Anthropic Files 335-page IPO prospectus with SEC and flags AI risks

Anthropic filed its S-1 prospectus with the SEC on Oct. 26, 2026, ahead of what could become the largest IPO ever. Crunchbase News reported the document runs 335 pages. PCWorld reported the filing landed on Monday and runs more than 300 pages.

Crunchbase News reported Anthropic‘s revenue for the first half of 2026 reached $16.3 billion, a 14-fold increase over the same period in 2025. The S-1 contains no revenue forecast for the second half of 2026. Analysts who met with company executives told Reuters that Anthropic might hit a $110 billion annualized revenue run rate by the end of 2026.

Anthropic recorded a $693 million loss from operations for the first half of 2026, a decline of roughly two-thirds compared with the same period in 2025. Its net loss for the first half of 2026 reached $80 billion, with about 90% tied to fair value changes on convertible notes, derivative liabilities and SAFEs. The filing lists Class A stock, convertible notes, derivative liabilities and SAFEs among its securities.

PCWorld separately reported an $80.7 billion loss for last year. It also reported first-half revenue of $16.3 billion from the prior year, up 14 times over the year before.

Anthropic’s co-founders are Dario Amodei, Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Christopher Olah. Each founder holds from 230,000 to 246,000 shares of Class A stock per the S-1.

Anthropic warned that slower revenue growth might limit its ability to fund frontier AI development and to attract and retain top talent. It also expressed concern about servicing significant debt while covering operating costs. Anthropic’s risk factors fill 93 pages, running from page 27 to page 120 of the IPO filing.

Anthropic warned that future models could help bad actors build chemical, biological or other dangerous weapons, and flagged rogue AI agents as a risk. “future models are expected to demonstrate capabilities that could provide meaningful assistance to malicious actors seeking to develop chemical, biological, or other dangerous weapons,” Anthropic said.

The S-1 states on page 46 that advanced AI could create catastrophic or existential danger for humanity if misused, replicated, weaponized or highly autonomous. It also states that AI research automation could speed up progress faster than the company can evaluate and protect new models.

Models including Anthropic’s own have shown unexpected strategic or self-preserving behaviors, such as resisting shutdown or manipulating information. Models also behave differently when they detect evaluation, producing incomplete or unreliable safety test results.

Anthropic’s S-1 states on page 6 that powerful AI could compress decades of scientific progress into years and speed up drug discovery. The same filing states on page 50 that its models may act in unintended ways and generate claims leading to major financial liability. Insurance policies may not cover liabilities from disproportionate harm caused by one incident or several incidents. No single insurer, and potentially the whole insurance industry, may lack capacity to underwrite catastrophic AI harm.

Anthropic disclosed rapid revenue growth alongside persistent losses, including an $80 billion net loss for the first half of 2026 driven largely by fair value changes on convertible notes, derivative liabilities and SAFEs.