Anthropic's Planned IPO Tests Its Governance Trust

Introduction

Anthropic is preparing for a public offering that could value the company at as much as $2 trillion, according to a Financial Times report republished by Ars Technica on September 4, 2026. A listing at that scale would put fresh pressure on an unusual feature of Anthropic's governance: the Long-Term Benefit Trust (LTBT), a body of outside trustees that holds no equity in the company yet has the right to appoint or dismiss a majority of its board. The trust was built into Anthropic's structure as a public benefit corporation, intended to keep the company's stated mission from being overridden by short-term profit pressure. Anthropic says it plans to preserve the LTBT's role after going public. That plan raises a question with no settled answer: can an outside oversight body built for a private startup hold its authority once public shareholders are in the room?

How the Trust Is Structured

The LTBT does not own any part of Anthropic. Its influence comes entirely from one power: appointing and removing the majority of the company's board. The trust has used that power to select four of Anthropic's seven directors, including Netflix co-founder Reed Hastings and Novartis CEO Vas Narasimhan.

The trust has room for five members but currently has three. Its chair is Neil Buddy Shah, CEO of the Clinton Health Access Initiative. Former Federal Reserve chair Ben Bernanke joined the trust in July 2026. Richard Fontaine, CEO of the Center for a New American Security, also serves as a trustee. A seat was briefly held by Mariano-Florentino "Tino" Cuellar, a former California Supreme Court justice, who left the trust after a few months to become Anthropic's chief global affairs officer — moving from the body meant to oversee the company into an executive role inside it.

That leaves the trust two members short of its own five-seat ceiling. People close to Anthropic describe Bernanke's addition as a step toward making the trust a more institutional body. Their stated ambition, according to those people, is for the LTBT to eventually function the way GAAP accounting standards did: a set of voluntary rules that became an industry-wide reference point rather than one company's internal arrangement.

One safeguard in the design points toward investors rather than trustees. Trustees can be removed if shareholders holding 85% of voting power agree to it — a supermajority that makes a quick override difficult. That supermajority requirement could itself change once Anthropic's shares are distributed across many public holders instead of a smaller group of private investors.

What the Trustees Actually Do

In practice, the trust's role has been mostly consultative. Trustees receive advance notice of major company actions, including new AI model launches. They meet weekly among themselves, meet with Anthropic's leadership as often as every other week, and attend board meetings.

Two cases illustrate how that access has been used. When Anthropic developed Mythos, a cybersecurity-focused model, trustees encouraged a limited rollout through the company's Glasswing Project rather than a wider release. Trustees have also weighed in on Anthropic's dispute with the US government over automated weapons.

In both instances, the trust influenced the conversation without issuing a binding order. It has not drawn a hard red line, and it has not forced Anthropic into a decision that sacrificed meaningful profit for the sake of its mission. Because no conflict of that scale has yet occurred, the trust's real authority — what happens when advice and business interest genuinely diverge — remains unproven.

The Public-Market Test

Going public changes who the trust answers to in practice, even if its formal powers stay the same on paper. Harvard Law professor Jesse Fried, a corporate governance expert, has described the arrangement as a "built-in conflict." In a July 2026 paper covering both Anthropic and OpenAI, Fried wrote that each company "raises funds from profit-seeking investors, then lets self-appointed individuals decide how much profit to sacrifice for the firm's mission," calling this "a deep and potentially unmanageable tension ... hard-wired into the firms' corporate DNA." Fried advised investors to "scrutinize both companies' arrangements ... and price shares accordingly."

University of Pennsylvania law professor Elizabeth Pollman points to a different limitation, noting that governance design is inherently incomplete. "It's nearly impossible to perfectly contract for all possible circumstances that could arise when managing competing interests within a firm," she said. She framed the open question this way: "Will this governance structure work in the way intended, serving dual or more interests over time? That's the real challenge."

One venture backer, speaking about the broader bet investors have made, put it more bluntly: "There was a judgment made by investors that capitalism would win in the end. Whatever you say, if you need a lot of money for compute and to compete [for the best model], investors assume there will be a business." Private investors backed Anthropic with full knowledge of the trust's powers, and several have cited its safety emphasis as part of their investment thesis — but that thesis has not yet been tested against a public shareholder base demanding returns from a company that is currently lossmaking and working toward long-term sustainability.

Anthropic's structure is generally viewed as carrying less governance risk than OpenAI's, largely because of the 85% removal threshold. OpenAI's November 2023 board attempt to fire Sam Altman remains the reference case for what can go wrong when a nonprofit-style oversight body collides with commercial reality.

Pros and Cons

The LTBT's strengths are structural. It cannot be diluted the way an equity stake can, since it holds no shares. Trustees get early visibility into major decisions, including model launches, rather than after-the-fact disclosure. The 85% removal threshold sets a high bar against a quick shareholder override, and hiring a figure like Bernanke signals an effort to professionalize the body over time.

The weaknesses are just as structural. The trust is running at three of five seats, a gap that limits its capacity right as scrutiny increases. Its conduct so far has been advisory rather than binding, so its real leverage in a genuine crisis is unknown. And the departure of a trustee into an Anthropic executive role shows how thin the line can be between independent oversight and the company it is meant to watch.

Outlook

People close to Anthropic frame the LTBT as a prototype other AI companies could eventually adopt, comparable to how accounting standards started as one industry's internal practice before spreading. Bernanke's appointment fits that framing, suggesting Anthropic wants trustees with institutional credibility beyond the AI industry itself. Whether that ambition survives contact with public markets depends on factors outside Anthropic's control, including how index funds and activist shareholders respond to a governance layer with no ownership stake. Filling the two open trustee seats before an IPO would be one concrete signal of how seriously Anthropic is treating the gap Fried and Pollman have both pointed to.

Conclusion

The Long-Term Benefit Trust gives Anthropic an unusual governance mechanism, but it has never faced a real profit-versus-mission fight. An IPO valuing the company at up to $2 trillion is the kind of environment where that fight becomes likely. Whether the structure holds will be settled by how it performs under that pressure, not by how it reads on paper.

Editor's Verdict

Anthropic's Planned IPO Tests Its Governance Trust earns a solid recommendation within the Claude space.

The strongest case for paying attention: no equity stake means the trust cannot be diluted like a shareholder position. That alone raises the bar for what readers should expect in this space. Reinforcing that, trustees receive advance notice of major actions, including new model launches, before they happen — practical value rather than just headline appeal. The broader signal worth registering is straightforward: Anthropic plans to keep the LTBT's authority intact even after an IPO that could value the company at up to $2 trillion. On the other side of the ledger, one constraint is real rather than a marketing footnote: the trust operates at three of five possible seats, leaving it understaffed heading into an IPO. It should factor into any serious decision. Layered on top of that, its role so far has been advisory; it has not drawn red lines or forced a real profit-versus-mission trade-off — which narrows the set of teams for whom this is an obvious yes.

For Anthropic and Claude users, alignment-focused teams, and developers already invested in the Claude ecosystem, this is a serious evaluation candidate, not just a curiosity to bookmark. For everyone else, the safer posture is to monitor coverage and revisit once the use cases that matter to your team are demonstrated in the wild.