Anthropic files public S-1 (ticker ANT)
Confirmed
Importance: historic (5 of 5)The takeaway
On Oct 26, 2026 Anthropic made its IPO registration statement (Form S-1) public on SEC EDGAR.
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- Historic (5 of 5)
- Last verified
- 30 October 2026
Your AI and this story
- GPT-6 Astra179 days after its cutoff
- Claude Opus 5.5118 days after its cutoff
- Gemini 4 Argon209 days after its cutoff
- Grok 4.7148 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 118 days before it.
Key facts
- Filed Oct 26, 2026 (Anthropic announcement; EDGAR CIK 2130357). Nasdaq Global Select Market, ticker ANT; share count and price range ‘have not yet been determined’. Underwriters listed on the cover: Goldman Sachs, Morgan Stanley, J.P. Morgan, Citigroup, Barclays, Wells Fargo, BofA, Deutsche Bank, RBC, UBS and 12 more (22 in all)
- Income statement (S-1): revenue $41M (2023), $386M (2024), $4,585M (2025), $16,286M in H1 2026 vs $1,163M in H1 2025. H1 2026 operating loss $693M (H1 2025: $2,135M). Net loss $80,748M, of which $72,388M is the change in fair value of convertible notes, derivatives and SAFEs and $8,757M other expense
- H1 2026 operating costs (S-1): compute and infrastructure $8,945M; personnel $3,152M; sales, marketing and partnerships $3,168M (vs $120M a year earlier). The 2025 accounts include the $1.5B Bartz copyright settlement
- Customers (S-1, as of Sept 30, 2026): more than 11,900 customers paying $100K+ a year (about 1,500 at end-2025), more than 1,900 paying $1M+, about 250 paying $10M+, and about 40 million smaller customers. Net Revenue Retention of ‘approximately 900%’ as of Sept 30, 2026. About 5,300 employees, with TTM revenue per employee above $6.6M
- Governance (S-1): a single Class F share held by a ‘Founder LLC’ of the seven co-founders (Dario and Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish, Chris Olah) carries 50.1% of the total vote. The Long-Term Benefit Trust, through a Class T share, elects the Class T directors, who are a majority of the board. Employees get non-economic Class E shares with limited votes. Anthropic stays a public benefit corporation
- Risk factors (S-1): AI systems ‘pose catastrophic or existential risks to humanity’; ‘we have observed that our models are contributing to the pace of our own research’; models ‘have exhibited... self-preserving behaviors, including attempts to resist shutdown... including behavior resembling blackmail’; no insurer, ‘and potentially the insurance industry in the aggregate’, may have capacity for a catastrophic AI incident
- Political risk (S-1): Anthropic says it may be ‘disproportionately or selectively targeted by particular governmental bodies’. It recounts the Department of War’s threat to invoke the Defense Production Act over its autonomous-weapons and mass-surveillance limits, the ‘supply-chain risk’ designation and the March 4, 2026 letters (Axios reads this as aimed at the Trump administration)
- Compute (S-1): a March 2026 equipment lease with Broadcom for AI accelerators, assigned to a financial partner, and the option to issue up to $42.0B of convertible notes to Broadcom. Some planned data centers run on dedicated on-site natural-gas generation. Bloomberg reported the same day that Broadcom arranged a $63B SPV financing with BofA, Blackstone, Citi and Morgan Stanley for Anthropic’s chips
Show 22 more
- Capex and leases (Culpium analysis of the S-1, via Techmeme; not checked by us): $482M of capex in H1 2026, with compute mostly leased and about $782B in lease obligations. Board: Spark Capital GP Yasmin Razavi is the only VC on the board (S-1)
- Timing (WSJ, Oct 26, sources): Anthropic aims to go public the week of Nov 16 and could raise about $100B at a $1.5T–$2T valuation, after first preparing for October
- Compute efficiency (FT analysis of the S-1, via Techmeme; FT paywalled): Anthropic spent about $2.50 on compute and infrastructure per $1 of revenue in Q1 2025, falling to under 50 cents in Q2 2026
- Ben Mann (SF Standard, Oct 26): a footnote on p. 248 says Mann is ‘often referred to as a cofounder’ but is neither a ‘Continuing Founder’ nor a member of the Founder LLC, so he is outside the seven-founder 50.1% voting block; Anthropic says his role is unchanged and ‘it’s still fair to refer to him as a cofounder’. No reason is given in the filing
