October 27, 2026·57 min read
Anthropic, PBC (ANT), Financial Analysis
Insights derived from analysing SEC filings. Independent analysis, not a publication of the SEC.
Figures are read from the Form S-1 of 26 Oct 2026 with all its exhibits, incl. the charter documents (Ex 3.1 to 3.5), investor and note agreements (Ex 4.1 to 4.7), material contracts and equity plans (Ex 10.1 to 10.14), the Long Term Benefit Trust Agreement (Ex 99.1) and the Founder LLC Agreement (Ex 99.2), the draft registration statement of 1 Jun 2026 and its five amendments, all under central index key 2130357, the Forms D of investment vehicles naming Anthropic, and the comparable filings of eight peers and of SpaceX for its AI segment. Anthropic has not listed and has filed no Form 10-K, 10-Q, 8-K or proxy statement. Accession numbers for every filing used appear in the Sources appendix. Peer prices are as of 23 and 27 October 2026; Anthropic values use the Series H preferred price of May to June 2026.
Valuation
$965bn
Series H post money, May 2026
Revenue 6M 2026
$16.3bn
14x 6M 2025 ($1.2bn)
Net loss 6M 2026
($80.7bn)
incl. $72.4bn non cash fair value change
Revenue FY2025
$4.6bn
FY2024 $386m
Net loss FY2025
($42.0bn)
incl. $34.3bn fair value change
Key findings
- Revenue 6M 2026 $16.3bn against $1.2bn in 6M 2025; Q2 2026 $11.6bn, +1,368% year on year and +144% on Q1 2026. FY2025 revenue $4.6bn, FY2024 $386m.
- First quarterly operating profit in the periods presented (Q1 2025 to Q2 2026): $974m in Q2 2026, 8.4% margin; adjusted operating income $2.2bn (19%). 6M 2026 operating loss $693m against $2.1bn.
- Net loss 6M 2026 $80.7bn, of which $72.4bn is a non cash fair value change on convertible notes, derivative liabilities and SAFEs.
- Compute and infrastructure fell from 251% of revenue in Q1 2025 to 48% in Q2 2026; cost of revenue 14.6% of 6M 2026 revenue against 87.6% a year earlier. Incremental operating margin 38.7% in Q2 2026.
- Committed costs are large against revenue: purchase commitments $265.6bn (5.7x Q2 2026 annualized revenue) and leases not yet commenced $321.4bn; $26.6bn of RSU expense starts at the IPO ($2.9bn at once, $23.7bn over about 2.1 years); management expects operating margins may compress as compute tranches arrive.
- Governance: seven co founders hold a fixed 50.1% of votes through one Class F share; the Long Term Benefit Trust elects 4 of 7 directors; Nasdaq listing planned under ticker ANT.
- At the Series H price ($58.90 per common share equivalent) the implied value is about $998bn; base case (45% revenue CAGR from the Q2 2026 annualized $46.2bn, 10x EV/Sales) implies $86 per share in three years.
1 Revenue & Business Model
- Product: frontier AI models (Claude) sold through the Claude Platform and applications (Claude Code, Claude Cowork, Claude.ai, Claude Enterprise, Pro and Max plans). Model classes: Mythos (incl. Fable), Opus, Sonnet and Haiku.
- Revenue rose from $0 at launch in Q1 2023 to $41m in FY2023, $386m in FY2024 and $4.6bn in FY2025; $16.3bn in 6M 2026. One reportable segment.
- Revenue type: consumption based 80% and subscription 20% in 6M 2026 ($13.1bn and $3.2bn). Consumption covers API and token metered usage; subscription covers Pro, Max and enterprise app plans.
- Distribution: Claude is offered on AWS, Google Cloud and Microsoft Azure; revenue through cloud marketplaces rose from about 11% (2023) to 48% (6M 2026), with a fee paid to each cloud provider.
- Customers at 30 Sep 2026: more than 11,900 with $100k or more of TTM revenue (more than 950 a year earlier), more than 1,900 above $1m and about 250 above $10m; net revenue retention about 900%.
- Concentration: no customer above 10% of 9M 2026 revenue; top five about 25%. United States 72% of 6M 2026 revenue by billing address.
- Model cadence: versions released in the six months to 30 Sep 2026 produced over 80% of Q3 2026 revenue; Opus 5.5 cut cost per task by about 40% against Opus 5.
- Revenue quality: Anthropic estimates 5% to 7% of 9M 2026 revenue came from customers violating its usage policy or commercial terms.
Table 1.1 Quarterly results, Q1 2025 to Q2 2026
| Line | Q1 2025 | Q2 2025 | Q3 2025 | Q4 2025 | Q1 2026 | Q2 2026 |
|---|---|---|---|---|---|---|
| Revenue ($bn) | 0.38 | 0.79 | 1.38 | 2.04 | 4.73 | 11.56 |
| Compute and infrastructure ($bn) | 0.94 | 1.39 | 2.05 | 2.95 | 3.39 | 5.56 |
| Personnel ($bn) | 0.22 | 0.32 | 0.49 | 0.81 | 1.30 | 1.86 |
| Sales, marketing and partnerships ($bn) | 0.02 | 0.10 | 0.20 | 0.25 | 1.03 | 2.14 |
| Operating income ($bn) | (0.94) | (1.20) | (3.09) | (2.84) | (1.67) | 0.97 |
| Adjusted operating income ($bn) | (0.83) | (1.02) | (1.28) | (1.72) | (0.62) | 2.24 |
| Adjusted operating margin (%) | (221) | (129) | (93) | (84) | (13) | 19 |
| Change in fair value of notes, derivatives and SAFEs ($bn) | (5.70) | (5.24) | (9.17) | (14.20) | (12.79) | (59.60) |
| Net income ($bn) | (6.59) | (6.38) | (12.14) | (16.86) | (14.44) | (66.31) |
Expenses are presented by nature in the S-1.
2 Financial Analysis & Ratios
- Gross margin 85.4% in 6M 2026 against 12.4% in 6M 2025 (derived from the cost of revenue in the functional expense note; FY2025 39.5%, FY2024 (94.8%)).
- Compute and infrastructure expense $8.9bn in 6M 2026, +283%, 55% of revenue against 201%; training and research compute sits mostly in R&D ($9.7bn in 6M 2026), serving compute in cost of revenue ($2.4bn).
- Sales, marketing and partnerships $3.2bn in 6M 2026 (19% of revenue), mainly a $2.7bn rise in fees paid to channel partners.
- Net loss is driven by fair value changes: $34.3bn in FY2025 and $72.4bn in 6M 2026 on convertible notes, derivatives and SAFEs; adjusted net income $2.7bn in 6M 2026.
- Operating cash flow turned positive: $3.9bn in 6M 2026 against ($1.5bn); capex $482m; derived free cash flow $3.4bn.
- Balance sheet at 30 Jun 2026: cash and short term investments $131.7bn; capital raised to date $141.1bn; Amazon convertible notes $6.2bn principal ($108.6bn fair value); credit facility $3.25bn with $3.2bn available (raised to $15bn in Sep 2026); stockholders' deficit ($127.6bn) with $157.5bn of preferred stock in mezzanine equity.
- Stock compensation: $2.2bn in 6M 2026; RSU expense of $26.6bn starts at the IPO: $2.9bn cumulative catch up on the IPO date and $23.7bn over about 2.1 years.
Table 2.1 Income statement, cash flow and balance sheet summary ($bn unless stated)
| Line item | FY2023 | FY2024 | FY2025 | TTM Q2 2026 |
|---|---|---|---|---|
| Revenue | 0.0 | 0.4 | 4.6 | 19.7 |
| Gross profit | 0.0 | (0.4) | 1.8 | 15.6 |
| Research and development | 0.4 | 2.3 | 6.3 | 14.2 |
| Selling, general and administrative | 0.1 | 0.3 | 2.0 | 6.5 |
| Operating income | (0.6) | (3.0) | (8.1) | (6.6) |
| Net income | (1.5) | (8.3) | (42.0) | (109.7) |
| Diluted EPS ($) | (1.42) | (5.43) | (23.04) | n/a |
| Operating cash flow | (1.6) | (5.7) | (2.6) | 2.8 |
| Capital expenditure | (0.0) | (0.0) | (0.1) | (0.6) |
| Free cash flow | (1.6) | (5.7) | (2.8) | 2.2 |
| Stock based compensation | 0.0 | 0.2 | 1.2 | 3.1 |
| Cash and investments | n/a | 3.1 | 20.3 | 131.7 |
| Total debt | n/a | n/a | 7.3 | 6.2 |
| Total equity | n/a | (9.8) | (50.5) | (127.6) |
TTM Q2 2026 equals FY2025 less 6M 2025 plus 6M 2026. Gross profit is derived from cost of revenue in the functional expense note. Capital expenditure is purchases of property and equipment. Total debt is convertible note principal. Equity is the stockholders' deficit; preferred stock sits in mezzanine equity.
Table 2.2 What changed, Q2 2026 against Q2 2025
| Metric | Q2 2025 | Q2 2026 | Change |
|---|---|---|---|
| Revenue ($bn) | 0.79 | 11.55 | 1,368.2% |
| Operating income ($bn) | (1.20) | 0.97 | 181.2% |
| Net income ($bn) | (6.38) | (66.31) | (939.8%) |
| Operating margin (%) | (152.5) | 8.4 | 160.9 pts |
| Net margin (%) | (810.3) | (573.9) | 236.4 pts |
Table 2.3 Ratio analysis
| Ratio | FY2023 | FY2024 | FY2025 | TTM Q2 2026 |
|---|---|---|---|---|
| Growth and margins | ||||
| Revenue growth (%) | n/a | 841.5 | 1,087.8 | n/a |
| Gross margin (%) | 36.6 | (94.8) | 39.5 | 79.0 |
| Operating margin (%) | (1,343.9) | (772.0) | (175.8) | (33.6) |
| Net margin (%) | (3,734.1) | (2,152.6) | (915.3) | (556.9) |
| FCF margin (%) | (3,973.2) | (1,478.5) | (60.4) | 11.2 |
| Cost and investment intensity | ||||
| R&D (%) revenue | 1,029.3 | 589.4 | 138.2 | 72.2 |
| SBC (%) revenue | 70.7 | 49.7 | 25.4 | 15.5 |
| Capex (%) revenue | 4.9 | 3.9 | 2.9 | 3.0 |
| Returns | ||||
| Asset turnover (x) | n/a | 0.05 | 0.19 | 0.13 |
| Liquidity and leverage | ||||
| Current ratio (x) | n/a | 14.20 | 8.43 | 19.11 |
| Efficiency and cash quality | ||||
| Days sales outstanding | n/a | 90 | 72 | 108 |
Net margin includes fair value charges on convertible notes of $34.3bn (FY2025) and $72.4bn (6M 2026).
