Who Owns Claude? Anthropic's Published Structure, Explained

On Google Trends, worldwide, over the past seven days and read on 23 October 2026, "who owns claude" is rising under the "claude bot" seed, in the same week that people were also asking who owns Grok and who owns Perplexity. The honest answer takes two sentences. Claude is a product; the company behind it is Anthropic. Anthropic is a public benefit corporation, and the documents it publishes describe governance rather than shareholdings, which means the ownership answer most people want is not a public document.

What is public is a set of statements about structure, and they are specific enough to be worth reading slowly. Three documents carry almost all of it, all read on 23 October 2026: Anthropic's company page, its 21 October 2026 announcement that a physicist had joined the Long-Term Benefit Trust, and its 18 October 2026 announcement of the Anthropic Public Benefit Fund. Everything below comes from those three, apart from facts I flag as coming from somewhere else. Where a question is not answered by them, I say so rather than fill the gap.

What the company says it is

Anthropic describes itself as a public benefit corporation "dedicated to securing its benefits and mitigating its risks," a phrase that appears in the footer of its own pages. The Long-Term Benefit Trust announcement puts the same idea in operational language: Anthropic is "structured to balance commercial growth with our public benefit purpose."

That is a claim about structure rather than sentiment, and the rest of the documents are best read as an attempt to show the structure working. A company that says it balances two objectives has to name who holds the balance and how the holding survives a bad quarter. Anthropic's answer is a trust with board authority and a fund that holds shares.

The Long-Term Benefit Trust

The Trust is the governance body. According to the 21 October announcement, it "has the authority to appoint a majority of Anthropic's board," and trustees "advise the board and our leadership on critical decisions, particularly those involving risks and societal impacts of AI." The same release states that trustees are independent of the company's management and investors, and that the Trust exists to help the company hold its balance over the long term.

On 21 October 2026, Anthropic announced that Dr Christopher Stubbs had joined. He is an experimental physicist, the Samuel C. Moncher Professor of Physics and of Astronomy at Harvard, and was Dean of Science there from 2018 to 2024. In 1998, as a member of the High-Z Supernova Search Team, he helped discover that the expansion of the universe is accelerating, work that later earned the team the Breakthrough Prize in Fundamental Physics. He was the inaugural project scientist for what is now the Vera C. Rubin Observatory, which began a ten-year survey of the southern sky in June. He is a member of JASON, the independent group of scientists that has advised the US government on defence and technology since 1960, and he has advised the US government on national security and arms control.

He is the Trust's first physicist. He joins Neil Buddy Shah, who chairs it, the national security expert Richard Fontaine, and the former Federal Reserve Chair Ben Bernanke.

The reasons a physicist would be a plausible trustee are visible in his own statement. "I use AI tools in my research every day and see what they make possible, and as an educator I have confronted some of the hard questions they raise," Stubbs said. "As a grandfather, I want to help ensure that AI is developed and deployed for the long-term benefit of humanity." He has also written recently about how to keep AI-assisted research reproducible at a time when models can give different answers to the same question — a technical version of the problem a governance body is supposed to notice.

Daniela Amodei, Anthropic's co-founder and president, framed the appointment around scientific temperament rather than credentials: "When Chris found evidence that the universe was expanding at a rate no one expected, he and his team followed the data even when it went against the prevailing wisdom of the field." Neil Buddy Shah's comment in the same release describes him as a leader of large, complex, collaborative scientific efforts across academia, government and the national laboratories.

The Public Benefit Fund, and the 5%-plus-2% mechanism

Three days earlier, on 18 October 2026, Anthropic announced the Anthropic Public Benefit Fund. The wording is precise and worth quoting, because most summaries of it are not. The Fund is "a new wholly owned entity with a structure that increases the stake as the company grows. Subject to stockholder approval, 5% of Anthropic's shares will be committed today, with an additional 2% added at the board's discretion for every 50% increase in the company's value above a set baseline."

Three things in that sentence do most of the work. The first 5% is subject to stockholder approval, so it is a commitment pending a vote rather than a completed transfer. The extra 2% is discretionary — it is added "at the board's discretion," not automatically. And the trigger is growth in the company's value above a baseline, not time, revenue or headcount. The baseline itself and the milestones that would add to the stake have not been published; Anthropic says it will share both, along with more about the Fund's structure and governance, "in the coming months."

