Reuters obtains Anthropic’s IPO prospectus
$2T+ target valuation, $4.6B 2025 revenue, 80 pages of risk factors incl. existential risk
Partly confirmed
Importance: major (4 of 5)The takeaway
On Sept 28, 2026 Reuters reported on Anthropic’s IPO prospectus.
Status
- Claim
Partly confirmed
- Our reporting
- Medium confidence
- Importance
- Major (4 of 5)
- Last verified
- 21 October 2026
Your AI and this story
- GPT-6 Astra151 days after its cutoff
- Claude Opus 5.590 days after its cutoff
- Gemini 4 Hydrogen181 days after its cutoff
- Grok 4.7120 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 90 days before it.
Key facts
- Reported by Reuters (exclusive) on the evening of Sept 28, 2026; follow-ups by CNN, Fortune, CNBC on Sept 29. Anthropic did not comment (CNN)
- Potential valuation: more than $2 trillion, over double its own estimated $965B valuation in May 2026
- 2025 revenue: nearly $4.6B (12x growth); operating loss >$8B; net loss ~$42B, including a ~$34B non-cash charge on financing convertible into shares
- 2025 compute and infrastructure spend: $7.33B (3x 2024), over half of $12.65B total operating expenses; $518B in cloud/compute/infrastructure obligations in coming years
- Cash, equivalents and short-term investments: $20.28B at Dec 31, 2025; nearly a quarter of 2025 revenue came from two customers
- About 80 of 261 pages are risk factors (48 on the business). They warn of ‘catastrophic or existential risk to humanity’ and models showing ‘self-preserving behaviors’, trying to ‘resist shutdown’, ‘conceal or manipulate information’ and behavior ‘resembling blackmail’
- FT (also reviewed the filing): Q2 2026 revenue alone was $11.5B, and Anthropic is on track for a second straight quarter of adjusted operating profit
- Listing likely after the November 2026 US midterms (Reuters sources); OpenAI confidentially filed for an IPO in June, and SpaceX listed on June 12 at a $1.77T valuation
Show 25 more
- Executive pay (Reuters exclusive, Oct 6): Dario Amodei earned about $18M in 2025, mostly stock (more than the Alphabet and Amazon CEOs, below Oracle and Nvidia); President Daniela Amodei $16.4M; both base salaries doubled to $1.4M in July 2026
- Governance (Reuters, Sept 29): the seven co-founders will initially hold 50.1% of total voting power through a new ‘Founder LLC’ meant to serve the common good; The Information (Sept 24) first reported the Palantir-style structure, which requires three founders to keep minimum stakes
- Reuters: revenue grew ~12x to ~$4.6B in 2025, about 25% from two customers; operating loss above $8B
- SpaceX/xAI: agreements for Nvidia-based compute worth up to $84.5B through 2029, largely cancelable on 90 days’ written notice (Reuters, The Information); Seeking Alpha reports this is nearly double earlier estimates
- Orbital compute: Anthropic said on May 6, 2026 it had ‘expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity’ (official). Crypto Briefing (Sept 29) says the prospectus describes ‘exploratory discussions’ on multi-gigawatt orbital compute; Reuters/FT reports checked here do not mention it, so this detail is unconfirmed
- Per-partner obligations (Reuters): at least $111.1B with Google (April 2026–July 2033), $110B with Amazon (May 2026–April 2036) and $31.4B with Microsoft (Nov 2026–May 2033, ‘non-cancelable except in the event of Microsoft’s uncured material breach’); Broadcom lease arrangements are non-cancelable by either party except on default. ‘If our actual spend falls short, we must pay Google the difference’
- Reuters (Sept 29): Anthropic expects to spend at least $518B over a decade with six infrastructure partners; about 80% is non-cancelable or payable regardless of usage. The filing says future AI demand will be ‘limited principally by the availability of compute’
- Reuters (Sept 29): Anthropic routed 47% of 2025 sales (~$2.16B) through cloud marketplaces of Amazon and Google, up from 11% in 2023 and 32% in 2024, and paid them ~$351M in distribution fees (~16 cents per marketplace dollar); ~60% of $909M in outstanding receivables was collected via the cloud providers
- Reuters (Sept 29): 2025 consumption-based revenue ~$3.8B and subscription revenue $789M; non-cancelable hosting/compute commitments were $54.6B at end-2025 before rising to over $417B in early 2026; 3.5 GW of dedicated compute