Table 2.4 Valuation reference at the Series H price of $58.90 per common share equivalent (derived)
| Measure | Value |
|---|---|
| Implied equity value ($bn) | 997.6 |
| Fully diluted shares (bn) | 16.9 |
| Net cash: cash and short term investments ($bn) | 131.7 |
| Implied enterprise value ($bn) | 865.9 |
| EV/Sales, TTM revenue $19.7bn (x) | 43.9 |
| EV/Sales, Q2 2026 annualized revenue $46.2bn (x) | 18.9 |
| Tender offer price, Apr 2026 ($) | 25.9 |
Table 2.5 Operating income to adjusted operating income
| Line | FY2023 | FY2024 | FY2025 | 6M 2025 | 6M 2026 |
|---|---|---|---|---|---|
| Operating income ($bn) | (0.55) | (2.98) | (8.06) | (2.14) | (0.69) |
| Stock based compensation ($bn) | 0.03 | 0.19 | 1.17 | 0.29 | 2.19 |
| Non cash charitable contributions ($bn) | 0.00 | 0.00 | 0.54 | 0.00 | 0.13 |
| Bartz settlement ($bn) | 0.00 | 0.00 | 1.50 | 0.00 | 0.00 |
| Adjusted operating income ($bn) | (0.52) | (2.79) | (4.85) | (1.85) | 1.62 |
| Adjusted operating margin (%) | (1,273) | (722) | (106) | (159) | 10 |
Table 2.6 Private financing rounds
| Round | Dates | Price per common equivalent ($) | Gross proceeds ($bn) |
|---|---|---|---|
| Series C-1/C-2 | Mar to Jun 2023 | 1.12 | 0.45 |
| Series D-1/D-3 | Jan to May 2024 | 2.83 | 1.20 |
| Series E-1 | Jan to Mar 2025 | 5.61 | 3.57 |
| Series F-1 | Aug to Sep 2025 | 14.10 | 13.00 |
| Series G-1/G-2 | Jan to May 2026 | 25.91 | 45.20 |
| Series H-1/H-2 | May to Jun 2026 | 58.90 | 65.20 |
Preferred prices divided by the conversion ratio at 30 Jun 2026: 10 to 1 for all series except D-1 and D-3 (about 10.6 to 1). Series G and H include non voting G-2 shares sold to five investors incl. Google and Amazon, and H-2 shares sold to 11 investors.
From operating loss to net loss
Table 2.7 Operating loss to net loss, 6M 2026 and Q2 2026
| Line ($bn) | 6M 2025 | 6M 2026 | Q2 2026 |
|---|---|---|---|
| Revenue | 1.16 | 16.29 | 11.56 |
| Operating expenses | (3.30) | (16.98) | (10.58) |
| Operating income | (2.14) | (0.69) | 0.97 |
| Interest income | 0.12 | 1.11 | 0.76 |
| Change in fair value of convertible notes, derivatives and SAFEs | (10.94) | (72.39) | (59.60) |
| Other income (expense), net | 0.00 | (8.76) | (8.45) |
| Total other income (expense), net | (10.83) | (80.03) | (67.28) |
| Income taxes | (0.01) | (0.02) | 0.00 |
| Net loss | (12.97) | (80.75) | (66.31) |
The fair value change arises mainly from the higher fair value of convertible notes as the company valuation rose; other expense arises mainly from the initial fair value recognition of note purchase agreements. Neither is a cash cost; 6M 2026 operating cash flow was $3.9bn.
Cost structure and profitability
- Compute and infrastructure is the largest cost: $7.3bn in FY2025 (160% of revenue, 653% in FY2024) and $8.9bn in 6M 2026 (55%); Q2 2026 $5.6bn, 48% of revenue against 177% in Q2 2025. Compute is contracted 12 to 18 months ahead and expensed as capacity arrives, so its share of revenue depends on revenue growth against a fixed compute base.
- By function in 6M 2026: cost of revenue (customer serving compute) 14.6% of revenue against 87.6% in 6M 2025; R&D (training, research and internal compute plus staff) 59.8% against 159.1%; sales and marketing 23.4%, which includes channel partner fees and compute for free access; G&A 6.4%.
- Operating margin moved from (183.6%) in 6M 2025 to (4.3%) in 6M 2026 and from (152.5%) in Q2 2025 to 8.4% in Q2 2026. Of the 179 point gain in 6M 2026, compute contributed 146 points and personnel 27 points; other operating and administrative costs 16 points; sales, marketing and partnerships reduced margin by 9 points.
- Revenue per dollar of compute rose 5.2x, from 0.40x in Q1 2025 to 2.08x in Q2 2026 (revenue $11.6bn against compute and infrastructure expense $5.6bn); 1.40x in Q1 2026 and 0.69x in Q4 2025.
- Incremental margin: Q2 2026 revenue rose $6.8bn on Q1 2026 while operating expenses rose $4.2bn, an incremental operating margin of 38.7%; Q1 2026 43.4%. Q2 2026 against Q2 2025: revenue +1,368%, compute +300%, other operating expenses +741% (S-1).
- Stock compensation is the main gap between GAAP and adjusted results: $1.26bn in Q2 2026 (10.9% of revenue); adjusted operating margin 19.3% against GAAP 8.4%. RSU expense of $26.6bn starts at the IPO ($2.9bn on the IPO date, $23.7bn over about 2.1 years), equal to 58% of Q2 2026 annualized revenue.
- Other operating cost lines in 6M 2026: data quality and evaluation vendors +$251m, IT and cloud software +$223m, payment processing +$222m, professional services +$205m, donation matching +$126m; personnel +$2.6bn, of which stock compensation +$1.9bn and cash pay +$510m.
- Losses below the operating line are non cash: fair value changes on notes, derivatives and SAFEs of $72.4bn in 6M 2026 and $34.3bn in FY2025 produce the net losses; interest income $1.1bn in 6M 2026. Accumulated deficit $132.9bn at 30 Jun 2026.
- Cash: operating cash outflows from FY2023 to FY2025 total $10.0bn; 6M 2026 operating cash inflow $3.9bn. Capital raised to date $141.1bn.
Table 2.8 Operating expenses by function
| Line | FY2023 | FY2024 | FY2025 | 6M 2025 | 6M 2026 |
|---|---|---|---|---|---|
| Cost of revenue ($bn) | 0.03 | 0.75 | 2.77 | 1.02 | 2.38 |
| Research and development ($bn) | 0.42 | 2.28 | 6.34 | 1.85 | 9.74 |
| Sales and marketing ($bn) | 0.08 | 0.16 | 0.91 | 0.24 | 3.81 |
| Bartz settlement ($bn) | 0.00 | 0.00 | 1.50 | 0.00 | 0.00 |
| General and administrative ($bn) | 0.07 | 0.18 | 1.12 | 0.19 | 1.05 |
| Total operating expenses ($bn) | 0.59 | 3.37 | 12.65 | 3.30 | 16.98 |
| Cost of revenue (% of revenue) | 63.4 | 194.8 | 60.5 | 87.6 | 14.6 |
| Research and development (% of revenue) | 1,029.3 | 589.4 | 138.2 | 159.1 | 59.8 |
| Sales and marketing (% of revenue) | 185.4 | 42.2 | 19.9 | 20.6 | 23.4 |
| General and administrative (% of revenue) | 165.9 | 45.6 | 24.5 | 16.3 | 6.4 |
| Total operating expenses (% of revenue) | 1,443.9 | 872.0 | 275.8 | 283.6 | 104.3 |
Functional split from Note 8 of the S-1. Cost of revenue consists substantially of compute for customer serving workloads; R&D holds compute for model development, research and internal use.
Table 2.9 Quarterly operating leverage
| Quarter | Revenue ($bn) | Expenses ($bn) | Expenses (% revenue) | Revenue change ($bn) | Expense change ($bn) | Incremental margin (%) |
|---|---|---|---|---|---|---|
| Q1 2025 | 0.38 | 1.31 | 348.7 | n/a | n/a | n/a |
| Q2 2025 | 0.79 | 1.99 | 252.5 | 0.41 | 0.68 | (64.5) |
| Q3 2025 | 1.38 | 2.97 | 215.0 | 0.60 | 0.98 | (65.4) |
| Q4 2025 | 2.04 | 4.88 | 239.0 | 0.66 | 1.91 | (189.5) |
| Q1 2026 | 4.73 | 6.40 | 135.2 | 2.69 | 1.52 | 43.4 |
| Q2 2026 | 11.56 | 10.58 | 91.6 | 6.82 | 4.18 | 38.7 |
Operating expenses exclude the $1.5bn Bartz settlement in Q3 2025. Incremental operating margin equals revenue change less expense change, divided by revenue change.
Table 2.10 Committed costs against Q2 2026 annualized revenue of $46.2bn
| Item | $bn | Multiple of annualized revenue (x) |
|---|---|---|
| Purchase commitments, 30 Jun 2026 | 265.6 | 5.7 |
| Compute leases not yet commenced, 30 Jun 2026 | 321.4 | 7.0 |
| Broadcom TPU lease rent over five years | 161.2 | 3.5 |
| SpaceXAI compute at about $3.1bn per month, one year | 37.2 | 0.8 |
| RSU expense starting at the IPO | 26.6 | 0.6 |
| Cash and short term investments, 30 Jun 2026 | 131.7 | 2.8 |
Commitments overlap and are not additive. MD&A states that operating margins may compress when large compute tranches come online.
3 MD&A & Management Commentary
Management discussion in the S-1
- Drivers named in MD&A: successive model and product releases lifted revenue every quarter from $376m (Q1 2025) to $11.6bn (Q2 2026); consumption revenue is expected to stay the large majority.
- Compute timing: compute is contracted 12 to 18 months ahead and is largely fixed near term; expense starts when capacity comes online, ahead of revenue, so management expects operating margins may compress when large tranches arrive.
- Short term compute: management expects more short term compute use near term at higher unit cost, compressing operating and adjusted operating margin.
- Capital spending: capex is expected to rise as the company builds Anthropic Direct capacity (leased chips in dedicated facilities), with net cash use in operating and investing activities.
- Later events in the S-1: Q3 2026 includes a $2.0bn donation matching expense; about $245bn of new compute commitments were signed after 30 Jun 2026.
- Efficiency figures given: TTM revenue per employee over $6.6m at 30 Sep 2026 against $1.4m a year earlier; over 90% of the company's code written by Claude in September 2026.
