Anthropic’s Potential $2 Trillion IPO Puts Its External Trustees to the Test

Anthropic’s possible move into the public markets is drawing attention not only because of the scale of its expected valuation, but also because of who would control the company’s board. The maker of Claude has created a Long-Term Benefit Trust, or LTBT, to protect its mission of developing AI for the long-term benefit of humanity. The trust holds no equity, yet it has the authority to appoint or remove a majority of Anthropic’s directors.

If the planned IPO reaches the upper end of current expectations, the company could be valued at roughly $2 trillion. That would bring an experimental governance model into an environment where shareholders, analysts and regulators are likely to demand clearer answers about profitability, accountability and decision-making authority.

Key points

• The LTBT can select or dismiss most of Anthropic’s board. It has chosen four of the company’s seven directors.
• The trust currently has three members, with room for as many as five. Its members include former Federal Reserve chair Ben Bernanke, Neil Buddy Shah of the Clinton Health Access Initiative and Richard Fontaine of the Center for a New American Security.
• Trustees receive advance notice of major actions, including the launch of new AI models. They meet weekly among themselves and regularly with executives, directors and founders.
• The group has discussed issues including the limited rollout of Anthropic’s Mythos cybersecurity model and the company’s dispute with the US government over automated weapons.

So far, however, the LTBT has largely operated as an advisory body. It has not yet had to impose a major constraint on Anthropic or force leadership to sacrifice a meaningful commercial opportunity for safety or another mission objective. That unresolved question is central to the credibility of the model.

Anthropic remains a loss-making company operating in an industry that requires enormous and continuing investment in computing capacity and talent. Private backers may have accepted the company’s safety-oriented structure while also assuming that it would eventually become a commercial powerhouse. Public investors could be less patient, especially if the company’s growth plans require large spending for an extended period.

Corporate-governance scholars see a structural tension. Capital comes from investors seeking returns, while decisions about how much profit to sacrifice for the mission may be made by external trustees with limited direct economic exposure. Elizabeth Pollman of the University of Pennsylvania has noted that no governance contract can anticipate every conflict that might arise, particularly in a sector shaped by intense corporate and geopolitical competition.

Anthropic’s design is viewed as less vulnerable than OpenAI’s governance system, which suffered a major crisis when its board attempted to remove Sam Altman in 2023 and subsequently lost the confidence of investors and employees. Anthropic also has a potential “kill switch”: trustees can be removed with support from shareholders holding 85 percent of the voting power. Yet that threshold and the broader structure could change as the company goes public.

The IPO will therefore be a test of more than Anthropic’s revenue prospects. It will show whether a mission-protecting body can retain meaningful authority inside a public company without creating paralysis or undermining investor rights. Success could make the LTBT a reference point for future AI governance. Failure could reinforce doubts that unusual control structures can reliably balance safety, public benefit and commercial pressure over time.

Source: Ars Technica AI