- IPO timing (Bloomberg, Oct 1): Anthropic could start formal IPO marketing as soon as the week of Nov 9, 2026 and begin trading before Thanksgiving (Nov 26). It still expects to list by year-end; valuation target up to ~$2T
- Broadcom (Reuters, Oct 1): up to $42B in convertible notes toward a $125.2B five-year TPU lease from 2027; see Anthropic prospectus: Broadcom to lend up to $42B toward a $125.2B TPU lease
- Offering size (Reuters, Sept 11): Anthropic was discussing raising up to $100B at roughly a $2T valuation, which would be the largest IPO ever (SpaceX raised ~$86.25B gross in June); Nvidia was in talks to invest up to $10B as an anchor investor. Terms were preliminary. Repeated by Motley Fool/Yahoo Finance on Oct 3, which put $100B at about 24% of the $518B commitments
- Founders’ pledge (Reuters via Yahoo Finance, Sept 30): the IPO filing says the seven co-founders pledge to donate 80% of their personal equity to charity; Dario Amodei’s 2025 pay was ~$18M and Daniela Amodei’s $16.4M, mostly stock and options
- Charity cost (The Information, Oct 4, paywalled; seen as a search snippet): Anthropic told prospective investors it booked a non-cash expense of more than $660M for stock used to match employees’ charitable donations from October 2025 to March 2026
- Oct 3–4 coverage wave (Mashable on ‘$40B+ losses’, The Times ‘Wall Street reality check’, PressTV on ‘existential risks’, BBN Times on Nasdaq gains) repeats the Sept 28–Oct 1 reporting; no public S-1 had been reported as of Oct 4
- Charity match detail (AI Weekly summarising The Information, Oct 4-5): the Q1 2026 portion was about $125M, ~10% of employee expenses and ~2% of operating costs; the charge is expected to reach billions after the IPO and dilute other shareholders; the co-founders are not eligible for the match. Calcbench: Anthropic’s 2025 charitable contributions (~$540M) top the Fortune 500 comparison, next BlackRock at $109M
- PitchBook analysis (Oct 2, republished by Morningstar Oct 5): about $34B of the ~$42B 2025 net loss is a non-cash remeasurement of convertible notes; the operating loss was >$8B, with $7.33B spent on compute/infrastructure (58% of opex). Q1 2026 revenue $4.73B; Q2 >$11.5B (preliminary). $2T is ~435x 2025 revenue, <44x annualized Q2 revenue and just under 31x the July run rate (>$65B)
- PitchBook commitment breakdown (~$518B, six partners): Broadcom-related equipment leases $161.2B, Google $111.1B, Amazon $110.0B, xAI up to $84.5B (mostly cancelable on 90 days’ notice), Microsoft $31.4B, AMD >$20B; ~80% noncancelable. Amazon/Google marketplaces produced 47% of 2025 revenue. Verdict: the leak supports a very large valuation but not yet $2T until the public S-1 shows gross margin and the payment schedule
- CNN (Oct 5), citing the WSJ: Anthropic is moving forward despite rising rates and the Iran-war energy shock; the IPO could come as soon as November and raise up to $100B at ~$2T, which would top SpaceX (~$86B raised at $1.77T) as the largest IPO ever
- Political pushback (Oct 5): Sen. Bernie Moreno (R-Ohio) wrote to Dario Amodei calling it hypocritical to ask Americans to invest while warning the product could cause extinction (see Sen. Bernie Moreno writes to Dario Amodei attacking Anthropic’s ‘alarmist’ superintelligence warnings ahead of its IPO)
- Bloomberg Opinion column (Oct 6), ‘What Anthropic’s Existential Risks Tell IPO Investors’ (URL slug: ‘prospectus is long on flash, short on information’); not read, paywalled
- Commentary, Oct 9 (headlines; articles not read in full): Bloomberg’s newsletter asked how investors should ‘price rogue AI risk’ given the risk factors, and investor Bradley Tusk asked on CNBC why Anthropic would ‘add public market pressure if safety is your top priority’; Mashable re-reported the ~$42B 2025 loss