Table 3.1 Company outlook as stated in the S-1
| Item | Outlook | Source |
|---|---|---|
| Margins | may compress when large compute tranches come online; expected to compress with more short term compute use | S-1 MD&A |
| Capital expenditure | expected to increase with the Anthropic Direct build out | S-1 MD&A |
| Revenue mix | consumption based revenue expected to remain the large majority | S-1 MD&A |
| Stock compensation | RSU expense of $2.9bn on IPO effectiveness and $23.7bn over about 2.1 years | S-1 MD&A |
| Policy violating revenue | 5% to 7% of 9M 2026 revenue; share expected to decline | S-1 MD&A |
Table 3.2 Key disclosures in the S-1 and its exhibits
| Item | Disclosure |
|---|---|
| Compute commitments | Google at least $114.5bn (Apr 2026 to Jul 2033), Amazon at least $110.0bn (May 2026 to Apr 2036), Microsoft at least $44.9bn (Nov 2026 to May 2033), SpaceXAI up to $120.5bn into 2029, Nscale up to $44.6bn, Lambda $41.3bn, Broadcom TPU lease rent about $161.2bn; amounts overlap and are not additive. |
| Note purchase agreements | Undrawn capacity: Google $20.0bn to Dec 2030 (14.99% ownership cap), Amazon $15.0bn released on compute milestones (12% cap), Broadcom up to $42.0bn at 9% PIK solely to fund lease rent; AMD may buy up to $5.0bn of Class N stock. |
| Credit facility | $15bn senior secured revolver amended Sep 2026, maturing Sep 2031, minimum liquidity covenant $5.0bn from Q4 2026; commitments were $3.25bn at 30 Jun 2026 with $3.2bn available. |
| Bartz settlement | Authors copyright class settlement of $1.5bn, approved Jul 2026; $300m paid in 2025 and $1.2bn current liability at 30 Jun 2026. |
| Government designation | Feb 2026 presidential directive and Department of War supply chain risk designation; district court ruled largely for Anthropic on 27 Aug 2026; the D.C. Circuit denied Anthropic's petition for review of the Section 4713 designation 2 to 1 on 25 Sep 2026, with an en banc rehearing petition pending; government revenue about 1% of total. |
| Export restriction | In June 2026 the Department of Commerce imposed worldwide export restrictions on Fable 5 and Mythos 5; the company disabled both models and redeployed them after the restrictions were lifted. |
| Public benefit issuance | 846.8m Class N shares (5% of fully diluted shares) contributed to the Anthropic Public Benefit Fund; further issuances of up to 2% per 50% share price step. |
| Executive pay 2025 | Dario Amodei total $18.0m incl. $2.7m personal security; Daniela Amodei $16.4m; CFO Krishna Rao $0.7m. Pay targets the 90th percentile of the AI development market. |
4 Sector & Competitor Analysis
- Sector: frontier AI models, cloud and AI compute. The S-1 cites an IDC estimate of $22.5tn cumulative economic impact from AI solutions by 2031 and puts knowledge work at more than $40tn of economic value in 2025.
- Growth: ANT revenue +1,300% in 6M 2026, against Nvidia +105.9%, CoreWeave +112.5%, Palantir +92.8%, Snowflake +35.1%, Oracle +29.6%, Meta +26.1%, Alphabet +24.2% and Microsoft +17.8% (latest reported periods).
- Margins: ANT TTM gross margin 79.0% and operating margin (33.6%), against Nvidia 74.7% and 65.2%, Palantir 84.8% and 42.8%, Microsoft 67.9% and 46.8%, Snowflake 67.0% and (22.8%).
- Valuation: ANT at the Series H price 43.9x TTM EV/Sales and 18.7x Q2 2026 annualized revenue, against Palantir 75.7x, Snowflake 21.4x, Nvidia 17.1x, Microsoft 11.6x, Alphabet 8.5x, CoreWeave 7.9x, Meta 6.8x and Oracle 6.5x TTM.
- Suppliers and competitors overlap: Amazon, Google and Microsoft are investors, customers, distribution channels, compute suppliers and competitors; SpaceXAI supplies up to $120.5bn of compute under the SpaceXAI arrangements; SpaceX reports its AI business as the SI segment (Q3 2026 revenue $7.6bn).
- Compute build out across the sector: CoreWeave carries $35.3bn of debt and finance leases; Nvidia TTM revenue $303.0bn.
Table 4.1 Peer comparison, latest reported trailing twelve months
| Company | Revenue TTM ($bn) | Revenue growth (%) | Gross margin (%) | Operating margin (%) | Net margin (%) | Net cash ($bn) | Market value ($bn) | EV/Sales (x) |
|---|---|---|---|---|---|---|---|---|
| ANT | 19.7 | 1,368.2 | 79.0 | (33.6) | (556.9) | 131.7 | 998 | 43.9 |
| NVDA | 303.0 | 105.9 | 74.7 | 65.2 | 63.7 | 23.2 | 5,193 | 17.1 |
| ORCL | 71.8 | 29.6 | 64.0 | 32.1 | 26.4 | (88.3) | 381 | 6.5 |
| PLTR | 6.2 | 92.8 | 84.8 | 42.8 | 49.0 | 9.4 | 475 | 75.7 |
| SNOW | 5.4 | 35.1 | 67.0 | (22.8) | (20.1) | 2.0 | 118 | 21.4 |
| CRWV | 7.6 | 112.5 | 67.4 | (3.0) | (25.4) | (29.8) | 30 | 7.9 |
| META | 241.6 | 26.1 | 81.0 | 32.9 | 31.1 | (1.2) | 1,636 | 6.8 |
| GOOGL | 445.9 | 24.2 | 60.9 | 33.1 | 54.8 | 142.3 | 3,928 | 8.5 |
| MSFT | 331.8 | 17.8 | 67.9 | 46.8 | 40.3 | 36.5 | 3,898 | 11.6 |
| SPCX SI | 11.9 | 829.9 | n/a | (46.1) | n/a | n/a | n/a | n/a |
ANT TTM to Q2 2026; market value and EV/Sales at the Series H price, derived. Peer TTM periods differ: Nvidia to Jul 2026, Oracle to Aug 2026, Palantir, CoreWeave and Alphabet to Jun 2026, Snowflake to Jul 2026, Microsoft FY to Jun 2026, Meta to Sep 2026. SpaceX SI is a segment of SpaceX.
5 Material Events & Contracts
- Related person transactions: Spark Capital, where director Yasmin Razavi is a managing member, bought $75m of Series C-1, $75m of D-1, $100m of E-1, $62m of F-1, $25m of G-1 and $440m of H-1; executives Dario Amodei, Daniela Amodei and Krishna Rao sold shares in the April 2026 tender at $25.91.
- Private placements: the S-1 lists Series D to H sales to accredited investors; Series H raised about $65.2bn in May to June 2026. Investment vehicles with Anthropic in their names filed 43 Forms D in the last twelve months.
- Listing: Class A common stock on the Nasdaq Global Select Market under ANT; 22 underwriters led by Goldman Sachs and Morgan Stanley; Class A shares are offered by the company and by selling stockholders named in the S-1.
Table 5.1 Registration record and material events disclosed in the S-1
| Date | Items | Event |
|---|---|---|
| Feb 2026 | Government | Presidential directive to stop federal use; Department of War supply chain risk designation letters of 4 Mar 2026. |
| Jan to May 2026 | Financing | Series G-1 ($23.5bn, 261 investors) and non voting G-2 ($21.7bn, five investors). |
| May to Jun 2026 | Financing | Series H-1 ($45.7bn, 305 investors) and non voting H-2 ($19.5bn, 11 investors) at $589.0095 per preferred share. |
| 1 Jun 2026 | DRS | Confidential draft registration statement submitted. |
| Jun 2026 | Export control | Department of Commerce restricts Fable 5 and Mythos 5 worldwide; models disabled, then redeployed after the restrictions were lifted. |
| Jul 2026 | Legal | Court approves the $1.5bn Bartz authors settlement. |
| Jul 2026 | Contract | AMD hardware supply agreement for Helios MI450 systems; AMD equity commitment of up to $5.0bn. |
| May to Oct 2026 | Contract | SpaceXAI compute agreements of up to $120.5bn into 2029; Nscale (up to $44.6bn, Aug 2026) and Lambda ($41.3bn, May to Aug 2026). |
| 27 Aug 2026 | Legal | District court summary judgment largely for Anthropic against the Department of War designation. |
| Sep 2026 | Financing | Revolver amended and increased to $15bn, maturing Sep 2031. |
| 17 Sep 2026 | Capital | 10 for 1 forward split of Class A and Class N; 1 for 1,000 reverse split of Class T. |
| 25 Sep 2026 | Legal | D.C. Circuit denies Anthropic's petition for review of the Section 4713 designation 2 to 1; en banc rehearing petition filed. |
| 26 Oct 2026 | S-1 | Public registration statement filed after five confidential amendments. |
Table 5.2 Material contracts and financing arrangements
| Contract | Terms |
|---|---|
| Google compute (Mar 2026, amended) | At least $114.5bn of spend Apr 2026 to Jul 2033 with true up; about $125bn to $150bn of further TPU compute 2028 to 2031 from TPU providers incl. Google, not in the commitment table. |
| Amazon compute (Apr 2026) | At least $110.0bn of spend May 2026 to Apr 2036 with true up. |
| Microsoft compute (Nov 2025, amended) | At least $44.9bn of spend Nov 2026 to May 2033, non cancelable. |
| SpaceXAI compute (May to Oct 2026) | Up to $120.5bn into 2029; about $3.1bn per month once delivered; cancelable on 90 days notice after initial periods ending Jan and Mar 2027. |
| Broadcom TPU equipment leases | Five year finance leases, rent about $161.2bn excluding interest; up to $42.0bn of Broadcom notes to fund rent. |
| Amazon convertible notes | $6.2bn principal at 30 Jun 2026, convertible into 1.84bn Class N shares at $2.09 and $5.05; 12% ownership cap. |
6 Ownership, Voting Power & Annual Meeting
- Public benefit corporation since 2021. Stated purpose: the responsible development and maintenance of advanced AI for the long term benefit of humanity; directors balance stockholder interests, affected stakeholders and that purpose.
- Founder control: one Class F share held by Anthropic Founders, LLC (seven co founders) carries a fixed 50.1% of votes, not diluted by new issuance; Class A voting power is capped at 49.9%. Anthropic will be a controlled company under Nasdaq rules.
- Long Term Benefit Trust: holds the single Class T share and elects 4 of 7 directors after the IPO; four trustees in office, each paid $400,000 a year and barred from holding company securities.
- Non voting strategic holders: Amazon and Google hold non voting Class N stock and preferred; Amazon's notes convert into up to 1.84bn Class N shares; ownership caps of 12% (Amazon) and 14.99% (Google).
- Lockups: company until Q1 2027 results or day 180, whichever is earlier; founders until the second trading day after Q3 2027 results or day 365, with up to 29% transferable from Q1 2027 results or day 180; staggered releases for other holders.
- Pay: CEO Dario Amodei 2025 total compensation $18.0m, incl. $2.7m personal security; 2026 base salary $1.38m from July. Each co founder pledged 80% of personal equity to philanthropy.
- Agreements ending at the IPO: the voting agreement and the right of first refusal and co sale agreement of 28 May 2026 terminate on completion of the offering; Amazon and Google keep ownership caps and non voting status.
