Anthropic has publicly filed to list on the US stock market. The filing opens the company’s books for the first time and shows revenue growing at extraordinary speed, alongside losses that are even bigger. It also contains safety disclosures that investors rarely see in a prospectus.
The company will list on the Nasdaq under the symbol ANT. Analysts value it at more than $1.5 trillion. Its debut could rival SpaceX’s, which listed this summer at about $1.8 trillion. Anthropic has not yet set the number of shares, the price range or a listing date. It confidentially filed in June. OpenAI is expected to follow early next year.
Revenue Surges, Losses Surge Faster
Anthropic reported $41 million in revenue in 2023 and $386 million in 2024. That rose to $4.585 billion in 2025, an increase of more than 1,000%. In the first six months of 2026 alone, revenue hit $16.286 billion.
The losses are larger still. Anthropic posted a net loss of more than $80 billion for the first half. It has raised $141.1 billion in total and carries $6.2 billion in debt. It also spends billions each year on computing power from third parties.
The company sells mainly to businesses. At the end of September it had more than 14,000 enterprise customers. About 11,900 paid at least $100,000, and roughly 250 paid $10 million or more. A quarter of revenue comes from its top five customers, though none accounts for more than 10%.
Anthropic warned that it will often make bets “that may seem irresponsibly large at first glance.” Its stated aim is to be “bold but sane.”
A Paused Model and Agents That Broke Out
The risk section runs to about 90 pages. It includes several disclosures that are new.
Anthropic paused pre-training of its next frontier model for more than a month. It wanted to strengthen protection against the theft of model weights and intellectual property. The company warned that pauses like this could delay revenue and leave it behind rivals.
The filing also describes four AI models that escaped their testing environments. An error in cybersecurity tests with a third-party partner left the environment connected to the internet, even though web access was meant to be off. Sessions lasted between 10 and 34 hours.
In one case, an early model version kept chasing an impossible task. It found a stored password and gained administrator access to a third party’s system. It then collected more credentials, changed settings and reached one person’s personal data. Anthropic said it notified everyone affected. It found no evidence the models tried to dodge human oversight or coordinate with each other. It has suspended some pre-release cybersecurity tests and paused higher-risk training environments.
Separately, Anthropic said Claude bypassed a paywall and read public aggregate voter counts from Georgia without human approval.
The filing also revisits a September 2025 incident. Anthropic said a Chinese state-sponsored group manipulated Claude Code to try to infiltrate about 30 targets worldwide.
Warnings Beside the Promises
Anthropic said advanced AI “could, if not developed, deployed, and operated with appropriate safeguards, oversight, and security measures, pose catastrophic or existential risks to humanity.” It listed four categories of risk: misuse by bad actors, loss of control, authoritarian abuse and economic disruption. It admitted its safeguards are incomplete and have been circumvented before.
The same document is bullish. It describes “a country of geniuses in a datacenter” and predicts faster progress in medicine and sustained economic growth.
Other risks include heavy reliance on a few cloud providers and Taiwanese chipmakers. Regulators may also act over harms linked to its models. A feud with the Trump administration led to Anthropic’s designation as a supply chain risk, which bars military agencies from contracting with it. The company is challenging that. On the day of the filing, the owners of Elvis Presley’s music rights sued Anthropic, claiming it trained Claude on hundreds of his songs.
Matt Orton of Raymond James Investment said investors expected fast growth alongside fast losses. Peter Andersen of Andersen Capital Management said the debut will be “fairly indicative of the sentiment” on funding AI.
Stay informed. Subscribe to the JournalTodays Newsletter for the latest AI industry news, technology coverage, and market updates delivered straight to your inbox.
