Tech BusinessLOT 261119 · 4 min read
Anthropic's $81.3 Billion IPO Tests Its Safety-First Promise
Anthropic's record $81.3B IPO values it near $1.5 trillion, but its Long-Term Benefit Trust lacks the authority experts say is needed to check founder power as commercial pressure mounts.
Key takeaways
- Anthropic raised $81.3 billion in the largest IPO in history, beating SpaceX's record, at a ~$1.5 trillion fully diluted valuation.
- The Long-Term Benefit Trust names four of seven board seats but lacks day-to-day or model-release authority per Anthropic's Form S-1.
- Trustees include Ben Bernanke, Richard Fontaine, Christopher Stubbs and chair Neil Buddy Shah — none known as AI experts.
- Co-founder Daniela Amodei chairs the nominating committee alongside Reed Hastings and investor Yasmin Razavi.
- Dissent includes a February safeguards-research departure, a September Center for AI Safety statement, and a $81.3B capital need driven by compute costs.

Anthropic raised $81.3 billion in the largest IPO in history, beating SpaceX's record for the most valuable public debut ever — and putting the AI lab's carefully cultivated safety reputation to its hardest test yet.
The offering values Anthropic at a fully diluted market cap of roughly $1.5 trillion, and it sets a precedent for every public AI lab debut that follows. But the money arrives with a paradox baked in: a corporate governance structure designed to shield the company from shareholder pressure also concentrates near-unilateral power in the hands of its founders, according to outside experts and a former employee.
What does the Long-Term Benefit Trust actually control?
Anthropic registered as a Public Benefit Corporation, obligating its board to balance mission against shareholder wealth. Its Long-Term Benefit Trust (LTBT), introduced in 2023, holds a special Class T stock giving trustees "authority to select and remove a portion of" the board, with a stated mission to "balance the financial interests of our stockholders and our public benefit purpose at key junctures."
The fine print is murkier. The LTBT technically chooses four of seven board seats, but one of those picks effectively requires founder approval — leaving the trust short of a clean majority. Any decision to nominate or remove a director requires 30 days' notice to a nominating and corporate governance committee stacked with insiders: co-founder Daniela Amodei as chair, Netflix co-founder Reed Hastings, and Anthropic investor Yasmin Razavi.
Nothing bars Anthropic from hiring a trustee outright. Mariano-Florentino Cuéllar, a former trustee, joined the company in August as Chief Global Affairs Officer. "If a trustee was looking a little bit too activist-y or something, they could just make that trustee an offer with millions of dollars in effective compensation," said Tyler Johnston, founder of watchdog The Midas Project.
Ann Lipton, a corporate governance expert at the University of Colorado Law School, questions both the trust's leverage and its expertise. "The trust, since it doesn't have a financial interest in Anthropic, could theoretically use its power with three slash four directors to keep the company on mission," she said. "One might question how likely that is to occur." The four listed trustees — former Fed Chair Ben Bernanke, foreign policy analyst Richard Fontaine, experimental physicist Christopher Stubbs, and chair Neil Buddy Shah — aren't known as AI specialists. "They're very prestigious people, but what does Ben Bernanke know about AI?" Lipton asked.
How does this compare to OpenAI's board?
Several experts drew parallels to OpenAI's nonprofit board before Sam Altman's ouster — a body that intervened, then watched Altman return days later while the board resigned. But the comparison flatters OpenAI's structure, not Anthropic's. OpenAI's nonprofit retained authority over day-to-day decisions and, even after restructuring, kept the power to delay or stop a model release on safety grounds. Anthropic's Form S-1 shows the LTBT has no equivalent authority; its role is largely appointing and removing certain directors.
"Neither the Trust or the trustees have any decisionmaking authority over day to day decisions," said Hamza Chaudhry of the Future of Life Institute. "So I do not have strong hopes for the LTBT being a key mechanism for governance after they go public."
The setup more closely resembles founder-centric structures like Mark Zuckerberg's Meta or Elon Musk's SpaceX. Anthropic declined to comment.
Where does the money go?
Compute. "Compute is the deepest business dilemma that the company faces," Amodei said last Friday in a livestreamed conversation with JPMorgan Chase CEO Jamie Dimon. "If you buy too much, of course you can financially overextend yourself. If you buy too little, then you can't support inference for customers, and even more importantly … you can't do the R&D."
The company frames the race explicitly. "You can't do safety from second place," policy chief Sarah Heck said in September. Anthropic, like its rivals, is pursuing recursive self-improvement — AI systems that autonomously build better versions of themselves — despite Amodei's own warning that "left unchecked, [RSI] could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all."
Internal dissent has been building. Anthropic's head of safeguards research left in February alleging "the world is in peril." Researcher Jacob Coxon claimed in September that Anthropic and OpenAI are "racing straight to self-improving superintelligence and gambling with our lives." Another safety researcher departed for auditor METR that month, writing that "AI companies are on track to impose an unprecedented level of risk on society" and that Anthropic avoiding a severe incident so far was "partly a matter of luck."
On Wednesday, the Center for AI Safety published a statement with signatories from Hugging Face and Harvard University alleging Anthropic "uses safety branding to court investors, employees, and the public — even as it has quietly weakened its safety pledges."
A former employee who left before the IPO put it bluntly: private companies are building a new form of intelligence "with less oversight, in many ways, than taco trucks currently have." Now, with $81.3 billion raised and public-market expectations attached, the pressure to justify spending on safety over speed — and to keep buying compute — only intensifies.
theverge.com (Source)
More from Rebecca Stone
Show full bio
Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,170 articles