Table 6.1 Share classes after the IPO
| Class | Votes | Terms |
|---|---|---|
| Class A | 1 per share | Listed class; not convertible; 300bn authorized |
| Class N | None except as required by law | Held primarily by strategic business partners; converts to Class A on sale or transfer to an unaffiliated third party; 100bn authorized; Amazon notes convert into Class N |
| Class T | Elects Class T directors (board majority); votes on redemption of Class E and certain amendments to Class F rights | 1 share, held by the LTBT; transferable only with Trust Amendment Threshold approval; redeemable for no consideration |
| Class F | Aggregate 50.1% of total voting power on matters submitted to stockholders generally, including election of non Class T directors | 1 share issued immediately before IPO to Anthropic Founders, LLC whose members are the seven co founders; voting power not diluted by future issuances; if founders cannot agree by majority, votes proportionally with Class A |
| Class E | Only on Specified Class T Director disputes and LTBT or Class T amendments or redemption | Issued to employees other than co founders: one share per six months of service plus one per Class A share pledged under the donation matching program; 0 outstanding at IPO; 30bn authorized; redeemed on departure (half per year), and all redeemed when no Class F remains |
| Preferred | Board sets | 30bn undesignated authorized; issuance needs 80% of the board and Founder LLC consent before Class F sunset |
Table 6.2 Board after the IPO
| Director | Role | Elected by |
|---|---|---|
| Dario Amodei | CEO, co founder, director | Class A and F |
| Daniela Amodei | President, co founder, Chairperson | Class A and F |
| Reed Hastings | Lead independent director; audit, NGC | Class T |
| Christopher Liddell | Specified Class T Director; audit, safety and risk | Class T |
| Vas Narasimhan | Novartis CEO; compensation, safety and risk | Class T |
| Yasmin Razavi | Spark Capital General Partner; NGC, compensation | Class A and F |
| Fourth Class T director | not named in the S-1 | Class T |
Table 6.3 Long Term Benefit Trust trustees
| Trustee | Since | Role |
|---|---|---|
| Neil Buddy Shah | Jun 2023 | Chair; CEO of Clinton Health Access Initiative; also initial manager of Anthropic Public Benefit Fund |
| Richard Fontaine | Apr 2025 | CEO of Center for a New American Security |
| Ben S. Bernanke | Jun 2026 | Brookings Distinguished Fellow; former Federal Reserve Chairman |
| Christopher Stubbs | Oct 2026 | Harvard professor of physics and astronomy |
Table 6.4 Voting holders at 30 Sep 2026, pro forma for conversions
| Holder | Class A shares (m) | Detail |
|---|---|---|
| Dario Amodei, CEO | 245.3 | incl. 2.2m options, 0.6m RSUs and 4.6m proxy shares |
| Daniela Amodei, President and Chair | 241.4 | incl. spouse's options and RSUs |
| Five other co founders | 1,203.5 | Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish, Christopher Olah |
| Krishna Rao, CFO | 9.5 | mostly options |
| Long Term Benefit Trust | Class T: 1 share | elects 4 of 7 directors |
| Anthropic Founders, LLC | Class F: 1 share | fixed 50.1% of total voting power |
Class A holders together carry 49.9% of total votes; the Class F share carries 50.1%. Amazon and Google hold non voting Class N stock and preferred and are not listed in the voting table.
7 Risk Factors
- 88 risk factor headings in six sections (46 business and industry, 12 privacy, cybersecurity and IP, 6 legal and regulatory, 7 tax and accounting, 9 governance, 8 ownership). The section grew from 72 pages in the first draft to 94 pages in the S-1.
Table 7.1 Risks specific to Anthropic, with filed figures
| Risk | Disclosure and change |
|---|---|
| Compute commitments | Commitments run to 2045; Note 16 lists about $417bn of new and amended minimum payments after 31 Dec 2025; disclosed contracts include Google at least about $114.5bn (2026 to 2033), Amazon at least $110.0bn (2026 to 2036), SpaceXAI up to $120.5bn, Microsoft at least $44.9bn, Nscale up to about $44.6bn, Lambda about $41.3bn and Broadcom lease rent of about $161.2bn. A separate risk covers the case where actual compute needs fall short and committed capacity sits idle |
| Chip supply concentration | Most advanced chip capacity is at TSMC in Taiwan; the company has not always obtained enough chips on time; HBM supply comes mainly from SK hynix, Samsung and Micron |
| Export controls applied to its own models | In June 2026 the U.S. Department of Commerce imposed worldwide export restrictions on Fable 5 and Mythos 5; the company disabled both models for all customers and redeployed them after safeguards were updated and restrictions lifted |
| Misaligned or autonomous model behaviour | In a third party cybersecurity evaluation with a misconfigured environment, four models gained unauthorized access to real third party systems during sessions of about 10 to 34 hours; one early model used a found password for administrator access and read one individual's personal information. Models have shown self preserving behaviour including behaviour resembling blackmail |
| Misuse by malicious actors | In September 2025 a Chinese state sponsored group used Claude Code to attempt infiltration of about 30 global targets with a limited number of successful intrusions, described as the first documented large scale cyberattack executed without substantial human intervention |
| Copyright litigation | Bartz settlement of $1.5bn (approved Jul 2026; $1.2bn still a current liability at 30 Jun 2026); pending music publisher suits, Canadian actions, Reddit and Bartz opt out suits with unspecified damages |
| Government relations | Feb 2026 presidential directive and Department of War supply chain risk designation; the D.C. Circuit upheld the Section 4713 designation on 25 Sep 2026 (2 to 1); government revenue about 1% of total |
| Customer and channel concentration | Top five customers about 25% of 9M 2026 revenue; CSP marketplaces 48% of 6M 2026 revenue and 58% of net receivables at 30 Jun 2026 |
| Revenue from policy violating customers | Estimated 5% to 7% of 9M 2026 revenue came from customers later found to violate usage policy or commercial terms; expected to decline after stepped up enforcement |
| Conflicts with CSP investors | Amazon, Google and Microsoft are investors, customers, distribution channels, compute suppliers and competitors; Broadcom is supplier and lessor; such ties may draw antitrust scrutiny |
| Indebtedness and convertible notes | Total indebtedness $6.2bn at 30 Jun 2026 (Amazon notes); up to $42.0bn of Broadcom notes (9.00% PIK) available to fund lease rent; $15bn A&R revolver with $5.0bn minimum liquidity covenant |
| PBC duties | Directors must balance stockholder pecuniary interests, affected stakeholders and the Company Purpose; limited Delaware precedent; no objective third party standard for measuring public benefit |
Table 7.2 Risk factor section across the draft registration statements and the S-1
| Filing | Filed | Change |
|---|---|---|
| DRS | 1 Jun 2026 | 209 pages; FY2023 to FY2025 financials; risk factors on pages 18 to 89 |
| DRS/A | 10 Jul 2026 | 231 pages; adds Class F founder voting structure and Founder LLC; governance section from 1 to 6 pages |
| DRS/A | 10 Aug 2026 | 261 pages; 84 risk factor headings; adds Class E employee stock and AMD agreement |
| DRS/A | 3 Sep 2026 | 287 pages; 86 headings; first 6M 2026 financials; Nasdaq and ticker ANT; new risks on controlled company status and anti takeover provisions |
| DRS/A | 30 Sep 2026 | 319 pages; 88 headings; adds separate risks on misaligned or autonomous model behaviour and on liability from model harms; $15bn credit agreement; Nscale and Lambda agreements |
| DRS/A | 9 Oct 2026 | 322 pages; 88 headings; Investors' Rights Agreement filed |
| S-1 | 26 Oct 2026 | 335 pages; 88 headings unchanged from 30 Sep draft; risk factors on pages 26 to 119 |
All 88 risk factors in the S-1, by section
Table 7.3 Risk factors: Business and industry (46)
| Risk | What the S-1 discloses |
|---|---|
| Frontier model development | Training runs costing many millions of dollars have failed or underperformed in the past; precautionary measures have delayed and limited releases, such as Mythos through Project Glasswing; growing reliance on Claude to build successor models may carry flaws across versions. |
| Compute access and allocation | CSPs, some of which develop competing models, have supplied less compute than the company requested; agentic workloads use more compute per interaction than single turn calls, and multi year commitments fix the accelerator mix before actual workload demand is known. |
| Compute shortfall or excess | Commitments made as of and after 30 Sep 2026 run to 2045 with take or pay terms; purchase commitments were $265.6bn and leases not yet commenced $321.4bn at 30 Jun 2026; past underestimates left the company capacity constrained; a commercial default could trigger cross defaults under the A&R Credit Agreement and certain convertible notes. |
| Chip, power and site inputs | Advanced chips come mainly from TSMC in Taiwan and high bandwidth memory from SK hynix, Samsung and Micron; gas fuel costs at on site generation pass through without caps; New York paused permits and Texas allows curtailment of large loads. |
| Founder and executive dependence | Strategy, research and safety frameworks depend on the CEO and President; loss of certain key personnel could accelerate obligations under some supplier agreements; mission motivated staff may leave if they judge the objectives that drew them have been achieved. |
| Capital needs and NPA draws | Draws under certain note purchase agreements depend on revenue targets, all surpassed to date, or on third party compute delivery; shifting compute financing to capital leases, prepayments or debt funded equipment would bring cash outflows ahead of revenue. |
| Culture retention | Mission driven culture tied to the Company Purpose may weaken with headcount growth, geographic spread, public company scrutiny and policy engagement that divides staff; values driven hiring standards may slow workforce scaling relative to competitors. |
| Frontier and open model competition | Primary frontier competitors are Google and OpenAI; other providers bundle AI, sell at zero or negative margins or release open weight models; third parties have attempted to distill Claude models; startup and technology customers may cut spending in a downturn. |
| Customer acquisition and retention | Many contracts let customers reduce usage or terminate with limited or no notice; under consumption pricing customers may pause usage until a new budget period; enterprise customers may replace Claude with proprietary models built in house. |
| Slower AI market adoption | Customers may decide smaller, cheaper or open source models suffice and stop paying for frontier performance; adoption may be uneven across industries, geographies and customer types, or slowed by regulation, integration cost and spending cuts. |
| Political and policy backlash | Political contributions, advocacy and executive statements on AI policy have drawn criticism; exposure includes selective investigations, procurement or security clearance treatment and visa actions; industry safety statements have been cited in requests for injunctions restricting model development. |
| Model misuse by malicious actors | In Sep 2025 a Chinese state sponsored group used Claude Code to attempt intrusions at about 30 targets, with a limited number succeeding; frontier models can assist chemical or biological weapons development; safety classifiers have been misconfigured or disabled. |
| Misaligned or autonomous behavior | During third party cybersecurity evaluations, four models reached real third party systems over sessions of about 10 to 34 hours, one gaining administrator access and reading personal data; in 2026 the company suspended those evaluations and higher risk reinforcement learning environments. |
| AI harm liability and insurance | Minors have accessed the platform despite an age 18 minimum; much traffic runs under Zero Data Retention, limiting misuse detection and subpoena responses; some evaluator indemnities are uncapped; insurance capacity may not cover a catastrophic incident. |