- Bloomberg (Anthony Hughes, Oct 8–9; read in The Nightly’s republication), ‘Anthropic IPO investors struggle to put a price on rogue AI risk’: investors weighing a valuation of up to $2T cannot quantify liability for severe AI scenarios. Sam Lessin (Slow Ventures) proposed that the US require AI labs to put $1 trillion in escrow for future clean-up costs: ‘The liabilities might be a $2 trillion or $4 trillion risk to society’. Joe Ziolkowski (Relm Insurance) said cover for AI firms is hard to find: ‘It is a good time to raise capital before some of these things become objectively quantifiable’. Harrison Rolfes (PitchBook) said safety measures add cost but could become a moat; Kevin Moss (Private Shares Fund) said safety talk is unlikely to derail the IPO. SpaceX’s $86.2B June IPO traded more than 20% above its offer price. Anthropic declined to comment
- Bear case (CNBC, Oct 7): research firm New Constructs (founder David Trainer) published a report on Oct 6 calling Anthropic ‘the most ridiculous IPO of 2026’ and ‘an unprecedented test of investor gullibility’. It values the company at $150B and says a $2T valuation would require double Nvidia’s trailing-year profit (over $190B): ‘We don’t think Anthropic has a viable business’, citing operating losses and competition from open-source models. The firm has not seen the filing and worked from the leaked figures. CNBC notes its record is mixed (right on WeWork in 2019, wrong on DoorDash in 2020); Anthropic did not comment. The CNBC story reached 238 points on Hacker News on Oct 12
- Updated draft with H1 2026 numbers (WSJ, Oct 20): revenue $16.3B (vs $1.1B in H1 2025), $80.7B net loss of which ~$72B non-cash, ~$9B compute spend, ~20% of revenue from two customers, up to $750B compute spend ‘in the coming years’; see Anthropic draft S-1: $16.3B H1 revenue, $80.7B paper loss
What happened
Reuters saw Anthropic’s IPO prospectus and reported its first detailed public financials. Revenue grew fast but costs grew faster, and the headline net loss is mostly a non-cash accounting charge. Most attention went to the unusually stark risk section: Anthropic tells investors its own models have shown self-preserving and manipulative behavior in tests, and that AI could pose existential risk.
Why it matters
If filed as reported, it would be the first IPO document from a frontier AI lab. It sets a valuation benchmark for OpenAI and puts catastrophic-risk language into securities disclosures, where it carries legal weight. Confidence is medium because the prospectus is not public and all figures come from Reuters’ reporting.
Sources
39 sources from 24 sites. Numbers match the chips in the text.
39 sources: 1 primary, 36 press, 2 reactions
Primary
- Anthropic: Higher usage limits and a SpaceX compute deal (May 6, 2026; orbital compute interest)anthropic.com, official
Press
- Bloomberg via The Nightly: Anthropic IPO investors struggle to put a price on rogue AI risk (Oct 9)thenightly.com.au, press
- Bloomberg Opinion: What Anthropic’s ‘Existential Risks’ tell IPO investors (Oct 6)bloomberg.com, press
- New York Post: Anthropic IPO filing reveals pay of CEO Dario Amodei and sister Danielanypost.com, press
- Quartz: Anthropic’s IPO filing shows CEO Dario Amodei earned $18 million last yearqz.com, press
- Reuters (Oct 6): Anthropic’s Amodei made $18 million last year, middle of the tech CEO packreuters.com, press
- PitchBook via Morningstar: Anthropic’s leaked financials reflect fast growth, but not a $2 trillion valuationmorningstar.com, press
- CNN: Anthropic expected to IPO despite market uncertainty, AI slowdown calls (Oct 5)cnn.com, press
- The Information: Anthropic’s big charity bill for shareholders (paywalled)theinformation.com, press
- AI Weekly: Anthropic’s charity-match charge hit $660M+ in six months to March (summarising The Information)aiweekly.co, press
- TechCrunch: Anthropic’s prospectus details losses, growth, and a warning that its AI could end humanity (citing FT and Reuters)techcrunch.com, press
- Financial Times on the Anthropic prospectusft.com, press
- Reuters via CNBC: Anthropic’s IPO prospectus shows sweeping AI vision, surging costscnbc.com, press
Show all 36 sources
- Reuters via Yahoo Finance: Exclusive — Anthropic’s IPO prospectusfinance.yahoo.com, press
- CNN: Anthropic says its AI models pose ‘existential risk to humanity’ in leaked IPO filingcnn.com, press
- CNBC: Anthropic warns of AI’s ‘existential risk to humanity’ in IPO filingcnbc.com, press
- Fortune: Anthropic’s leaked IPO prospectus details steep losses, rapid growth, and a fear that AI could end humanityfortune.com, press
