THE CRUNCH
Anthropic has filed its S-1 to go public on the Nasdaq, with trading possibly starting as soon as next month at a valuation somewhere between $1.5 trillion and $2 trillion. That would immediately place it among the most valuable companies on the planet. The filing, hundreds of pages long, offers the first detailed look at the economics of a frontier AI lab, and it rests on a series of very ambitious assumptions about what AI will do to the future.
On the Decoder podcast, SemiAnalysis analyst Joey Brookhart, whose firm built an outside model of Anthropic's finances in July, said the filing showed a company that has scaled efficiently, with more than $70 billion of annualised recurring revenue and healthy gross margins, compute margins and free cash flow. He suggested the numbers helped drive a rally in semiconductor and compute stocks in the days after the filing.
The catch is concentration. Brookhart noted that the two leading labs together sit at around $120 billion of annualised recurring revenue, drawn from a very narrow market: the top one percent of customers, some spending $50,000 to $100,000 per employee, with much of Anthropic's business tied to coding. Reaching the trillion-dollar revenue ambitions Anthropic's leadership has floated would, in his view, require broader white-collar adoption, where AI shifts from augmentation to replacing labour in some areas.
That transition is the open question, both for this filing and the wider industry. Enterprise budgeting cycles have not yet meaningfully included these labs, and competition is thickening, with low switching costs in parts of coding and enterprises wary of handing data to a single lab exploring on-premises options.