Anthropic’s $2 trillion IPO puts powerful external trustees in spotlight

Ars Technica, drawing on Financial Times reporting, focuses not on the IPO headline number but on Anthropic's Long-Term Benefit Trust: an external body with no equity that can appoint or dismiss a majority of directors. The trust has already named four of seven board seats, including Netflix co-founder Reed Hastings and Novartis CEO Vas Narasimhan.

That structure is meant to survive a public debut. Trustees currently number three of a possible five — chaired by Clinton Health Access Initiative CEO Neil Buddy Shah, with former Federal Reserve chair Ben Bernanke and Center for a New American Security CEO Richard Fontaine. They get advance notice of major moves, including new model launches, meet weekly among themselves, and meet leadership as often as every other week.

The commercial stakes are blunt. Anthropic is loss-making and heading toward markets where shareholders will demand a path to profit. The trust's job is to keep the public-benefit mission intact under that pressure. Whether it can actually force a hard trade-off has never been tested.

Signal Desk's read: the $2 trillion valuation talk is bait; the governance design is the real product being sold to — and stress-tested by — public markets. Anthropic wants the LTBT treated like a proto-GAAP for AI: a voluntary template that becomes industry infrastructure. Bernanke's July appointment is a credibility signal for that ambition. Ambition is not evidence the mechanism works when money and mission collide.

Reporting close to the trust says it has so far operated largely as an adviser. It encouraged a limited Mythos cybersecurity rollout through the Glasswing Project and weighed in on the company's dispute with the U.S. government over automated weapons. Useful access. Not a red-line veto. Harvard Law's Jesse Fried calls the setup a built-in conflict: profit-seeking capital funding a firm whose self-appointed guardians can decide how much profit to sacrifice for mission. That tension is not a bug in the PowerPoint; it is the operating system.

Compared with OpenAI's 2023 board crisis, Anthropic's design looks less brittle on paper. Trustees can be removed with 85 percent shareholder voting support — a "kill switch" that may change after listing. Early private backers accepted the structure and, in some cases, treated safety branding as part of the thesis while assuming capitalism would still win because compute bills force a real business. Public investors will be less sentimental.

The likelier outcome is not sudden trustee heroics or a clean OpenAI-style collapse. It is pricing: shares that embed a governance discount until the LTBT either constrains a commercial push or proves it will not. Until a profit-versus-purpose fight is visible, treat "mission-preserving IPO" as branding, not a demonstrated control system.

Anthropic is a public benefit corporation whose unusual trustee model sits alongside OpenAI's nonprofit-rooted structure as Silicon Valley's main experiment in mission guardians outside ordinary fiduciary boards. OpenAI's November 2023 attempt to fire Sam Altman — and the subsequent board overhaul — remains the cautionary case investors will map onto any AI IPO prospectus.