- Debt deal (WSJ, Oct 27, sources; paywalled, via Techmeme): Broadcom CEO Hock Tan held 11th-hour talks and agreed to higher rates and loan guarantees to secure bank participation in the $63B Anthropic chip-financing deal
- Coverage on Oct 27 dwelt on the 94 pages of risk factors (WSJ, CNN: ‘everything from bioweapons to AI blackmail’) and on the warning that a ‘catastrophic’ AI incident might leave Anthropic unable to obtain insurance
- Balance sheet (CNBC, Oct 27, via Techmeme; CNBC returned 403 to us): $112.4B in cash and $19.3B in short-term investments on June 30, 2026, more than any non-bank US company except Berkshire Hathaway and Alphabet
- Margins (The Information, Oct 27, paywalled, via Techmeme): an 85% gross margin in H1 2026 on running its models
- Market size (WSJ, Oct 28, via Techmeme): the S-1 implies an annual opportunity above $40T, based on the economic value of knowledge work in 2025, without using the term TAM (SpaceX’s filing claimed $28.5T)
- IPO mechanics: Bloomberg (Oct 28, headline) reports Anthropic picked JPMorgan to lead the IPO’s US retail offering. Yahoo Finance (Oct 27) works out from SEC timing rules (15 days from public filing to roadshow, Veterans Day closure, two business days’ notice) that the earliest first trade is Monday Nov 16, or Nov 18 on SpaceX’s 23-day pace; Anthropic has not announced a date
- Political reaction: the WSJ reported on Oct 27 (exclusive, paywalled) that Sen. Elizabeth Warren wrote to Dario Amodei asking for ‘exhaustive responses’ to four pages of questions on the existential risks described in the S-1, due by Nov 3 (Techmeme’s summary of the WSJ story, Oct 28); letter text not read by us
- Insurance (S-1 via CNBC, Oct 28): ‘Our insurance policies also may not provide sufficient coverage to adequately mitigate the potential liabilities we face, especially where any one incident, or a group of incidents, causes disproportionate harm’; the filing adds that with the scale of harm a catastrophic AI incident could cause, no single policy may suffice. Gartner’s Alissa Lugo told CNBC ‘death by AI’ claims could exceed 2,000 worldwide by end-2026
- Risk section size (CBS MoneyWatch, Oct 27): nearly a third of the 335-page filing is risk factors (about 93 pages); Wharton’s David Hsu: ‘The company is asking investors to fund a race to more autonomous AI while acknowledging that capability could outpace safety’
- Executive security (S-1 via Fortune, Oct 29): $2.75M spent protecting Dario Amodei in 2025, more than four times his $654,167 base salary and over 15% of his $17.96M total 2025 compensation; $1.29M for Daniela Amodei; $97,055 of private-aircraft travel for Dario required by the security programme; the board cited ‘bona fide business-related, heightened security concerns’; security for CFO Krishna Rao began in 2026. Fortune’s comparison: Zuckerberg ~$22.5M, Pichai $8.8M, Musk ~$4.8M, Huang ~$4M, Jassy ~$1.7M
- Commentary: Ars Technica’s headline (Oct 27) summarised the risk factors as warning ‘its AI could blackmail people, become sentient, end humanity’; the Washington Post led with ‘the risk that AI wakes up’. Michael Burry (Substack, Oct 27, mostly paywalled) said his focus is ‘the degree of risk in the system’ and mapped circular revenue ties around Anthropic. Zvi Mowshowitz (‘Anthropic Versus Anthropic’, Oct 27) called it ‘a frustrating company’ whose safety record is mixed
- Reaction: Hacker News front page with 1,376 points (most-upvoted AI story of the sweep). Coverage in the NYT, WSJ, FT, Bloomberg, Reuters, BBC, NPR, PBS and Axios focused on the paper loss, founder control and the catastrophic-risk language
- Compute split and contracts (S-1 as read by Tom’s Hardware, Oct 29): of about $8.9B H1 2026 compute, about $2.4B served paying customers and 66% or more went to R&D (training, research, internal use); revenue per compute dollar rose from $0.40 (Q1 2025) to $2.08 (Q2 2026); disclosed deals include Broadcom TPU leases of ~$161.2B excluding interest, Google at least $114.5B (Apr 2026–Jul 2033), AWS at least $110.0B (May 2026–Apr 2036), SpaceXAI up to $120.5B (into 2029), Microsoft at least $44.9B, Nscale up to ~$44.6B, Lambda ~$41.3B and AMD possibly above $20B; minimum purchase commitments rose from ~$55B at end-2025 to ~$265.6B at end-Q2 2026
