Anthropic Files for a Nasdaq Listing, Revealing an $80.7 Billion First-Half Loss

Anthropic has filed its S-1 registration statement with the Securities and Exchange Commission and plans to list on the Nasdaq under the ticker symbol "ANT." The Verge reported the filing on October 26, 2026.
The filing reports $16.3 billion in revenue and an $80.7 billion net loss for the first half of the year. For comparison, Reuters reported in late September that Anthropic lost $42 billion in 2025 and that its operating loss had widened to $8.06 billion. Reuters also reported that the IPO could value the company at more than $2 trillion.
The path to the public filing ran through a confidential submission. Anthropic confidentially filed with the SEC in June, ahead of the S-1 now on the public record. A confidential filing means regulators review the draft before the public can see it. Timing reports moved over the late summer and early autumn. In late August, Reuters reported that the company planned to unveil its prospectus after Labor Day and had targeted a listing in late September or early October. In early September, Reuters reported that marketing of the offering was expected to begin in mid-October at the earliest. On October 21, Yahoo Finance reported, citing three people familiar with the matter, that the company was expected to publicly file its S-1 as soon as that Friday. The public filing has since come, roughly a month behind the earliest of those targets.
Reuters also reported that Anthropic plans to warn prospective IPO investors that advanced AI could pose catastrophic or existential risks to humanity. Risk language is a standard feature of a risk-factors section, the part of a prospectus that lists what could go wrong. In this case, however, it is written by a frontier model developer about its own core product, and it will sit alongside the revenue and loss figures in front of institutional buyers.
The broader context here is the shape of the numbers. A first-half net loss of $80.7 billion against $16.3 billion in revenue is a loss profile that depends on continued access to capital and on investors accepting a long horizon before margins turn. Compute commitments, meaning the cost of the processing power used to train and run models, are the obvious drivers. Anyone modelling the company will want the cost of revenue, the depreciation schedule on AI accelerators (the specialised chips that do the computation), and the contractual obligations once they are read in full, rather than the headline totals.
The valuation reported by Reuters, above $2 trillion, would place a company founded in 2021 in a bracket that few public software or semiconductor companies occupy. That comparison matters less than the question of what the market is being asked to price: a revenue line growing quickly against losses that scale with it. Public-market investors have priced AI infrastructure before, and the cloud buildout of the last decade offers a useful reference for how long capital intensity can persist before it eases. The difference is the scale of the cash burn now disclosed in a single document.
In my view, the listing matters most to the people who build on the company's tools. Once public, Anthropic will report results every quarter and answer to a board accountable to shareholders. Those obligations can shape pricing, product cadence and how much loss the company is willing to absorb on any given tier. For the wider market, the filing offers the first detailed look at the financial structure of a frontier lab operating at this scale, and competitors and customers will read it closely.
The ticker "ANT" is short. For a company whose products reach developers through APIs (application programming interfaces, the connection points that let other software use a model) and enterprise contracts, the listing will test whether public investors value the model layer differently from the compute layer beneath it. That answer will emerge over the coming months, as the offering is marketed and priced and the full prospectus is read line by line.


