power of our capital stock" ^[13]. Anthropic's reported class would clear half the vote on its own ^[20]. TechCrunch noted a second difference. Super-voting shares are common among tech founders, and TechCrunch wrote: "What's unusual is the group approach." ^[14]

Amodei announced the founders' giving pledge in January, in an essay on the risks of AI. "All of Anthropic's co-founders have pledged to donate 80% of our wealth," he wrote ^[15]. In the same essay he wrote, "The thing to worry about is a level of wealth concentration that will break society." ^[16]

For a Claude buyer, control after the listing would sit in two places. The founders would hold the shareholder vote. The Long-Term Benefit Trust, which added former Fed chair Ben Bernanke in July ^[9], would pick most of the board ^[6]. The reports do not say how decisions on model behaviour or usage policy would be divided between the two.

A team deciding how far to commit to Claude can sort each dependency along two lines. The first separates dependencies that need Anthropic to keep its current course from those that need it to change. The second separates decisions that go to a shareholder vote from decisions the board makes. A dependency that needs Anthropic to hold course on a shareholder matter rests on the founders' majority ^[1]. One decided at board level rests on the trust's appointees ^[6]. A dependency that needs change requires the founders' agreement in the vote column and the trust-picked majority's in the board column. On the plan as reported, outside shareholders settle none of the four cells while three founders keep the minimum stake ^[4].