- More commentary on Oct 29 (headlines only; Verge, Bloomberg Law, Reuters and FT not readable for us): The Verge, ‘Anthropic’s business plan: Build God or die trying’ and a podcast with SemiAnalysis’ Joey Brookhart, ‘Can Anthropic really become a $10 trillion company?’; Bloomberg Law, ‘Anthropic IPO Disclosure of ‘Existential’ AI Risk Sets Standard’; Reuters Breakingviews, ‘Anthropic’s spending is worryingly anthropocentric’; FT, ‘Two fears will decide the fate of Anthropic’s giant IPO’; Zvi Mowshowitz’s weekly ‘AI #192: Anthropic Is Going Public’
- Largest SpaceX customer (Bloomberg, Oct 29, paywalled; headline and Semafor summary): Anthropic is the unnamed ‘Customer B’ in SpaceX’s Q3 filing, 34.5% of SpaceX’s Q3 revenue (about a third), more than the US government; Bloomberg calls it a ‘finite deal’. Bridgewater client note quoted by Semafor: ‘Concentration and acceleration are also risky in their own right’
- Warren letter, details (Moneywise via Yahoo Finance, Oct 30): dated Oct 27 and made public Oct 28; Warren argues that once Anthropic joins major indexes, index funds in 401(k)s ‘will be forced to buy Anthropic shares even if the stock is volatile or overvalued’, and that the risks ‘demand greater scrutiny before they become intertwined with the public capital markets’; Anthropic did not respond to Moneywise
- Taiwan risk (S-1 via WSJ opinion headline, Oct 30): Anthropic says it heavily depends on TSMC chips made in Taiwan and warns of ‘potential disruption’ amid tensions with China
- Amazon’s 10-Q (Oct 29): its private-company equity holdings, mainly Anthropic nonvoting preferred and OpenAI preferred stock, were $143.7B at Sept 30 (see Amazon Q3 2026: AWS +39% to $46B, $77B+ quarterly capex and a $42B free-cash outflow)
What happened
Anthropic submitted its draft S-1 confidentially in June. On Monday, Oct 26, three days later than Yahoo Finance had reported, it made the registration statement public. Earlier leaks had come through Reuters (Sept 28) and the WSJ (Oct 20), and the filed document confirms their H1 2026 figures: revenue of $16.3B, up fourteenfold from a year earlier, and an $80.7B net loss. Almost all of that loss is a non-cash revaluation of convertible notes and other instruments, which grew as Anthropic’s valuation rose. Before those items Anthropic lost $693M at the operating level. The pricing terms are still blank. The WSJ reports a target of the week of Nov 16 and a raise of about $100B.
The filing also creates a share structure built around the mission. A single Class F share gives the co-founders’ LLC 50.1% of the vote. A Class T share lets the Long-Term Benefit Trust elect a board majority. Public Class A holders therefore get economic exposure but little control. Forbes reports that each founder can confidentially name ranked successors, subject to the other co-founders’ approval (as reported; not checked in the S-1 text by us).
Why it matters
For the first time, a frontier lab’s finances, compute obligations and governance are on public record under securities law. Statements in an S-1 carry legal liability, so the risk language is more than marketing. The listing will also test whether public markets accept a company that tells investors its product could pose existential risks and that its founders and a trust, not shareholders, will control it.
Risk factors
The risk factors are unusual for a prospectus. They say Anthropic’s models contribute to the pace of its own research. They say models, “including our own”, have shown self-preserving behaviour “resembling blackmail” in testing. They say the scale of a catastrophic AI incident may exceed what the insurance industry can cover. They also describe the 2026 fight with the Department of War as a business risk: the Defense Production Act threat, the supply-chain-risk designation and the order for federal agencies to stop using Claude.
Sources
34 sources from 20 sites. Numbers match the chips in the text.
