Seven Founders, 14% of the Equity, 50.1% of the Vote: Anthropic Formalizes Its Pre-IPO Power Structure

In late August, Anthropic was reported to be “preparing” supervoting shares for its founders ahead of a potential IPO. This week the preparation became a formal proposal with numbers attached. According to The Information, and confirmed across Reuters and TechCrunch reporting, the company is asking shareholders to approve — “in the coming days” — a special class of shares that would hand CEO Dario Amodei and his six co-founders a combined 50.1% of the vote on most corporate matters, despite the group owning roughly 14% of the equity in total, about 2% each.

The vote is expected within days, and its outcome will lock in the governance architecture of what is widely expected to be one of the largest IPOs in history. Anthropic was valued at $965 billion in May; it now trades around $1.5 trillion on the secondary market, and the listing is expected to price off that number.

The structure, in plain terms

The proposal creates a founder share class carrying disproportionate voting power but, importantly, no additional economic value. The founders do not get a bigger slice of profits or dividends — only a bigger say. Three conditions and caveats define how it works:

• The three-of-seven rule. The 50.1% bloc holds only as long as at least three of the seven founders retain an undisclosed minimum stake. Departures dilute the founders’ collective grip; enough departures dissolve it.
• The trust keeps its teeth. The Long-Term Benefit Trust (LTBT) — the independent body Anthropic created in 2023 — still appoints the majority of the seven-seat board. Founder voting control operates underneath the trust’s board authority, not over it.
• Founders go from two board seats to three. The group’s direct board representation grows even as the trust retains the appointing power.
• Employees get tie-breaker stock. Employees will hold their own class of stock with vote-splitting rights on certain issues — a mechanism designed to prevent the founder bloc from being a pure dictatorship of seven.

Reuters characterizes the setup as Palantir-style, a reference to the founder-controlled voting architecture Alex Karp’s company carried into its own 2020 direct listing. But the more familiar analogies are Zuckerberg’s Meta and Evan Spiegel’s Snap — super-voting shares are a standard founder-control instrument at this point. What is genuinely unusual here is the group approach: not one founder-king, but a rotating, conditional, seven-person coalition.

Why the founders say they need it

The stated rationale is mission continuity. Anthropic’s leadership has argued for three years that its governance is built to survive shareholder pressure — that the Claude maker should not be subject to the same quarterly-earnings incentives that push public companies toward aggressive capability shipping and away from safety spending. A public float would dilute founder influence over time; the supervoting class preempts that.

There is also a personal dimension TechCrunch highlighted: the seven co-founders have pledged to give away 80% of their wealth — a commitment Amodei announced in January alongside a warning that AI-driven wealth concentration could destabilize society. If your economic stake is destined to shrink via philanthropy, voting power is the only lever left. The share class preserves control precisely for people who intend to own less.

The context of Anthropic’s scale makes the stakes concrete. The company’s annualized revenue run rate topped $65 billion mid-year — roughly double OpenAI’s at the time — with Q2 revenue of $11.6 billion and its first quarter of positive adjusted operating income. An IPO at a $1.5 trillion valuation would be the largest AI company listing ever attempted, and possibly the largest US IPO ever. Whoever holds 50.1% of the vote in that company holds a historically unprecedented concentration of power over frontier AI development.

The governance stack, now complete

What Anthropic is assembling is a three-layer structure no other frontier lab has attempted:

1. Founders with conditional majority voting control — operational and strategic direction, insulated from activist investors and dilution.
2. The Long-Term Benefit Trust with board-appointment power — mission oversight that neither founders nor public shareholders can vote away.
3. Public benefit corporation status — the legal charter obliging the company to balance commercial success against public benefit, giving the trust’s mandate legal force.

Each layer checks a different failure mode. Supervoting shares protect against hostile capital. The trust protects against founder drift. The PBC charter binds both to a mission that is not merely aspirational. The employee tie-breaker stock adds a fourth, thinner layer against founder-group groupthink.

Skeptics will note the tension: a structure designed to prevent any single actor from dominating frontier AI now concentrates 50.1% of votes in seven individuals. Anthropic’s answer is that the concentration is conditional (three-of-seven), subordinate to the trust, and time-limited by the founders’ own giving pledges. Whether those guardrails hold through the first activist campaign, the first takeover rumor, or the first founder falling-out is exactly what the next year of public-market scrutiny will test.

What to watch

The shareholder vote is expected within days, and approval is likely — the founders’ 14% plus supportive institutional holders such as Google and Amazon, who have consistently backed management, should clear the threshold. The real questions start after the S-1:

• Does the LTBT’s authority survive the listing intact, or does underwriter pressure carve into the trust’s board-appointment majority?
• How do index funds price a three-class structure with a conditional founder bloc — the same institutional investors who have campaigned against multi-class share structures for a decade?
• Does the three-of-seven condition ever bind? If two founders depart early, the bloc falls below the threshold — and the governance story changes overnight.

For an industry still arguing about who should control frontier AI, Anthropic’s IPO was always going to be the first real-world stress test of mission-locked governance at public-market scale. This week’s proposal is the final blueprint before that test begins: seven founders, 14% of the equity, and — if shareholders say yes — 50.1% of the vote.

---

Figures in this article are drawn from The Information, Reuters, and TechCrunch reporting published September 24–26, 2026. Anthropic has not publicly commented on the vote beyond the shareholder proposal itself.