New York City Pension Fund Challenges Anthropic’s $2 Trillion IPO Over AI Governance
New York City Comptroller Mark Levine has urged artificial intelligence developer Anthropic to postpone its planned U.S. initial public offering until the company addresses concerns about AI safety, corporate governance and the pace of technological development.
The intervention, outlined in an Oct. 14, 2026, letter to Anthropic CEO Dario Amodei and reported publicly Oct. 15, comes as the San Francisco-based company prepares for a potential November listing at a reported valuation exceeding $2 trillion. Levine oversees investment responsibilities involving New York City’s public pension funds, which collectively hold more than $325 billion in assets.
New York Pension Funds Question Anthropic’s IPO Readiness
Levine’s intervention connects institutional investor oversight with concerns about the risks of advanced artificial intelligence.
New York City’s pension funds have approximately $7 million in indirect exposure to Anthropic through private equity investments, according to the comptroller’s office.
In his official letter to Anthropic CEO Dario Amodei, Levine requested greater transparency concerning independent audits, operational safeguards and the company’s approach to controlling increasingly capable AI systems.
He also questioned whether Anthropic’s existing governance arrangements provide sufficient accountability to prospective public shareholders.
Anthropic operates as a Delaware public benefit corporation and maintains a Long-Term Benefit Trust intended to support its broader public-interest commitments.
Levine raised concerns about potential concentrations of decision-making authority among founders and trustees, particularly if the company’s proposed public-market structure limits shareholder influence.
AI Safety Disclosures Raise Governance Questions
The comptroller cited Anthropic’s own warnings about risks associated with increasingly autonomous AI models.
Those concerns include cybersecurity vulnerabilities, manipulation of information and potential misuse of advanced systems.
Levine also referred to reports that Anthropic’s draft IPO documentation contains extensive disclosures concerning catastrophic AI risks.
The letter does not establish that these potential outcomes will occur. Rather, it argues that investors need clearer information about how Anthropic intends to identify, mitigate and govern such risks.
For prospective institutional shareholders, the central issue is whether the company’s internal oversight mechanisms are sufficiently robust for public-market accountability.
$2 Trillion Valuation Raises the Stakes
Anthropic’s potential IPO comes amid rapid commercial expansion and substantial investor expectations.
At a recent pre-IPO investor meeting, the Claude developer reportedly disclosed third-quarter revenue of approximately $16.8 billion, up 45% from the previous quarter’s $11.6 billion.
The company has also reportedly committed hundreds of billions of dollars toward computing infrastructure over the coming decade.
Such commitments make capital requirements, profitability and operational resilience important considerations for prospective shareholders.
The reported $2 trillion valuation remains a potential IPO target, not a confirmed public-market valuation.
What Institutional Investors Should Watch
Levine’s request is not a regulatory order and does not independently prevent Anthropic from proceeding with its listing.
However, it introduces an additional governance issue for pension funds, venture capital investors and other institutions evaluating exposure to frontier AI companies.
Anthropic has previously described its Long-Term Benefit Trust governance structure as a mechanism for balancing shareholder interests with broader public benefits.
Investors evaluating the proposed offering will need to assess the relationship between board independence, shareholder voting rights, safety oversight and financial performance.
A delay could affect the timing of liquidity opportunities for existing private investors, while additional disclosures could influence institutional participation and valuation expectations.
Anthropic had not publicly confirmed a response to Levine’s request in the reporting reviewed.
