• NYC Comptroller Mark Levine urges Anthropic to postpone its IPO until AI safety and governance risks are addressed.
  • Levine oversees public pension funds with over $325 billion in assets and warns of severe investor damage from a post-listing AI incident.
  • Anthropic still targets a November IPO at a valuation above $2 trillion, following blowout quarterly revenue of $16.8 billion.

A Pension Fund Pushback

Anthropic’s plan for what could be the largest IPO in history has met an unexpected hurdle: a powerful pension fund official demanding that the AI developer hit the brakes. New York City Comptroller Mark Levine, who oversees public pension funds with more than $325 billion in assets, sent a letter to Anthropic on October 14 asking the company to delay its listing until it can demonstrate robust AI safety and governance protections. The intervention, first reported by Reuters (TRI), adds a prominent institutional voice to growing concerns that the breakneck pace of AI development is outpacing oversight.

Levine’s letter specifically calls for detailed safety audits, transparent controls, implementation deadlines, and an explanation of how these measures square with CEO Dario Amodei’s own public calls to slow advances in frontier AI. “Without concrete commitments, public investors would be exposed to risks that even the company’s own disclosures flag as potentially catastrophic,” Levine argued, according to people familiar with the matter. An Anthropic spokesperson did not immediately respond to requests for comment.

Blockbuster Growth Meets Safety Questions

The pressure comes as Anthropic’s financials are nothing short of spectacular. According to an October 14 investor presentation, the company generated $16.8 billion in third-quarter revenue, up about 45% from the prior quarter. That follows $11.6 billion in the second quarter and $4.7 billion in the first. The company also reported adjusted profitability for the third quarter, though it posted a nearly $4.6 billion operating loss for full-year 2025, excluding certain liability adjustments. To fuel its ambitions, Anthropic has committed at least $518 billion over a decade to cloud services, computing, and infrastructure.

Yet those numbers carry a hefty price tag and considerable risk. The company’s draft prospectus reportedly includes roughly 80 pages of risk disclosures, including warnings about potentially catastrophic AI outcomes. Levine’s letter argues that a major AI safety incident after listing could severely damage pension beneficiaries and other investors. The White House’s recently signed voluntary AI safety accord, which calls for internal controls and independent auditors but lacks enforcement teeth, is insufficient, Levine said in a CNBC (VSNT) appearance on October 15.

IPO Timing and Governance Tensions

Anthropic is reportedly targeting an IPO as early as November at a valuation above $2 trillion. If completed, the offering would test whether public market investors believe the company’s extraordinary revenue growth can outweigh the enormous costs and existential risks of advanced AI. The listing would also test a novel governance structure: a “Founder LLC” of seven co-founders would control a Class F share carrying 50.1% voting power over key matters, while Anthropic would remain a Delaware public benefit corporation and its Long-Term Benefit Trust would elect four board directors.

That structure aims to insulate long-term safety decisions from quarterly earnings pressure, but it also limits ordinary shareholders’ influence. Levine questions that accountability gap. “Regulatory stability and clear lines of responsibility are what institutional investors need before they commit capital,” he said, according to the letter. The comptroller’s intervention is not a regulatory halt, and Reuters still expects the IPO to proceed in November. However, the standoff highlights a growing debate over whether AI safety should be a measurable investment criterion rather than a corporate promise.

Broader Implications

The outcome could set a precedent for other AI listings. OpenAI is expected to go public in 2027, and how Anthropic resolves the tension between mission-driven control and public accountability may shape investor expectations for the entire sector. A delay could postpone access to public equity financing and dampen sentiment toward other AI offerings. Conversely, stronger safeguards and clearer governance could bolster confidence.

For now, all eyes are on Anthropic’s next move. The company has not yet published its IPO prospectus, and its response to Levine remains pending. Whether it accelerates or adjusts its timeline, the episode signals that even the most hyped AI companies face rising scrutiny over the risks embedded in their own technology.

Correction: An earlier version of this article misstated the third-quarter revenue figure. It was $16.8 billion, not $16.8 million.