- Reuters: Anthropic’s IPO prospectus shows sweeping AI vision, surging costsreuters.com, press
- Reuters: Anthropic leaders to control AI lab via ‘Founder LLC’reuters.com, press
- The Information: Anthropic seeks Palantir-style voting control for seven co-founderstheinformation.com, press
- Seeking Alpha: Anthropic agrees to pay SpaceX nearly double prior estimates for computeseekingalpha.com, press
- Crypto Briefing: Anthropic could pay SpaceX up to $84.5B for computing capacity through 2029 (orbital-compute talks)cryptobriefing.com, press
- The Information: Anthropic discloses up to $84.5 billion in SpaceX compute agreementstheinformation.com, press
- Reuters via TradingView: Anthropic’s $518 billion AI buildout (full text)tradingview.com, press
- Reuters: Anthropic’s $518 billion AI buildout hinges largely on deals that cannot be canceled, filing showsreuters.com, press
- Reuters via US News: Anthropic IPO prospectus lays bare deep dependence on Big Tech partnersmoney.usnews.com, press
- Reuters: Anthropic IPO prospectus lays bare deep dependence on Big Tech partnersreuters.com, press
- Bloomberg: Anthropic targets November IPO, eyes SpaceX-sized debut (Oct 1)bloomberg.com, press
- Yahoo Finance: Anthropic reportedly looking to IPO as early as mid-Novemberfinance.yahoo.com, press
- Bloomberg: Nvidia mulls $10 billion Anthropic IPO backing, Reuters says (Sept 11)bloomberg.com, press
- The Decoder: Nvidia wants to pour up to $10 billion into Anthropic’s record-breaking IPOthe-decoder.com, press
- Motley Fool: Anthropic could raise up to $100 billion in its November IPO (Oct 3)fool.com, press
- Yahoo Finance: Anthropic founders reportedly pledge to give away 80% of equity (Sept 30)finance.yahoo.com, press
- Benzinga: Anthropic IPO founders pledge 80% of their equity to charitybenzinga.com, press
- Bloomberg (newsletter): Anthropic IPO forces question of how to price rogue AI riskbloomberg.com, press
- Mashable: Anthropic IPO docs reportedly reveal over $40 billion in losses last yearmashable.com, press
- CNBC: Research firm calls Anthropic’s IPO ‘unprecedented test of investor gullibility’ (Oct 7)cnbc.com, press
Reactions
- CNBC (video): Bradley Tusk on Anthropic IPO: why add public market pressure if safety is your top priority?cnbc.com, video
- Hacker News discussion of the CNBC story (238 points on Oct 13)news.ycombinator.com, discussion
Changes
- Pointer to the Oct 20 WSJ report on the updated draft S-1 (new entry Anthropic draft S-1: $16.3B H1 revenue, $80.7B paper loss)
- Added New Constructs’ ‘most ridiculous IPO of 2026’ report via CNBC (Oct 7), found through Hacker News
- Added the Bloomberg piece on pricing rogue-AI risk (readable copy found in The Nightly), with the $1T escrow proposal and insurer quotes
- Added Oct 9 commentary links (Bloomberg rogue-AI-risk newsletter, Bradley Tusk on CNBC, Mashable)
- Linked the video “The Stock Market Is Being Used as a Dumping Ground”
- Added Bloomberg Opinion, NY Post and Quartz follow-ups on pay and disclosures
Show all 17 changes
- Added executive pay figures from the prospectus (Reuters)
- Added PitchBook’s analysis of the leaked financials (loss breakdown, per-partner commitments), CNN/WSJ Oct 5 IPO status and the Moreno letter
- Charity-match breakdown via AI Weekly (The Information is paywalled; figures are second-hand)
- Added the up-to-$100B raise and Nvidia’s possible $10B anchor stake (Reuters, Sept 11), the founders’ 80% equity pledge and 2025 executive pay, and The Information’s $660M charitable-match expense (snippet only)
- Added Bloomberg’s November IPO timing and the Broadcom $42B convertible loan
- Added the Reuters URL for the Big Tech dependence story
- Added the SpaceX orbital-compute interest (official May 6 statement; prospectus mention reported only by Crypto Briefing, unconfirmed)
- Added Reuters’ cloud-channel figures (47% of sales via Amazon/Google, ~$351M distribution fees) and revenue mix
- Added the Reuters breakdown of the $518B obligations (Google $111.1B, Amazon $110B, Microsoft $31.4B, Broadcom leases; ~80% non-cancelable) and the up-to-$84.5B SpaceX/xAI compute agreements through 2029 (The Information)
- Added the Founder LLC voting structure and customer concentration
- Filed
