Anthropic, the maker of the Claude AI models, has filed a more than 300-page prospectus for its planned initial public offering, the most detailed public look yet at how a pure-play AI lab makes money. The company plans to list Class A shares on the Nasdaq under the ticker ANT, and analysts expect a valuation of more than $1 trillion.
According to the filing, Anthropic booked $16.29 billion in revenue in the first half of 2026, up from $1.16 billion a year earlier, with cost of revenue of $2.38 billion, implying a gross margin of about 85%. Growth slowed in the third quarter, to roughly 47% over the previous quarter after about 144% in the second quarter. The company said it has "historically operated in a compute-constrained environment."

The structure has drawn the most debate. Chief executive Dario Amodei and six other co-founders each own more than 200 million shares, while the founders collectively control a single Class F share that carries 50.1% of the total voting power on shareholder matters. NYU finance professor Aswath Damodaran called the voting design "deviously self-serving."
Anthropic is organized as a public benefit corporation. Analysts said the filing leaves open questions about how much of its computing commitments will go to training new models versus serving customers, a split that will shape future margins.