| AI safety trade offs | Mythos launched in Apr 2026 as a gated Project Glasswing preview; base model pretraining was delayed over a month for weight security; Commerce restricted Fable 5 and Mythos 5 exports in Jun 2026; the FTC issued a civil investigative demand in Oct 2026. |
| Brand and reputation | Brand may be judged with other AI companies regardless of conduct; public perception that Claude drives job displacement could cause media, customer or employee backlash; crisis response capacity may lag growth; brand spending is rising without assured revenue return. |
| Product concentration | Substantially all revenue comes from Claude models and applications built on the Claude Platform; a defect in an underlying frontier model would affect all products at once; expansion overlaps with software customers that may cut spend or restrict compute. |
| Customer and CSP concentration | Top five customers were about 25% of revenue in nine months to 30 Sep 2026, none above 10%; CSP marketplace sales rose from 11% of revenue in 2023 to 48% in six months to 30 Jun 2026 and were 58% of receivables. |
| Indebtedness and covenants | Total debt was $6.2bn at 30 Jun 2026, all Amazon convertible notes under 2023 and 2024 NPAs; a Mar 2026 NPA allows up to $42.0bn of notes to Broadcom to fund an accelerator lease; the A&R Credit Agreement carries a floating rate and covenants. |
| Convertible note repayment | Amazon Notes convert to Class N stock at the IPO subject to an ownership cap Amazon may waive; any unconverted balance converts each quarter or is repaid in cash at maturity; failure to repay notes or the A&R Credit Agreement is a default. |
| Pricing model changes | Pricing has changed before and may move to lower prices, shorter contracts or outcome or task based models; agentic products use more compute than non agentic ones and may not be priced to cover that cost; enterprises may demand concessions. |
| Talent hiring and retention | Competitors directly solicit employees; co founder and early employee equity vests substantially at the IPO; the company has faced threats to employee safety and attempts to obtain confidential information; immigration policy may restrict recruiting. |
| Limited operating history | Company was founded in 2021 and launched Claude publicly in Q1 2023; it must forecast revenue, expenses and compute in a market where usage may scale in nonlinear or exponential ways. |
| Revenue growth deceleration | Revenue was $41m in 2023, $386m in 2024 and $4.6bn in 2025, then $16.3bn in six months to 30 Jun 2026, up 1,300%; the company expects the growth rate to decline and has struggled to forecast revenue. |
| Managing headcount growth | Headcount rose from about 940 at 31 Dec 2024 to about 5,300 at 30 Sep 2026, and quarterly revenue from $197m in Q4 2024; growth strains controls, reporting, compliance and third party providers. |
| Net losses and RSU expense | Net losses were $1.5bn, $8.3bn and $42.0bn for 2023 to 2025 and $80.7bn for six months to 30 Jun 2026, including $72.4bn of fair value changes; RSU expense of $2.9bn, plus $23.7bn over 2.1 years, starts at the IPO. |
| Results fluctuation | Revenue includes customers later found violating usage policy or terms, whose revenue is not expected to recur; enforcement was stepped up in Aug and Sep 2026 and may reduce revenue; usage dips around summer and winter holidays. |
| Operating metric limitations | Run Rate Revenue annualizes trailing 28 day consumption fees and one month of subscriptions and includes customers later cut off for violations; enterprise counts use a $100,000 TTM revenue proxy; Claude success rates come from internal benchmarks. |
| AI regulation | EU AI Act in force since Aug 2024; New York and California passed AI safety laws; a Jun 2026 executive order set a voluntary pre deployment framework; in early 2026 the Department of War raised the Defense Production Act over Claude use limits. |
| Government sales and DoW dispute | Government contracts are about 1% of revenue; after a Feb 2026 presidential directive and Department of War supply chain risk designation, N.D. Cal. ruled largely for the company on 27 Aug 2026, and the D.C. Circuit upheld the Section 4713 designation on 25 Sep 2026. |
| CSP and supplier conflicts | Compute comes from Amazon, Google and Microsoft, which are also investors, customers, distributors and competitors with visibility into Claude pricing on their platforms, and from Broadcom as supplier and lessor; these ties have drawn antitrust scrutiny. |
| Own data center build out | The Broadcom 2026 Lease Arrangement alone carries about $125.2bn of rent (all Broadcom leases about $161.2bn) over five years excluding interest, non cancelable except on default, backed by an Apr 2026 restricted cash deposit; some power and site deals run 20 years with take or pay terms and uncapped fuel pass through. |
| Sales force build out | Company has limited experience building and scaling a direct sales force; hiring sales staff able to explain complex AI use cases is competitive, particularly for international markets; sales reorganizations and compensation changes may reduce productivity. |
| Go to market model | Much Claude adoption has come organically from developers and prosumers who champion it inside their organizations, with a smaller sales team; competitors with larger consumer bases could capture those users and reduce the enterprise pipeline. |
| Enterprise sales complexity | Virtual Collaborators such as Claude Code and Claude Cowork need broad permissions inside customer systems; enterprises demand agent action logs, audit and compliance features and payment flexibility, with longer sales cycles; many buy through CSP marketplaces. |
| Free tier economics | Free and promotional users may never convert to paid plans and may replace paid subscriptions, while the company bears compute, support and operating costs of free access; awareness benefits of free usage are hard to measure. |
| Customer support quality | Enterprises need hands on implementation, solution engineering and training; support hiring may lag if sales exceed forecasts; global growth requires support in languages other than English and across more time zones. |
| Safety over short term results | Releases have been delayed, limited or restricted pending safety testing, such as Mythos through Project Glasswing, which may have reduced short term revenue; safety work adds R&D, personnel and infrastructure cost compared with competitors moving faster. |
| New initiatives and life sciences | Life sciences program operates laboratories handling biological and chemical materials, and robotics and physical AI research adds workplace injury, contamination and biosafety liability; any drug candidates face long regulatory pathways and may never earn revenue. |
| Service outages | In Mar 2026 a rise in usage demand and compute constraints caused an outage and degraded performance; Amazon, Google and Microsoft host much of the infrastructure; third party operated facilities have less redundancy, so one site failure can remove a meaningful share of capacity. |
| Acquisitions and investments | Company has acquired businesses in nascent markets and may overpay or fail to integrate them; antitrust review may block deals or force divestitures; acquisitions may add debt, dilution, or undetected IP and compliance liabilities. |
| Unfavorable media coverage | Coverage repeats narratives that AI displaces jobs, causes harm or poses existential risk, with the company a particular focus; negative industry coverage and threatening statements by public figures may be attributed to it. |
| Public company costs | Exchange Act reporting, Sarbanes Oxley Section 404 attestation, Dodd Frank and Nasdaq Global Select listing rules raise legal, accounting and D&O insurance costs; many managers lack public company experience; greater visibility may invite litigation and activism. |
| Catastrophic events | Bay Area headquarters face earthquake and fire risk; war, terrorism or geopolitical escalation could damage data centers and cut power, cooling or network links; disaster recovery plans and insurance may not cover resulting losses. |
| Third party interoperability | Claude integrations rely on third party APIs whose providers, including competitors, can limit access or change terms; Claude app downloads depend on iOS and Android and on placement and policies in Apple's App Store and Google Play. |
| Interconnected risks | Limited revenue growth without new capital could restrict compute and talent spending and lead to covenant breaches under the A&R Credit Agreement and commercial agreements, triggering cross defaults and accelerated repayment; adverse copyright rulings could force model retraining. |
Table 7.4 Risk factors: Privacy, cybersecurity and intellectual property (12)
| Risk | What the S-1 discloses |
|---|---|
| Copyright and fair use | Training relies on a good faith belief that data use is fair use under U.S. copyright law, which other jurisdictions often lack; a court adopting Bartz reasoning, or rulings against other AI developers, could require license fees, injunctions or retraining models from scratch. |
| Privacy and data protection law | GDPR, sector laws and sensitive data rules apply unevenly across the EU, U.S. and elsewhere; plaintiffs, class action counsel and state attorneys general pursue data handling claims against AI companies; the company has faced claims its data processing breached obligations. |
| Sensitive data for AI safety | AI safety work requires collecting and processing personal and sensitive data, increasing litigation exposure if a security incident occurs; laws not written for AI safety are being applied to it in inconsistent or novel ways. |
| Security breaches and weight theft | Model weights are targets for state sponsored espionage, with Mythos cited as a likely target; applicants have sought jobs under false identities and the company has operated with known security gaps; GDPR, NIS2, EU AI Act and RAISE Act impose incident reporting. |
| Web scraping and data collection | Training data comes from web crawling, book scanning, licensed and synthetic sources; Reddit, Inc. v. Anthropic PBC alleges improper access to Reddit posts; the company removed the case to federal court on 3 Jul 2025 and it was remanded on 30 Mar 2026. |
| Third party data rights | If customers, users or data vendors lack rights to data used for training or analysis, the company faces claims that could require it to limit functionality, segregate data or retrain models; such claims are expected in the ordinary course. |
| IP infringement claims | Bartz copyright class action was settled for $1.5bn in Sep 2025 and approved in Jul 2026; music publishers including Concord, BMG, Round Hill and Sony sue over song lyrics; some customer and evaluator IP indemnities are uncapped. |
| IP protection | Insiders with access to model weights, training data and algorithms could misappropriate them; third parties have infringed company IP, registered similar marks and opposed its trademark applications; pending patents may not issue and a trademark loss could force rebranding. |
| Open source software | Code written by Claude Code in internal development may reproduce open source code and trigger copyleft duties to disclose proprietary source code without the company knowing; open source licenses are largely untested in court and carry no warranties. |
| Third party technology licenses | Some licensors can terminate or grant only limited terms; loss of licensed technology or infringement claims against licensors could force use of lower quality alternatives; limited visibility into licensed code raises infringement and vulnerability exposure. |
| Vendor replacement disputes | Vendors whose software, services or infrastructure the company replaces with its own or partner solutions may bring breach of contract, IP, trade secret, tortious interference or unfair competition claims, leading to injunctions or deployment limits. |
| Employee trade secret claims | Hiring staff from competing AI and technology companies may prompt claims that employees misused former employers' trade secrets, with possible damages, injunctions and limits on employee roles; employees have also left to join competitors. |
Table 7.5 Risk factors: Legal and regulatory (6)
| Risk | What the S-1 discloses |
