Subscribe
News
2 min read

Anthropic's Earliest Backer Says the Frontier Should Slow Down

Dustin Moskovitz, a board observer and one of Anthropic's first investors, told the NYT he invested 'to promote safety, not racing' as Wall Street hands the lab $81.3 billion.

Figure raising an open palm against a speeding machine built from a rising bar chart, Olivetti brochure style
Figure raising an open palm against a speeding machine built from a rising bar chart, Olivetti brochure style

Dustin Moskovitz was one of Anthropic's earliest investors, and the bet is expected to be worth billions when the company begins trading Thursday at an IPO valuation near $1.5 trillion. The Facebook co-founder is not celebrating.

In a statement to the New York Times, Moskovitz, who served as a board observer at Anthropic, a nonvoting seat, distanced himself from what the offering represents for the industry: "Cari and I invested in Anthropic to promote safety, not racing. AI is moving too fast. Anthropic and everyone else on the frontier should slow down and stop racing."

The safety arbitrage, repriced

Moskovitz's position is the most explicit statement yet from a financial winner of the safety-first thesis. He donated early to OpenAI and fell out with its leaders as the lab commercialized. Anthropic was founded by Dario Amodei and other OpenAI alumni precisely on the premise that a safety-priority lab could out-earn the race, and its long-term benefit trust and public benefit corporation structure were designed to make that commitment durable through a public listing.

The tension is now on the balance sheet. Wall Street is handing Anthropic more than $80 billion of new capital after an upsized IPO priced at $83 per share, raising $81.3 billion, the largest offering on record. The company's stated mission hierarchy asks it to restrain the pace of frontier development; its investor base, including one of its earliest backers, is asking for the same restraint and holding the opposite lever.

The NYT report notes that Moskovitz has repeated the worry across social media posts in recent days, framing Anthropic and other frontier labs as advancing faster than society can control, even as the IPO machinery accelerates them further.

A figure raising a hand against an oncoming data train

Why an investor's dissent matters here

Public benefit corporations can be criticized as branding, but Anthropic's structure gives dissenting shareholders an actual venue: the long-term benefit trust oversees the safety mandate and can, in principle, weigh release decisions against revenue pressure. Moskovitz's statement is a preview of what that argument sounds like from outside the building, from someone whose money is now exposed to the outcome in both directions.

Whether anything changes depends on the comparison that Anthropic's rivals keep drawing it into: a lab that slows down at the frontier cedes capability and market share to labs that do not. The lab's own employees, per separate NYT interviews published ahead of the debut, describe muted enthusiasm and internal debate about exactly that pace. An IPO resolves that debate in one direction by default, with fresh capital and quarterly eyes on it.

Moskovitz cannot force the other direction. He can only say, loudly, on the eve of listing, that this is not what he bought.

Sources

News