Anthropic is preparing to go public, and its filings spell out who will make the big decisions once outside investors own shares. The company’s seven co-founders have set up a dedicated entity called Founder LLC, which will hold a single share of Class F stock carrying 50.1% of the total voting power. That means the founders keep control over key decisions such as electing the board, even after the stock trades on a public market. Reuters first reported the plan in an exclusive on September 28, 2026, and the listing, once expected in September or October, has reportedly slipped to November while the paperwork is finalized.
How the voting power is structured
The plan relies on a multi-class share structure, where different classes of stock carry different voting rights. Most of the control sits with one class, while the shares sold to the general public carry far less weight.
| Element | What the filing describes |
|---|---|
| Founder LLC | An entity run by the seven co-founders, including CEO Dario Amodei |
| Class F stock | A single share held by Founder LLC, worth 50.1% of all votes |
| Class A common stock | The shares ordinary investors can buy, at one vote per share |
| Other share classes | Four more classes with limited or minimal voting rights, aimed at strategic partners |
| Legal form | A Delaware Public Benefit Corporation (PBC) |
A Public Benefit Corporation is a company that is legally allowed to weigh a broader public purpose alongside shareholder returns. Anthropic’s filing uses that status to explain why its leaders may balance investor interests against the interests of humanity as a whole. The filing describes the co-founders as distinctly equipped to be stewards of the company’s mission.
The founders’ seats are not permanent. A co-founder leaves Founder LLC after quitting, dying, selling too many shares, or being fired for cause. The extra voting power begins to sunset only when two or fewer co-founders or their successors remain. In practice, as long as at least three of the seven are still in place, the majority vote stays with the founder group.
▲ One Class F share versus many Class A shares
The warning to public shareholders
The most striking part of the filing may be what it tells future Class A shareholders. It states that management decisions may conflict with short-, medium- or long-term financial interests and business performance, and that this could hurt the value of Class A stock. Read plainly, the company is reserving the right to put its mission and safety choices ahead of the share price.
The filing also acknowledges that advanced AI carries a risk serious enough to threaten humanity. At the same time, it presents the company’s main advantage as a low-ego, truth-seeking culture in which frontier AI capability and safety reinforce each other. Two concrete choices are offered as evidence of that approach:
- Restricting or delaying certain capabilities, such as the Mythos Preview model focused on cybersecurity
- Choosing not to build image and video generation models so that compute can go to safety and research
Executive pay appears in the filing as well. According to its Summary Compensation Table, Dario Amodei received nearly $18 million in 2025, mostly in stock and option awards. Co-founder Daniela Amodei was the second-highest-paid executive at $16.4 million. The co-founders have pledged 80% of their personal Anthropic equity to charitable causes.
A $2 trillion company run by seven people
The founders are reportedly seeking a valuation of around $2 trillion at the IPO. Even at that size, majority voting control would remain with seven people. The structure invites two very different readings.
The case for founder control is laid out in the filing itself. If a technology could be dangerous, the argument goes, the people steering it need distance from quarterly market pressure so they can make safety-first calls that may cost money.
The case against focuses on the same point. Seven individuals would hold largely unchecked authority over a technology their own company describes as a potential threat to humanity, while public investors supply capital but get little say. A further argument holds that ordinary investors may actually be a useful check, because their returns depend on a world full of living customers.
▲ Competing views on founder control
The criticism also reaches beyond one company. One argument is that warnings about AI’s catastrophic potential can end up reinforcing how important AI companies seem, and with it their valuations. Another is that technology shifts that benefit society in aggregate still produce concentrated losses for specific groups, much as online education put pressure on tenured professors and email and messaging apps reduced the need for postal carriers. Which side is right appears to be less a question of fact than a value judgment about who should manage AI risk and how.
What to watch
Anthropic’s IPO plan writes founder control for the sake of safety directly into the company’s legal structure. Founder LLC keeps 50.1% of the vote after listing, and Class A shareholders are told in advance that leadership decisions may work against the share price.
For readers following the story, a few points are worth checking as more details emerge:
- The risk factor section of the prospectus, especially the language on voting rights and management discretion
- The voting differences between share classes and the exact conditions that end founder control
- How much room the PBC charter gives leadership to put other goals ahead of shareholder returns
- Whether the stated safety-first principles show up in actual product decisions
Whatever anyone decides about the stock, knowing whose judgment steers the company is useful for both the people who use its AI and the people weighing whether to own a piece of it.