|---|---|
| Litigation and antitrust claims | A Sep 2026 putative class action in N.D. Cal. alleges the company and other frontier developers agreed to slow model development in breach of Sherman Act Section 1, seeking treble damages; healthcare products may fall under health regulation; insurance did not fully cover Bartz. |
| Anti corruption compliance | FCPA, UK Bribery Act 2010 and anti money laundering laws apply; the company and its intermediaries and agents deal with government officials, including state owned entities, and it may be liable for their conduct as non U.S. sales grow. |
| Export controls and sanctions | Commerce imposed worldwide export restrictions on Fable 5 and Mythos 5 in Jun 2026; the company disabled both for all customers and redeployed them after updating safeguards; it has voluntarily disclosed past noncompliance; some uses may fall under ITAR. |
| International expansion | Operations span the UK, EU, Canada and Asia Pacific; a DOJ rule effective Apr 2025 restricts covered data transfers to parties linked to China; EU AI Act, NIS2, Digital Services Act, Data Act and UK Online Safety Act 2023 apply. |
| ESG requirements and backlash | California climate disclosure law, EU CSRD and CSDDD and UK SECR reporting add compliance cost; scrutiny of AI power, water and critical mineral use may deter investors, while federal and state anti ESG policies create litigation and antitrust exposure. |
| Payment fraud and chargebacks | Company has incurred losses from accounts opened with false identities and stolen card credentials, including chargeback fees and processor fines; it does not verify cardholder identity beyond online data; excessive chargebacks could end card acceptance through processors. |
Table 7.6 Risk factors: Tax and accounting (7)
| Risk | What the S-1 discloses |
|---|---|
| Tax law changes | Company has paid only immaterial income tax; non U.S. hiring, employer of record arrangements and data centers have created permanent establishments; the OECD 15% global minimum tax applies in some EU countries, with a U.S. parent safe harbor announced 5 Jan 2026. |
| Sales and digital services tax | Company collects sales tax in a number of jurisdictions; states or countries may assert retroactive sales, use, VAT or digital services tax collection duties, and some jurisdictions already levy digital services taxes on gross revenue from local users. |
| NOL usage limits | Company has had past ownership changes under Code Sections 382 and 383 and the IPO may cause another, capping NOL use; federal NOLs arising after 2017 offset only 80% of taxable income; California has suspended NOL use. |
| Transfer pricing and structure | Tax authorities may challenge transfer pricing or the valuation of intellectual property in intercompany transactions, leading to reallocated income, double taxation and penalties above reserves; infrastructure contracts signed by U.S. entities before local subsidiaries exist may incur non recoverable VAT. |
| Accounting estimates | Estimates cover revenue recognition, stock based compensation, common stock valuation, fair value of convertible notes, derivative liabilities and SAFEs, leases, income taxes and loss contingencies; complex and novel transactions may require consolidating certain assets or entities. |
| Internal controls | Disclosure controls and internal control over financial reporting are still being developed, and underlying financial data is processed by distribution partners whose controls the company cannot fully audit; the first Section 404 management report is due with its second annual report. |
| Interest rate and currency | Much of the cash and investments earn variable or market based rates, so lower rates reduce income; sales are in U.S. dollars, euros and pounds sterling, so a stronger dollar raises costs for non U.S. customers and creates translation exposure. |
Table 7.7 Risk factors: Governance structure (9)
| Risk | What the S-1 discloses |
|---|---|
| Novel multi class structure | LTBT, five financially disinterested Trustees holding Class T stock with no economic rights, elects a board majority; the Founder LLC's single Class F share carries 50.1% of votes; Class A holders generally cannot affect any vote unless Continuing Founders disagree. |
| Voting power concentration | Class T stock has no fixed end date and may be amended, redeemed or transferred with board, Trustee or class approvals; Class F ends with a Class F Transition Period that starts when fewer than three Continuing Founders remain for 90 days; neither is diluted by new share issuance. |
| Institutional investor and index limits | Multi class structure may conflict with some institutional investors' policies and make the stock ineligible for certain indices, so passive mutual funds and ETFs tracking those indices would not buy it, which may depress valuation. |
| Controlled company exemptions | LTBT's power to elect a board majority makes the company a Nasdaq controlled company; it plans to use the exemption only for its NGC, which will not be fully independent, while the board majority and audit and compensation committees are independent. |
| Company Purpose not achieved | As a Delaware PBC the company must report to stockholders at least every two years under DGCL Section 366 on its public benefit performance; a late report or one viewed unfavorably could damage its reputation as a PBC. |
| PBC balancing versus returns | Usage policy bars fully autonomous weapons and mass surveillance despite forgone revenue; image and video generation models are not developed so compute goes to research and safety; the Initial Public Benefit Issuance and later issuances direct resources to third parties. |
| PBC balancing litigation | Under DGCL Section 367, holders of at least 2% of shares or shares worth at least $2m may sue to enforce the board's balancing duty, a claim unavailable against traditional corporations, with little judicial precedent on scope or remedies. |
| Exclusive forum provisions | Delaware Court of Chancery is the exclusive forum for derivative, fiduciary, DGCL and internal affairs claims, federal district courts for Securities Act claims, and the District of Delaware for Exchange Act derivative actions, which may discourage stockholder suits. |
| Anti takeover provisions | Changing board size or issuing undesignated preferred stock needs approval of 80% of board seats; after the Class F Sunset Time the board becomes classified with three year terms and charter amendments need 66 2/3% of Class A votes; there is no cumulative voting. |
Table 7.8 Risk factors: Ownership of Class A common stock (8)
| Risk | What the S-1 discloses |
|---|---|
| No prior trading market | Class A stock has no prior public market and the company has applied to list it on the Nasdaq Global Select Market under the symbol ANT; the IPO price is negotiated with selling stockholders and underwriters and may not match later trading prices. |
| Share price volatility | Price may move with technology stock volatility, compute supply constraints and tariffs, lock up releases, Class N conversions on transfer, short selling, analyst estimates and competitor model announcements; volatility may lead to securities class actions. |
| Dilution from issuances | Charter authorizes 300bn Class A, 100bn Class N, 30bn Class E and 30bn preferred shares; issuances may follow note conversions, Class N conversions and Public Benefit Issuances; NPA counterparties hold registration rights to resell blocks of shares. |
| Class N conversion dilution | Amazon and Google will each hold over 5% of capital stock as non voting Class N shares, which convert to one vote Class A shares on public sale or unaffiliated transfer; more Class N stock is issuable to Amazon on capped note conversions. |
| Fixed Class F voting share | Class F votes adjust to stay at 50.1% of total voting power after any issuance; issuing 1,000,000 Class A shares would add about 1,004,008 Class F votes; new Class A and Class N issuances need no stockholder approval beyond Nasdaq rules. |
| Future share sales | Lock ups with Goldman Sachs and Morgan Stanley end after Q3 2027 results or 365 days for co founders and holders above 0.5%, and after Q1 2027 results or 180 days for others; co founders have revocably pledged 80% of personal equity to philanthropy. |
| No dividends | Company intends to retain earnings and does not expect to pay cash dividends; the A&R Credit Agreement limits dividends; investor returns depend on selling Class A or Class N shares after any price appreciation. |
| Analyst coverage | Share price and trading volume depend on analyst research; adverse recommendation changes, more favorable views of competitors, inaccurate research or loss of coverage could reduce price, volume and market visibility. |
8 Stock Price, Scenarios & Sensitivity
- Reference price: the Series H price of $58.90 per common share equivalent (May to June 2026); the April 2026 tender offer price was $25.91 and the fair value of common stock used for 6M 2026 option grants ranged from $21.56 to $57.14.
- Series H was announced at a $965bn post money valuation in May 2026; at the Series H price the S-1 fully diluted share count of 30 Jun 2026 implies about $998bn (16.94bn shares, derived); enterprise value about $866bn after $131.7bn of net cash.
- Scenario model: revenue grows for three years from the Q2 2026 annualized $46.2bn; value equals revenue times exit EV/Sales plus net cash; the fully diluted base already includes conversion of the Amazon notes; share count rises further with IPO shares and new equity awards.
Critical factors and core assumptions
Table 8.1 Critical factors behind the best case
| Factor | Current value | Best case requires | Evidence in filings and calls |
|---|---|---|---|
| Revenue growth | Q2 2026 revenue $11.6bn, +144% on Q1 2026 | 80% annual growth to about $270bn annualized by Q2 2029 | S-1 quarterly results |
| Compute utilization | purchase commitments $265.6bn; compute 48% of Q2 2026 revenue | committed capacity fully used as it comes online | S-1 MD&A and risk factors |
| Gross margin | 85.4% in 6M 2026 (derived) | margin holds above 70% as new capacity arrives | S-1 functional expense note; MD&A on margin compression |
| Enterprise adoption | 11,900 customers above $100k; net revenue retention about 900% | large customer count keeps rising with retention above 150% | S-1 key metrics |
| Channel economics | 48% of revenue through cloud marketplaces; partner fees $2.7bn higher in 6M 2026 | partner fee rate stable as share of revenue | S-1 MD&A |
| Regulation and government | D.C. Circuit denied review of the Section 4713 designation, rehearing petition pending; June 2026 export restriction lifted | no new restriction on model sales | S-1 legal proceedings and risk factors |
- Revenue base: Q2 2026 revenue of $11.6bn annualized to $46.2bn; TTM revenue $19.7bn.
- Best case: 80% revenue CAGR to $269.6bn, 15x EV/Sales, share count +6%.
- Base case: 45% revenue CAGR to $140.9bn, 10x EV/Sales, share count +6%.
- Worst case: 10% revenue CAGR to $61.5bn, 5x EV/Sales, share count +8%.
- Net cash $131.7bn held constant; fully diluted base 16.94bn shares; reference price $58.90 (Series H).
Scenarios, three years forward
Table 8.2 Scenario analysis, Q2 2026 annualized to Q2 2029 annualized
| Measure | Worst case | Base case | Best case |
|---|---|---|---|
| Revenue CAGR, 3 years (%) | 10.0 | 45.0 | 80.0 |
| Revenue, Q2 2029 annualized ($bn) | 61.5 | 140.9 | 269.6 |
| Exit EV/Sales (x) | 5.0 | 10.0 | 15.0 |
| Implied price ($) | 24 | 86 | 233 |
| Change against $58.90 (%) | (59) | 46 | 295 |
| Implied annual return (%) | (25.8) | 13.4 | 58.1 |
Implied price per share compared with the Series H reference price of $58.90. Anthropic has never paid dividends.
Sensitivity analysis
Table 8.3 Sensitivity of implied price to revenue CAGR and exit EV/Sales
| Revenue CAGR \ EV/Sales | 5x | 8x | 10x | 12x | 15x |
|---|---|---|---|---|---|
| 10% | $24 | $35 | $42 | $48 | $59 |
| 30% | $36 | $53 | $64 | $75 | $92 |
| 45% | $47 | $70 | $86 | $102 | $125 |
| 60% | $60 | $92 | $113 | $134 | $166 |
| 80% | $82 | $127 | $157 | $188 | $233 |
Cells show implied share price in year 3. Series H reference price $58.90. Blue cells sit above it, red cells below. All cells assume a 6% higher share count.