The entity is being set up as an LLC rather than a 501(c)(3) or another format, which Anthropic says is "to give adjustability between different forms of giving." It is explicit about one limit: none of the Anthropic equity held by the LLC will be used for lobbying or political giving, in the US or abroad. The uses it sketches are looser — large donations to traditional nonprofits, direct distributions to specific populations — and it gives labour-market impacts as one example. Anthropic says it has already committed $200 million to studying and addressing that.

The stated logic for the ratchet is unusual enough to repeat. Anthropic notes that it is common practice for technology companies to offer escalating compensation awards to their executives, and says it is directing that mechanism to the public benefit instead. Whatever you think of the analogy, it explains the design: the fund is meant to grow in step with the thing it is funded from.

Where the roughly 20% figure comes from

The 18 October post also supplies a number that circulates without its context: roughly 20% of Anthropic shares will be dedicated to the public good once the Fund is added to existing commitments. The post explains the two components. When Anthropic was founded, the co-founders pledged to donate 80% of their equity holdings. Separately, the company has run an equity donation match programme since founding, through which employees can pledge equity that Anthropic matches; approximately half of employees participate.

Combine those with the Fund announcement, the post says, and roughly 20% of Anthropic shares end up dedicated to the public good, plus whatever additional percentage the valuation milestones add. That is Anthropic's own aggregate figure, published by the company. It is not something I can reconstruct from the parts, and I have not tried to.

Control and ownership are not the same question

This is where most "who owns" answers go wrong, and it is worth separating carefully. On control, the documents publish something concrete: the Trust has the authority to appoint a majority of the board. On money, they publish the Fund's 5% and the co-founder pledge. On ownership itself, meaning who holds which shares and in what proportion, they publish nothing at all.

What is not established by anything Anthropic has published: the full cap table, the identity and size of any investor's stake, how voting rights are allocated among holders, or which person or entity ends up with the largest block. Those questions may have answers. The answers are just not in the public record, and anyone who supplies them with a number attached is either working from a filing I have not seen or making one up.

One role is safe to state and easy to confuse with ownership: Dario Amodei is Anthropic's chief executive. A title describes what someone does. It does not describe what someone holds.

What the first Public Benefit Report adds

The week of 21 October also brought Anthropic's first Public Benefit Report, described by the company as "a public account of how we're working toward our charter and where we're falling short." Two structural details matter more than the contents. The Trustees helped define the five charter pillars the report measures the company against, and the report closes with the trustees' own assessment of progress. That is a different arrangement from a company grading itself: the same body with board-appointment authority is signing off on the scorecard.

What this means if you are just trying to pick a provider

Governance documents answer a question about accountability, not a question about which endpoint to call. They tell you who is supposed to hold a company to its stated purpose over a long horizon. They tell you nothing about price, context length, uptime or which model to use for a task, and they should not be read as if they did.

Those questions have their own published numbers. Claude Opus 5.5 is on our catalogue at $4.00 per million input tokens and $20.00 per million output tokens, with a 1M-token context and cache reads at $0.20 per million. It arrives through one OpenAI-compatible endpoint that carries more than 200 models under a single key, with automatic failover across providers and provider list prices passed through at 0% markup. We host models; we do not run Anthropic's apps or Claude Code, and corporate structure at the model vendor is not something a gateway can change.

How to check an ownership claim about any AI lab

The habit that pays off here is asking what kind of claim you are being handed. A statement about board authority is a governance claim and can be checked against a named announcement with a date. A statement about a percentage stake is an ownership claim and needs a filing or a published figure. A statement about who "really" controls a company is usually neither, and is the one most likely to be invented.

The second habit is looking for the aggregate someone is quoting, and asking what it includes. Anthropic's roughly 20% figure is a good example: it combines the Fund, a co-founder pledge made at founding, and an employee matching programme, and it is explicitly contingent on milestones that have not been set out yet. Repeat it as "roughly 20%, per Anthropic, subject to stockholder approval and milestones," or do not repeat it.

The third is noticing what a document declines to say. Anthropic is unusually forthcoming about the mechanics of the Fund — the percentage, the trigger, the entity type, the prohibition on political use — and completely silent on the cap table. That combination is not an accident of editing. It tells you which questions the company considers answerable in public and which it does not, and that distinction is more informative than any speculative answer would be.

Which leaves the original query in a mildly unsatisfying place. Claude is owned, in the ordinary legal sense, by whoever holds Anthropic's shares, and Anthropic has not published who they are. What it has published is the machinery around them: a trust that can appoint a board majority, a fund that grows with the company, and a report in which the trustees assess the company against pillars they helped write. For a question that sounds like gossip, that is a surprising amount of structure to find.