'Pause the IPO': NYC Pension Boss to Anthropic
The man who watches over the retirement money of New York's firefighters, teachers and police officers has read Anthropic's IPO paperwork. His verdict: do not list this company. Not yet. Not until the artificial intelligence it builds stops looking, by its own description, like a hazard.
That is the thrust of a letter New York City Comptroller Mark Levine sent to Anthropic chief executive Dario Amodei this week, urging the AI lab to pause its march toward a stock market debut that could value it at more than $2 trillion. CNBC first reported the letter's contents on Thursday; Reuters independently reviewed the letter the same day.
For months, the conversation around Anthropic's listing has been about the numbers — a $16.8 billion quarter, bankers whispering about a raise above $100 billion, a valuation that would make it one of the most valuable companies on Earth. Levine has changed the subject. His question is not whether the IPO will work. It is whether it should be allowed to.
Photo: Anthropic CEO Dario Amodei at 10 Downing Street, May 2023. This week, NYC Comptroller Mark Levine quoted the CEO's own "slow down" essay back at him. Credit: Simon Walker / No 10 Downing Street, CC BY 2.0 via Wikimedia Commons.
The letter, in his own words
Levine did not ask politely. The letter opens, according to CNBC's account, with a flat request: "I write to urge Anthropic to pause its planned initial public offering until the company can fully and transparently address the artificial intelligence safety, governance, and development pace concerns it has raised."
Then came the line that will follow Amodei around: "Prudence should not only be shown with respect to the pace of the frontier, it should be also shown with respect to the timing of an IPO," Levine wrote, as quoted by Agence France-Presse. He finished the thought with a direct command: "I therefore urge Anthropic to pause the IPO process so that the company may fully address these key issues, ones which investors must also assess."
On CNBC on Thursday, Levine said Anthropic and rival OpenAI both needed to get their safety under control, and that the companies' own filings admitted how much work remained. "I see no reason to rush into this," he said, as relayed by Seeking Alpha. "They should slow down, focus on safety, and go public when they can ensure investors."
It is an unusual beat for Levine: he is the city's fiscal watchdog, managing the pension funds of more than 750,000 people, with assets that totalled $330 billion at the end of August, according to AFP — one of the largest pools of retirement money in America.
Amodei's own essay, thrown back at him
What gives the letter its sting is where Levine sourced his ammunition. The comptroller effectively raided Anthropic's own files. Central to his case is an essay Amodei published on September 12 — "We Must Pace the Frontier" — a roughly 3,800-word call for the industry to slow down the development of advanced frontier models. Amodei has spent the autumn preaching restraint; Levine is asking him to live it.
The timing of the essay makes for shoddy optics, and Levine appears to know it. His letter pointed to the essay's publication coming just after the September 9 resignation of Jacob Coxon, a former Anthropic researcher who left citing existential concerns about the company's direction — a departure that was widely reported last month. The letter also leans on Anthropic's own prospectus, which warns that advanced AI could pose catastrophic or existential risks, with risk-factor disclosures that one report on the letter's contents puts at roughly 80 pages.
The letter also dismissed the industry's recent safety accord with the White House — signed by Anthropic and five others — as "insufficient to meet the moment," as reported. A handshake with Washington does not substitute for fixing the thing.
Why $7 million moves a $2 trillion listing
On paper, Levine's leverage looks comical: roughly $7 million in Anthropic exposure through private equity holdings — a rounding error against $330 billion. But the letter is not about the $7 million. It is about what happens the morning after the listing.
Once Anthropic goes public, its shares will be swept into the major stock indices, including the Russell 3000, which New York City tracks for its own investments. At that point, as the comptroller's office explained to AFP, the funds would "automatically become long-term investors when the company is integrated into the index fund." Every city pensioner would own Anthropic by default, through the index, with no way to opt out. A pension boss cannot watch that happen without asking what exactly is being bought.
The argument has a cold logic. If Anthropic's models really carry the catastrophic risks its own prospectus describes, then an index fund holding them is not a passive investment. It is a bet that the risks get managed before they get monetised. Levine is asking, in effect, whether the retirees are consenting to that bet. They are not, and cannot.
Anthropic's busiest — and worst-timed — week
The letter landed in the middle of the most choreographed week in the company's history. On October 14, Anthropic held a pre-IPO investor day in San Francisco, where — as we reported — executives let prospective investors interrogate them ahead of a listing that Bloomberg says could begin formal marketing the week of November 9. Days earlier, the company had disclosed a third quarter that stunned even the optimists: $16.8 billion in revenue, up 45 percent on the prior quarter, and a first-ever quarterly profit on an adjusted basis.
The week did not go entirely to plan. Reuters reported on Thursday that some investors who attended the October 14 event were quietly discussing a valuation closer to $1.5 trillion — down from the $2 trillion target — and that others were underwhelmed that revenue growth had slowed to 45 percent after more than doubling the quarter before.