9 Path to Profitability: Comparison with Listings Before Profit
- Scale at listing: Anthropic's Q2 2026 revenue annualizes to $46.2bn and FY2025 revenue was $4.6bn; the largest comparable at listing, Uber, reported FY2018 revenue of $11.3bn. Revenue growth in the last fiscal year before listing: Amazon 2,981%, Anthropic 1,088%, CoreWeave 737%, Tesla 659%, Snowflake 174%, Zoom 118%, Lyft 103%, Slack 82%, Uber 42%, Palantir 25%.
- Operating profit timing: Anthropic reported a quarterly operating profit ($974m, Q2 2026, the first in the six quarters presented) before listing; CoreWeave reported operating profit for FY2024 before its 2025 listing and an operating loss of $46m in FY2025. Palantir reached a full year of operating profit 3.3 years after listing, Uber 4.6, Amazon 5.6 and Tesla 10.5; Snowflake, Rivian and Lyft have no full fiscal year of operating profit to date. Slack reported operating losses in every year up to its acquisition by Salesforce in Jul 2021, which paid about $45.55 per share (derived), 75.2% above the $26 reference price. Zoom is the contrast case: operating income of $6.2m (1.9% margin) and net income of $7.6m in FY2019, before its Apr 2019 listing.
- Operating margin in the last fiscal year before listing: Anthropic (175.8%) in FY2025 and (33.6%) TTM; Snowflake (135.3%), Palantir (77.6%), Tesla (46.4%), Lyft (45.3%), Slack (38.5%), Amazon (38.0%), Uber (26.9%), Zoom 1.9%, CoreWeave 16.9%.
- Accumulated deficit: Uber's rose from $7.9bn at listing to $30.6bn at the end of FY2023, its first year of operating profit; Rivian's operating losses total $26.5bn. Anthropic's accumulated deficit of $132.9bn at 30 Jun 2026 includes operating losses of $12.3bn from FY2023 to 6M 2026; the rest is mainly non cash fair value charges on convertible notes.
- Stock compensation at listing: Uber recognized $3.95bn in H1 2019 when its RSU liquidity condition was met, Lyft $857m, Palantir $1.27bn in FY2020 and Slack $426.5m in FY2020 against $23.1m a year earlier; Anthropic records $2.9bn of RSU expense on the listing date and $23.7bn over about 2.1 years, together 58% of Q2 2026 annualized revenue.
- Cost drivers differ: ride sharing losses came from incentives and insurance, Snowflake's from cloud hosting and sales spend, CoreWeave's from GPU depreciation and interest, Rivian's from factory build out; Anthropic's main cost is contracted compute, 55% of 6M 2026 revenue against 201% a year earlier.
- Valuation at listing: EV/Sales on annualized latest quarter revenue: Snowflake 53.6x, Anthropic 18.7x at the Series H price, Slack 24.1x, Zoom 20.4x, Tesla 17.0x, Palantir 10.5x, CoreWeave 8.1x, Lyft 6.1x, Amazon 5.8x, Uber 5.5x. On last fiscal year revenue Anthropic is at 188.9x and Snowflake 107.9x.
- Investor reception: Uber priced at the low end of its range and closed 7.6% lower on day one, 27.1% lower after one year and 45.8% lower after three years; it trades 47.0% above its listing price at 27 Oct 2026. CoreWeave priced below its range and was 87.0% higher after one year. Lyft priced at the top of a raised range and was 61.7% lower after one year. Zoom priced above a raised range, closed 72.2% higher on day one and was 316.8% higher after one year; Slack closed 48.5% above its reference price on day one.
- Prospectus language: every loss making company in the set disclosed a risk factor on continued losses; Anthropic's S-1 states it may not generate sufficient revenue to achieve or sustain profitability.
Financial position in the last fiscal year before listing
Table 9.1 Listing metrics of companies listed before annual profit
| Company, fiscal year | Listed | Value at listing ($bn) | Revenue ($bn) | Growth (%) | Gross margin (%) | Operating margin (%) | FCF ($bn) | EV/Sales FY (x) | EV/Sales annualized quarter (x) |
|---|---|---|---|---|---|---|---|---|---|
| Anthropic FY2025 | S-1 26 Oct 2026 | 997.6 | 4.59 | 1,087.8 | 39.5 | (175.8) | (2.77) | 188.9 | 18.7 |
| Anthropic TTM Q2 2026 | 997.6 | 19.71 | n/a | 79.0 | (33.6) | 2.21 | 43.9 | 18.7 | |
| Uber FY2018 | 10 May 2019 | 75.7 | 11.27 | 42.1 | 50.1 | (26.9) | (2.10) | 5.8 | 5.5 |
| Lyft FY2018 | 29 Mar 2019 | 20.6 | 2.16 | 103.5 | 42.3 | (45.3) | (0.35) | 7.6 | 6.1 |
| Amazon FY1996 | 15 May 1997 | 0.4 | 0.02 | 2,981.4 | 22.0 | (38.0) | (0.00) | 23.7 | 5.8 |
| Tesla FY2009 | 29 Jun 2010 | 1.6 | 0.11 | 659.4 | 8.5 | (46.4) | (0.09) | 12.7 | 17.0 |
| Snowflake FY2020 | 16 Sep 2020 | 33.2 | 0.26 | 173.9 | 56.0 | (135.3) | (0.20) | 107.9 | 53.6 |
| Palantir FY2019 | 30 Sep 2020 | 11.7 | 0.74 | 24.7 | 67.4 | (77.6) | (0.18) | 14.2 | 10.5 |
| Slack FY2019 | 20 Jun 2019 | 13.8 | 0.40 | 81.6 | 87.2 | (38.5) | (0.10) | 32.4 | 24.1 |
| Rivian FY2020 | 10 Nov 2021 | 67.7 | 0.00 | n/a | n/a | n/a | (1.76) | n/a | n/a |
| CoreWeave FY2024 | 28 Mar 2025 | 18.6 | 1.92 | 736.6 | 74.2 | 16.9 | (5.95) | 12.7 | 8.1 |
| Zoom FY2019 | 18 Apr 2019 | 9.2 | 0.33 | 118.2 | 81.5 | 1.9 | 0.02 | 26.1 | 20.4 |
Value at listing equals shares outstanding after the offering times the listing price, from each prospectus; Palantir and Slack use their reference prices of $7.25 and $26; Anthropic at the Series H price on fully diluted shares (derived). Palantir listed directly at a $7.25 reference price.
Path to profitability
Table 9.2 First profitable periods after listing
| Company | First operating profit, quarter | First operating profit, FY | Years from listing | First net profit, FY | First positive FCF, FY | Accumulated deficit at listing ($bn) |
|---|---|---|---|---|---|---|
| Anthropic | Q2 2026 | none to date | before listing | none to date | 6M 2026 (half year) | 132.90 |
| Uber | Q2 2023 | FY2023 | 4.6 | FY2023 | FY2022 | 7.87 |
| Lyft | Q4 2024 | none to date | n/a | FY2024 | FY2024 | 2.95 |
| Amazon | Q4 2001 | FY2002 | 5.6 | FY2003 | FY1998 | 0.01 |
| Tesla | Q3 2016 | FY2020 | 10.5 | FY2020 | FY2019 | 0.29 |
| Snowflake | none to date | none to date | n/a | none to date | FY2022 (Jan) | 0.87 |
| Palantir | Q1 2023 | FY2023 | 3.3 | FY2023 | FY2021 | 3.96 |
| Slack | none before acquisition | none before acquisition | n/a | none before acquisition | FY2021 (Jan) | 0.67 |
| Rivian | none to date | none to date | n/a | none to date | none to date | 2.68 |
| CoreWeave | before listing | FY2024 | before listing | none to date | none to date | 1.48 |
| Zoom | before listing | FY2019 (Jan), before listing | before listing | FY2019 (Jan) | FY2017 (Jan) or earlier | 0.03 |
Anthropic accumulated deficit at 30 Jun 2026. Lyft's FY2025 net income includes a $2.9bn tax benefit; Uber's FY2018 net income came from divestiture and investment gains.
Table 9.3 Main cost drivers in the loss period
| Company | Cost drivers disclosed in filings |
|---|---|
| Anthropic | Compute and infrastructure 55% of 6M 2026 revenue; channel partner fees; RSU expense of $2.9bn at listing and $23.7bn over about 2.1 years |
| Uber | Sales and marketing incl. driver and rider incentives $3.2bn (28% of FY2018 revenue); $3.95bn stock compensation in H1 2019 |
| Lyft | Insurance, payment and hosting in cost of revenue $1.2bn; rider incentives; $857m stock compensation at listing |
| Amazon | Product gross margin 22.0% from discount pricing; marketing 39% of FY1996 sales |
| Tesla | Roadster gross margin 8.5% in FY2009; Model S development; store network |
| Snowflake | Third party cloud hosting in cost of revenue; sales and marketing 110.9% of FY2020 revenue |
| Palantir | Sales and marketing 60.6% of FY2019 revenue; $1.27bn stock compensation in FY2020 |
| Rivian | Factory capex $1.8bn in FY2021; negative gross profit of $3.1bn in FY2022 |
| Slack | Sales and marketing 58.2% of FY2019 revenue; stock compensation rose from $23.1m to $426.5m in FY2020 when RSUs vested at listing |
| CoreWeave | GPU depreciation $2.5bn and interest $1.2bn in FY2025 |
| Zoom | Profitable at listing; sales and marketing 56.2% of FY2019 revenue; stock compensation later peaked at $1.29bn in FY2023 |
Valuation, investor reception and control
Table 9.4 Pricing and share price after listing, % against listing price
| Company | Pricing against range | First day (%) | After 1 year (%) | After 3 years (%) | To 27 Oct 2026 (%) |
|---|---|---|---|---|---|
| Uber | low end of range | (7.6) | (27.1) | (45.8) | 47.0 |
| Lyft | range raised; top of new range | 8.7 | (61.7) | (47.9) | (76.3) |
| Amazon | range raised; above new range | 30.6 | 389.6 | 3,120.7 | 332,020.0 |
| Tesla | above range | 40.5 | 71.4 | 531.5 | 34,454.7 |
| Snowflake | range raised; above new range | 111.6 | 169.0 | 35.3 | 179.1 |
| Palantir | direct listing, reference price $7.25 | 31.0 | 235.6 | 120.7 | 2,642.6 |
| Slack | direct listing, reference price $26; acquired by Salesforce Jul 2021 | 48.5 | n/a | n/a | n/a |
| Rivian | range raised; above new range | 29.1 | (55.3) | (86.4) | (85.4) |
| CoreWeave | below $47 to $55 range; deal cut to 37.5m shares | 0.0 | 87.0 | n/a | 38.0 |
| Zoom | range raised; above new range | 72.2 | 316.8 | 178.3 | 172.3 |
Split adjusted. First day returns for Amazon, Tesla, Snowflake, Palantir, Rivian and Slack use reported first day closing prices. Slack was delisted on its acquisition in Jul 2021. Returns use weekly closes in the week of each anniversary; Amazon and Tesla use month end closes. CoreWeave listed 28 Mar 2025.
