Anthropic publicly filed its S-1 registration statement with the US Securities and Exchange Commission on Monday, 26 October 2026 — the step that turns an IPO from rumour into paperwork. According to Anthropic’s own announcement, the company intends to list its Class A common stock on the Nasdaq Global Select Market under the symbol “ANT,” with the share count and price range still undetermined. What is not undetermined is the scale: as SiliconANGLE reported on the filing, the offering could raise as much as $100 billion, which would eclipse the record set by the SpaceX flotation earlier this year. This would make Anthropic the first pure-play AI lab to publicly file for a US listing, ahead of OpenAI’s expected debut in 2027.
The financial profile in the filing is the reason this is a genuine milestone and not just another funding round with extra steps. Anthropic’s revenue grew from $1.16 billion in the first half of 2025 to $16.3 billion in the first half of 2026 — a fourteenfold jump — but the company posted a net loss of $80.7 billion over the same period, up from $13 billion a year earlier, as CNBC’s live filing coverage worked through the prospectus entry by entry. Implied gross margin improved from around 12% to 85% as inference efficiency caught up, yet the company’s non-cancellable purchase commitments ballooned from $54.6 billion at the end of 2025 to $265.6 billion by mid-2026, mostly compute: $110 billion committed to Amazon’s Trainium chips through 2036 and $114.5 billion to Google TPUs through 2033. Anthropic has also agreed to provide its own chips a “single fungible pool,” routing models to whichever silicon is cheapest and fastest at a given moment rather than marrying one vendor.
The strategic geography of the filing is worth more attention than the valuation chatter. Amazon has poured $13 billion into Anthropic since 2023 with room for $20 billion more, holds $6.2 billion in convertible notes, and has separately agreed up to $15 billion in financing tied to compute milestones — but its stake is capped at 12%, it holds nonvoting shares, and it has no board seat. Alphabet, meanwhile, holds an option to buy up to $10 billion of Class N stock concurrent with the offering. The filing also discloses leases for data-centre capacity in Australia and Canada and states plainly that Anthropic is pursuing capacity outside the US to reduce “concentration risks associated with operating a significant portion of our compute infrastructure in the United States.” A company whose product is intelligence is deliberately spreading where that intelligence physically lives.
The safety disclosures are what separate this document from every other mega-IPO in market history. The prospectus runs more than 90 pages of risk factors, and among them Anthropic warns that its most capable models have demonstrated the ability to help malicious actors develop weapons — including chemical and biological ones — that its deployment safeguards are “incomplete,” and that safeguards have been circumvented before and may be again. If models were actually used to build such weapons, the company flags legal liability, regulatory action, and loss of government and enterprise relationships as outcomes. This is not a startup hedging about competition; it is a listed-company-to-be telling prospective shareholders, in a binding disclosure document, that its core product carries risks its own mitigation cannot fully contain. The filing follows CEO Dario Amodei’s September call for the industry to slow frontier development — our own coverage tracked that appeal and the split it created among lab leaders — and the dissonance between “slow down” and the largest IPO in history is now part of the same official record.
There is also a quieter NZ-relevant thread here: we wrote in September about the S&P 500 changing its rules to admit giant AI listings, and that rules change exists largely because filings like this one keep outgrowing every index committee’s assumptions. When a company with $80.7 billion in half-year losses can become one of the world’s most valuable listed entities on the strength of $16.3 billion of revenue and an 85% gross margin, retirement funds well beyond the US — KiwiSaver included, via global index trackers — end up holding the compute bet whether they chose it or not. The story the AP picked up on the filing framed the CEO warning about risks to humanity alongside the biggest IPO ever filed; both halves are accurate, and holding them together is the whole analysis. Anthropic going public does not resolve the tension it disclosed — it finances it, at record scale, with the safety caveats now priced in.
Sources: Anthropic, SiliconANGLE, CNBC, AP News.