Anthropic did take one safety step this week. On October 15, the company announced it had given six researchers from METR, a nonprofit that stress-tests frontier AI models, hardware, office space and internal-tool access roughly on par with its own risk-assessment teams. METR can publish its findings, and Anthropic cannot veto them, though it can demand redactions in narrow categories like customer data. A genuine concession to transparency — and precisely the kind of thing Levine's letter implies is not enough.
Photo: NYPD officers in front of the Charging Bull on Wall Street, New York. Once Anthropic lists, NYC's pension funds would automatically own it through the Russell 3000 index — no choice, no opt-out. Credit: Bruno Sanchez-Andrade Nuño, CC BY 2.0 via Wikimedia Commons.
OpenAI just did this — voluntarily
Here is the detail that makes Levine's letter more than a lone crank's complaint: Anthropic's closest rival has already done exactly what he is asking. In a Fortune interview last month, OpenAI chief executive Sam Altman said the company would not go public in 2026. "I actually think that, given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that," Altman said. Pressed on whether that meant 2027, he answered simply: "I would say not 2026."
Reuters' own reporting has OpenAI expected to list in 2027. The role reversal is striking. For years, OpenAI was the speed-obsessed one and Anthropic the safety-first lab founded by exiles who thought OpenAI moved too fast. Now OpenAI is the one hitting the brakes while Anthropic sprints for a Thanksgiving listing. If you wanted a symbol of how much the industry has tilted, Levine's letter is it.
Meanwhile, prosecutors in eight US states sued OpenAI over child safety this week, alleging the company concealed risks and failed to protect children. Anthropic, for its part, did not respond to Reuters' requests for comment on Thursday — and our Friday-morning checks found no public response from the company since.
The uncomfortable arithmetic
Strip the letter to its bones and you get a valuation question. Anthropic is asking public investors — including, automatically, 750,000 New York pensioners — to pay up to $2 trillion for a company whose own draft prospectus warns that its products could cause catastrophic harm, and which has some $518 billion in future cloud and compute commitments. Bankers say the offering could raise more than $100 billion. Levine's argument is that the prospectus reads less like an investment case and more like a confession — and that listing it anyway, at the largest valuation ever attempted, is botched governance.
Levine's position is not anti-AI but pro-diligence: this IPO is structured so the safety work and the listing run in parallel, with the money landing before the assurances. He wants the order reversed — assurances first, listing after.
Our verdict: the letter will not stop the IPO. It will stain it.
Mark Levine cannot block the IPO; he commands no regulator. The listing machine will grind on. But colossally inconvenient facts have a way of sticking to a prospectus, and Levine has now planted them in the official record, weeks before marketing begins. Every banker on the deal will know it exists. If the IPO later hits trouble — a model incident, a safety scandal, a Coxon-style resignation at the wrong moment — this letter becomes the exhibit. That is the shoddy-risk bargain Anthropic is being asked to reconsider: not whether it can list, but whether listing at $2 trillion, with its own risk warnings quoted back at it by a $330 billion fiduciary, is the victory it looks like. Our read: Levine is asking the only question nobody on the roadshow wants answered.
Your questions answered
Who is Mark Levine?
Mark Levine is the Comptroller of New York City — the city's chief fiscal watchdog. The office oversees municipal finances and manages the pension funds of more than 750,000 city workers, teachers, police officers and firefighters, with about $330 billion in assets.
What does the letter actually ask for?
Levine urges Anthropic to "pause the IPO process so that the company may fully address these key issues, ones which investors must also assess" — the AI safety, governance and development-pace risks the company itself has flagged. He is not asking for the IPO to be cancelled, only delayed until the risks are addressed.
Why does a $7 million investment matter?
Because the $7 million is about to grow automatically. Once Anthropic lists, its shares enter indices like the Russell 3000 that NYC's funds track, making the pension funds involuntary long-term holders. Levine is acting as the fiduciary for retirees who will own the stock whether they like it or not.
Has Anthropic responded?
Not publicly, as of Friday morning. Anthropic did not respond to Reuters' requests for comment on Thursday, and our own Friday-morning checks found no company statement on the letter. We will update this article if one arrives.
When is the Anthropic IPO expected?
Media reports point to November 2026 — Bloomberg says formal marketing could begin the week of November 9, with trading before Thanksgiving, after the US midterm elections. Anthropic has not confirmed a date, and a target valuation of over $2 trillion has only been reported, not announced.
Sources: Reuters (reporting by Manya Saini and Utkarsh Shetti, October 15, 2026); Agence France-Presse via Daily Sabah (October 15, 2026); CNBC (first report of the letter, via Reuters and Seeking Alpha relays); Bloomberg (IPO timeline reporting, October 2026); Bloomberg Law; Fortune via PYMNTS and T2Online (Altman interview, September 2026); AIStockWire and CryptoCompass for letter-context and METR reporting. Quotes from the letter are as reported by CNBC, Reuters and AFP; the company has not published the letter itself.
About the author: TechPulse USA's news desk covers the AI industry's collisions with law, money and power — from courtroom filings to credit markets — with every fact checked against primary sources before publication.
Comments
Post a Comment