Anthropic's IPO: a valuation of up to $2 trillion that its own revenue doesn't explain

By Zendoric · October 3, 2026.

Anthropic has taken its first formal step toward Wall Street. On June 1, 2026 it confidentially filed with the SEC —the US securities regulator— the draft of its S-1 form, the document that opens the door to going public without yet obliging it to do so, as the company itself announced. The Wall Street Journal reported on September 18 that the target date had moved from October to November 2026; Reuters had earlier pointed to mid-October. Neither is an official date, and both could change again.

The underlying numbers are already on the table. In May, Anthropic closed a $65 billion round that valued it at $965 billion, and the company said at the time that its "run-rate" revenue —the annualized projection obtained by multiplying the latest month of billing by twelve, not an audited figure— had passed $47 billion. That puts the valuation at around 20.5 times that revenue. Recent reports gathered by Reuters now cite an IPO figure of up to $2 trillion, which would amount to 42.6 times revenue that, by then, will probably also have grown. It is, in any case, a considerable jump in multiple even for a sector accustomed to aggressive valuations.

The enterprise story Anthropic wants to sell investors is not just Claude's growth in consumer, but its deployment inside large companies: in May it announced a joint venture with Blackstone, Hellman & Friedman and Goldman Sachs expressly devoted to embedding Claude in corporate processes. That is the argument it needs to justify such a multiple: selling software to developers is not enough, it has to show that enterprise revenue is recurring and defensible against competition from OpenAI and the big Chinese models that, as we have been reporting, close the quality gap with every passing quarter.

What will really matter, however, is not the date but the content of the public S-1 when it arrives: operating losses, gross margins, compute spending, cash generation and customer concentration. A company can grow very fast and still be expensive if sustaining that growth requires burning enormous amounts of capital on data centers; that is, in fact, the norm among frontier AI labs, not the exception. Other coverage of this same move, such as Seeking Alpha's, sums it up with a line we share: Anthropic's IPO is, at bottom, a bet on how far AI can go, not merely a financial transaction.

Our read is that this listing is the next link in a financing chain we had already been flagging at Zendoric in connection with Anthropic's infrastructure deals with its compute providers, or with the stake Alphabet books in its accounts for its investment in the lab: AI capex is increasingly financed through cross-holdings among suppliers, customers and investors, so that one party's valuation depends on another's. Opening up to the public markets adds one more layer to that web, and makes it public: that is good for transparency —it forces the disclosure of losses and margins that private rounds kept hidden— but it also multiplies the risk of contagion if the market decides, at some point, that these multiples do not hold up.

For the retail investor there is also a diversification trap: anyone who already holds Amazon, Alphabet or Microsoft —all three with stakes or infrastructure alliances in the AI sector— is already exposed to this same bet. Adding Anthropic shares does not diversify, it concentrates. It is a nuance that SmartAsset, as an outlet specializing in financial planning, rightly underlines, and one worth keeping in view amid the enthusiasm over the IPO date.

In the short term, then, the honest thing to say is that this deal will be judged on figures, not narrative: if the S-1 shows that Claude's enterprise business generates real margins and not just growth financed with debt and giant rounds, the $2 trillion valuation will begin to look reasonable; if not, it will be another reminder that the capital sustaining the AI race moves faster than the profitability that justifies it. In the long run, however, companies like Anthropic gaining access to public capital markets is ultimately good news for the underlying thesis we defend: the more money —and the cheaper— that can be channeled into the compute and research that make more capable models possible, the sooner we will reach the point where that capability translates into medicines, diagnoses and real abundance, and not just multiples on a spreadsheet.