Analysis
The governance structure Anthropic built in 2023 as a safety mechanism is about to become a disclosure item in the largest IPO ever attempted. Its Long-Term Benefit Trust owns no shares, but a special class of stock lets its trustees appoint and remove a majority of the board outright -- four of seven directors today -- which Ars Technica notes puts three people with no financial stake in charge of the company public investors would be buying into.
The current trustees are Neil Buddy Shah, CEO of the Clinton Health Access Initiative and the trust's chair; Richard Fontaine, chief executive of the Center for a New American Security; and Ben Bernanke, the former Federal Reserve chair, who joined on July 9, 2026. Mariano-Florentino Cuellar, a former California Supreme Court justice, left the trust on August 4 to become Anthropic's first chief global affairs officer -- a move from overseer to executive that critics of the structure will cite.
Pulse covered Anthropic's $15 billion credit facility last week as the company built its pre-listing balance sheet. What has changed since is the focus: the financing questions are largely settled, and attention has shifted to control.
“A $2 trillion debut would require public investors to accept both a valuation roughly double the May private mark and a control structure that gives them no recourse.”
The numbers behind the listing:
- Targeted valuation: up to $2 trillion
- Prior private mark: $965 billion, set in May 2026
- Revenue run rate: roughly $65 billion in July, up from about $9 billion at the end of 2025
- Strategic ownership: Amazon roughly 21%, Alphabet about 15%
Dual-class structures are ordinary in tech listings -- Meta, Alphabet and Snap all went public with founders holding voting control, and index providers responded by restricting some non-voting shares from major benchmarks. What is not ordinary is control vested in outsiders with no economic interest, accountable to a charter rather than to shareholders. The closest analogues are foundation-controlled European corporates like Novo Nordisk and Ikea's ownership structure, not US technology listings.
The practical question for institutional buyers is what the trust can actually do. If it can replace directors over a disagreement about deployment pace or a safety decision that costs revenue, that is a risk factor with no precedent in a US mega-cap listing. Anthropic's answer is that the structure is the product: enterprise customers and governments buy Claude partly because of the company's safety posture, and the trust is what makes that posture credible rather than promotional.
The bear case is simpler than governance theory. A $2 trillion debut would require public investors to accept both a valuation roughly double the May private mark and a control structure that gives them no recourse. Underwriters -- Morgan Stanley, Goldman Sachs and JPMorgan are leading -- will price that discount somewhere, and it will show up as a lower multiple rather than as a negotiated change to the trust.
The document to read when the S-1 goes public is the risk factor describing the trust's powers and any sunset provisions. That language, not the valuation range, is what determines whether large index funds and pensions can hold the stock at size.