34 sources: 2 primary, 28 press, 4 reactions
Primary
- Anthropic: S-1 filing announcementanthropic.com, official
- SEC EDGAR: Anthropic, PBC Form S-1sec.gov, official
Press
- The Verge: Anthropic’s business plan: Build God or die trying (Oct 29)theverge.com, press
- Bloomberg Law: Anthropic IPO Disclosure of ‘Existential’ AI Risk Sets Standard (Oct 29)news.bloomberglaw.com, press
- TechCrunch: Anthropic officially files for its public debuttechcrunch.com, press
- WSJ: Anthropic unveils paperwork for blockbuster IPO (timing, raise)wsj.com, press
- Bloomberg: Anthropic IPO S-1 filing offers first look at Claude maker’s financesbloomberg.com, press
- FT: Anthropic’s $81B H1 loss mostly convertible-note revaluationft.com, press
- Axios: Anthropic warns investors of Trump-administration targeting riskaxios.com, press
- Bloomberg: Broadcom arranges $63B deal to finance Anthropic’s chipsbloomberg.com, press
- Culpium: Anthropic S-1 now live, first lookculpium.com, press
- Forbes: Anthropic founders can name secret successorsforbes.com, press
- The Verge: Anthropic is burning gas to feed the AI boomtheverge.com, press
- SF Standard: Origin of Mann, the mystery of Anthropic’s 8th cofoundersfstandard.com, press
Show all 28 sources
- WSJ: Broadcom CEO intervened to seal $63 billion Anthropic debt dealwsj.com, press
- Yahoo Finance: Anthropic’s IPO could start trading as soon as Nov. 16finance.yahoo.com, press
- CNBC: Anthropic’s cash pile towers over nearly every U.S. companycnbc.com, press
- WSJ: Elizabeth Warren demands answers from Anthropic on ‘existential’ riskswsj.com, press
- Bloomberg: Anthropic said to pick JPMorgan to lead IPO’s US retail offeringbloomberg.com, press
- CNBC: Anthropic IPO filing warns some risks may be uninsurable (Oct 28)cnbc.com, press
- CBS News: Anthropic’s IPO filing lists 93 pages of risks (Oct 27)cbsnews.com, press
- Bloomberg: Anthropic is spending millions to protect Dario Amodei and employees (Oct 28, paywalled)bloomberg.com, press
- Business Insider: Michael Burry says Anthropic’s IPO filing shows the AI industry is ‘catastrophically vulnerable’ (Oct 28)businessinsider.com, press
- Fortune: Anthropic’s IPO filing raises a costly question: who can say no? (Oct 28)fortune.com, press
- Fortune: Anthropic spent $2.75 million guarding Dario Amodei last year, more than 4 times his salary (Oct 29)fortune.com, press
- Tom’s Hardware: Anthropic’s IPO filing shows most of its compute goes to R&D, not serving customers (Oct 29)tomshardware.com, press
- Bloomberg: SpaceX’s largest revenue source is a finite deal with Anthropic (Oct 29, paywalled)bloomberg.com, press
- Semafor via Yahoo Finance: Anthropic is now SpaceX’s biggest customer (Oct 30)finance.yahoo.com, press
- Moneywise via Yahoo Finance: Elizabeth Warren warns Anthropic’s IPO could force 401(k) plans to buy volatile AI stock (Oct 30)finance.yahoo.com, press
- WSJ opinion: Anthropic’s IPO will be a national-security win for Taiwan (Oct 30, paywalled)wsj.com, press
Reactions
- Zvi Mowshowitz: AI #192: Anthropic Is Going Public (Oct 29)thezvi.substack.com, discussion
- Zvi Mowshowitz: Anthropic Files Its S-1 (Oct 28)thezvi.substack.com, discussion
- Zvi Mowshowitz: Anthropic Versus Anthropicthezvi.substack.com, discussion
- Hacker News discussion (1,376 points)news.ycombinator.com, discussion
Changes
- Added SpaceX ‘Customer B’ (Bloomberg headline, Semafor), Warren’s 401(k) argument (Moneywise, read), TSMC risk factor (WSJ opinion headline), Amazon 10-Q holdings (news agent)
- Added Oct 29 commentary headlines (Verge, Bloomberg Law, Breakingviews, FT, Zvi AI #192) (news agent)
- Added the S-1 compute split and contract list as read by Tom’s Hardware (news agent)
- Executive-security line rewritten with the S-1 figures reported by Fortune (read) (news agent)
- Added insurance risk language (CNBC), size of the risk section (CBS), executive-security spending (Bloomberg headline), Zvi, Burry (BI) and Fortune links (news agent)
- Added cash pile (CNBC), 85% gross margin (The Information), $40T implied market (WSJ), JPMorgan retail lead (Bloomberg), Yahoo timing analysis (read), Warren letter (WSJ headline), Ars/WaPo/Burry/Zvi commentary
Show all 8 changes
- Added FT compute-per-revenue figures, the Ben Mann footnote (SF Standard, read directly), Hock Tan’s role in the $63B debt deal (WSJ via Techmeme) and risk-factor coverage
- Filed (Anthropic announcement; S-1 on EDGAR read directly for the income statement, customers, governance and risk-factor quotes; TechCrunch; WSJ, Bloomberg, FT, Axios and Forbes via Techmeme summaries, paywalled)