Anthropic S-1 discloses existential AI risks alongside $16.3B half-year revenue

Anthropic, the company behind the Claude model family, has filed a Form S-1 ahead of a public listing, and the document is drawing attention for what it concedes as much as for what it promises. According to a dev.to analysis published on 27 October, the prospectus runs to 335 pages, and its risk section — spanning roughly pages 27 through 120 — includes an acknowledgement that the company's own products could pose existential risks to humanity.

An unusually candid filing

As the dev.to post frames it, most IPO filings are built to reassure investors, whereas Anthropic's reads as an unusually direct account of how fragile frontier AI development can be. The analysis calls it the most extensive public disclosure a leading AI lab has produced to date. Whether that candour is matched by the final offering terms remains to be seen, since the post notes that the share count and price range have not yet been published.

The loss that is mostly an accounting artefact

The headline number from the filing is a net loss of $80.7 billion for the first half of 2026. The dev.to breakdown explains that $72.4 billion of that comes from a non-cash adjustment: convertible notes, derivatives and SAFEs were marked up in value as private markets revalued Anthropic upward, and that revaluation registers as a loss on paper even though the company's implied worth rose.

Looked at on an operating basis, the picture is different. The dev.to figures show an operating loss of $693 million for the first half of 2026, and positive operating income of $974 million in the second quarter. Revenue for the six-month period reached $16.3 billion, a 1,304 percent jump from $1.16 billion a year earlier. The multi-year trend is steep as well: $41 million in 2023, $386 million in 2024 and $4.6 billion in 2025.

The cost side shows where the money goes. Compute and infrastructure spending hit $8.9 billion in the first half of 2026 alone, per the dev.to summary, after totalling $7.3 billion across all of 2025.

The offering being targeted

According to dev.to, Anthropic is seeking a valuation between $1.5 trillion and $2 trillion, with a Nasdaq listing under the ticker ANT possible as soon as 16 November. The post suggests the company could raise up to $100 billion, which would surpass SpaceX's June debut as the largest IPO on record. It also argues the deeper story is not the money being raised but the spending already committed, since the business model rests on enormous, continuous investment in computing power — a strength the analysis simultaneously identifies as the model's main vulnerability.

These details come from a single secondary account of the filing rather than from the S-1 itself, and final terms may shift before the listing.

Why it matters

If the dev.to reading holds up, this filing sets a disclosure benchmark for the AI industry. Any lab that follows Anthropic to the public markets will now be measured against a prospectus that names existential risk in its own risk factors, and regulators, customers and competitors will all have that text to point to.

It also matters for how the market interprets AI financials. An $80.7 billion loss headline and a profitable quarter can describe the same company once derivative revaluations are separated from cash economics, and investors who miss that distinction will misprice the sector. For developers, the filing quantifies the trade-off that defines frontier AI right now: revenue is compounding at extraordinary rates, but nearly all of it is being ploughed back into compute, which keeps upward pressure on the infrastructure costs that ultimately shape API pricing and tooling budgets across the ecosystem.