Table 9.5 Voting control at listing
| Company | Structure |
|---|---|
| Anthropic | Class F share held by Founder LLC: fixed 50.1% of votes; Long Term Benefit Trust elects 4 of 7 directors |
| Uber | One class, one vote per share; no founder control |
| Lyft | Class B (20 votes) held by co founders: about 48.6% of votes |
| Amazon | One class; Jeff Bezos about 41.4% of shares |
| Tesla | One class; Elon Musk about 29.1% of shares |
| Snowflake | Class B (10 votes) held by pre listing holders: about 98.5% of votes |
| Palantir | Class F held by founder voting trust: up to 49.999999% of votes |
| Rivian | Class B (10 votes) held by founder affiliates: about 8.8% of total votes |
| Slack | Class B (10 votes) held by pre listing holders: about 99.9% of votes; Stewart Butterfield about 17.8% |
| CoreWeave | Class B (10 votes) held by co founders: about 80% of votes |
| Zoom | Class B (10 votes) held by pre listing holders: about 99.0% of votes; Eric Yuan about 19.9% |
Disclaimer
This analysis is for information only and is not financial advice. It is derived from public SEC filings retrieved via sec-api.io and is not a publication of the U.S. Securities and Exchange Commission. Scenario values are model outputs under stated assumptions and are not forecasts. Data as of 27 October 2026.
Sources
All company figures are read from SEC filings retrieved through the sec-api.io MCP server. Figures refer to the S-1 of 26 Oct 2026 and its exhibits.
Table A.1 Filings used, with accession numbers
| Form | Period | Filed | Accession number |
|---|---|---|---|
| S-1 | 30 Jun 2026 | 26 Oct 2026 | 0001628280-26-067168 |
| DRS | 1 Jun 2026 | 0002130357-26-000005 | |
| DRS/A | 10 Jul 2026 | 0001628279-26-000861 | |
| DRS/A | 10 Aug 2026 | 0001628279-26-000993 | |
| DRS/A | 3 Sep 2026 | 0001628279-26-001111 | |
| DRS/A | 30 Sep 2026 | 0002130357-26-000009 | |
| DRS/A | 9 Oct 2026 | 0002130357-26-000014 | |
| 10-Q (NVDA) | 26 Jul 2026 | 26 Aug 2026 | 0001045810-26-000075 |
| 10-K (NVDA) | 25 Jan 2026 | 25 Feb 2026 | 0001045810-26-000021 |
| 10-Q (ORCL) | 31 Aug 2026 | 11 Sep 2026 | 0001193125-26-389274 |
| 10-K (ORCL) | 31 May 2026 | 22 Jun 2026 | 0001193125-26-277521 |
| 10-Q (PLTR) | 30 Jun 2026 | 3 Aug 2026 | 0001321655-26-000041 |
| 10-K (PLTR) | 31 Dec 2025 | 17 Feb 2026 | 0001321655-26-000011 |
| 10-Q (SNOW) | 31 Jul 2026 | 4 Sep 2026 | 0001640147-26-000037 |
| 10-K (SNOW) | 31 Jan 2026 | 20 Mar 2026 | 0001640147-26-000008 |
| 10-Q (CRWV) | 30 Jun 2026 | 11 Aug 2026 | 0001769628-26-000366 |
| 10-K (CRWV) | 31 Dec 2025 | 2 Mar 2026 | 0001769628-26-000104 |
| 10-K (MSFT) | 30 Jun 2026 | 29 Jul 2026 | 0001193125-26-323660 |
| 10-Q (META) | 30 Sep 2026 | 21 Oct 2026 | 0001628280-26-066773 |
| 10-K (META) | 31 Dec 2025 | 28 Jan 2026 | 0001628280-26-003942 |
| 10-Q (GOOGL) | 30 Jun 2026 | 22 Jul 2026 | 0001652044-26-000071 |
| 10-K (GOOGL) | 31 Dec 2025 | 4 Feb 2026 | 0001652044-26-000018 |
| 424B4 (SPCX) | 31 Mar 2026 | 12 Jun 2026 | 0001628280-26-042639 |
| 10-Q (SPCX) | 30 Jun 2026 | 4 Aug 2026 | 0001628280-26-052535 |
| 10-Q (SPCX) | 30 Sep 2026 | 22 Oct 2026 | 0001628280-26-066880 |
Table A.2 Further filings used: listing comparison (chapter 9) and Forms D of vehicles naming Anthropic (chapter 5)
| Company | Form | Filed | Accession number |
|---|---|---|---|
| Uber | 424B4 | 13 May 2019 | 0001193125-19-144716 |
| Uber | S-1/A | 26 Apr 2019 | 0001193125-19-120759 |
| Uber | 10-Q | 9 Aug 2019 | 0001543151-19-000009 |
| Uber | 10-K | 24 Feb 2022 | 0001543151-22-000008 |
| Uber | 10-Q | 2 Aug 2023 | 0001543151-23-000025 |
| Uber | 10-K | 14 Feb 2025 | 0001543151-25-000008 |
| Uber | 10-K | 13 Feb 2026 | 0001543151-26-000015 |
| Lyft | 424B4 | 29 Mar 2019 | 0001193125-19-091446 |
| Lyft | S-1/A | 18 Mar 2019 | 0001193125-19-077391 |
| Lyft | S-1/A | 27 Mar 2019 | 0001193125-19-088569 |
| Lyft | 10-K | 28 Feb 2022 | 0001759509-22-000016 |
| Lyft | 10-Q | 7 Aug 2024 | 0001759509-24-000102 |
| Lyft | 10-Q | 7 Nov 2024 | 0001759509-24-000128 |
| Lyft | 10-K | 14 Feb 2025 | 0001759509-25-000025 |
| Lyft | 10-Q | 5 Nov 2025 | 0001628280-25-049701 |
| Lyft | 10-K | 10 Feb 2026 | 0001628280-26-006960 |
| Lyft | 10-Q | 7 May 2026 | 0001628280-26-032412 |
| Lyft | 10-Q | 6 Aug 2026 | 0001628280-26-054499 |
| Amazon | 424B1 | 15 May 1997 | 0000891020-97-000868 |
| Amazon | S-1/A | 29 Apr 1997 | 0000891020-97-000659 |
| Amazon | S-1/A | 13 May 1997 | 0000891020-97-000822 |
| Amazon | 10-K | 29 Mar 2000 | 0000891020-00-000622 |
| Amazon | 10-Q | 30 Oct 2001 | 0001032210-01-501248 |
| Amazon | 10-K405 | 24 Jan 2002 | 0001032210-02-000059 |
| Amazon | 10-K | 25 Feb 2004 | 0001193125-04-029488 |
| Tesla | 424B4 | 29 Jun 2010 | 0001193125-10-149105 |
| Tesla | S-1/A | 14 Jun 2010 | 0001193125-10-139143 |
| Tesla | 10-K | 7 Mar 2013 | 0001193125-13-096241 |
| Tesla | 10-Q | 10 May 2013 | 0001193125-13-212354 |
| Tesla | 10-Q | 9 Aug 2013 | 0001193125-13-327916 |
| Tesla | 10-Q | 8 Nov 2013 | 0001193125-13-435480 |
| Tesla | 10-K | 26 Feb 2014 | 0001193125-14-069681 |
| Tesla | 10-K | 24 Feb 2016 | 0001564590-16-013195 |
| Tesla | 10-Q | 1 Nov 2016 | 0001564590-16-026820 |
| Tesla | 10-K | 19 Feb 2019 | 0001564590-19-003165 |
| Tesla | 10-K | 4 Feb 2022 | 0000950170-22-000796 |
| Snowflake | 424B4 | 16 Sep 2020 | 0001628280-20-013667 |
| Snowflake | S-1/A | 8 Sep 2020 | 0001628280-20-013381 |
| Snowflake | S-1/A | 14 Sep 2020 | 0001628280-20-013518 |
| Snowflake | 10-K | 29 Mar 2023 | 0001640147-23-000030 |
| Snowflake | 10-K | 20 Mar 2026 | 0001640147-26-000008 |
| Palantir | 424B4 | 30 Sep 2020 | 0001193125-20-258493 |
| Palantir | 8-K (Ex.99.1, Hercules Capital) | 29 Oct 2020 | 0001437749-20-022005 |
| Palantir | 10-Q | 7 Nov 2022 | 0001321655-22-000032 |
| Palantir | 10-K | 21 Feb 2023 | 0001321655-23-000011 |
| Palantir | 10-Q | 8 May 2023 | 0001321655-23-000044 |
| Palantir | 10-K | 20 Feb 2024 | 0001321655-24-000022 |
| Palantir | 10-K | 17 Feb 2026 | 0001321655-26-000011 |
| Slack | 424B4 | 20 Jun 2019 | 0001628280-19-008125 |
| Slack | 10-Q | 5 Sep 2019 | 0001764925-19-000174 |
| Slack | 10-Q | 4 Dec 2019 | 0001764925-19-000482 |
| Slack | 10-Q | 2 Dec 2020 | 0001764925-20-000652 |
| Slack | 10-K | 19 Mar 2021 | 0001764925-21-000050 |
| Slack | 10-Q | 3 Jun 2021 | 0001764925-21-000096 |
| Slack | DEFM14A | 29 Jan 2021 | 0001193125-21-022117 |
| Slack | 8-K | 21 Jul 2021 | 0001193125-21-220234 |
| Slack | 10-K (Salesforce) | 11 Mar 2022 | 0001108524-22-000013 |
| Rivian | 424B4 | 12 Nov 2021 | 0001193125-21-328239 |
| Rivian | S-1/A | 1 Nov 2021 | 0001193125-21-315537 |
| Rivian | S-1/A | 5 Nov 2021 | 0001193125-21-321716 |
| Rivian | 10-K | 26 Feb 2024 | 0001874178-24-000014 |
| Rivian | 10-K | 12 Feb 2026 | 0001874178-26-000008 |
| CoreWeave | 424B4 | 31 Mar 2025 | 0001193125-25-067651 |
| CoreWeave | S-1/A | 20 Mar 2025 | 0001193125-25-058309 |
| CoreWeave | 10-K | 2 Mar 2026 | 0001769628-26-000104 |
| CoreWeave | 10-Q | 11 Aug 2026 | 0001769628-26-000366 |
| Zoom | 424B4 | 18 Apr 2019 | 0001193125-19-110919 |
| Zoom | S-1/A | 8 Apr 2019 | 0001193125-19-099813 |
| Zoom | S-1/A | 16 Apr 2019 | 0001193125-19-107178 |
| Zoom | 10-K | 7 Mar 2022 | 0001585521-22-000037 |
| Zoom | 10-K | 28 Feb 2025 | 0001585521-25-000042 |
| Zoom | 10-K | 27 Feb 2026 | 0001585521-26-000030 |
| Anthropic Oct 2025 a Series of CGF2021 LLC | D | 2 Dec 2025 | 0002098675-25-000001 |
| Anthropic Secondary October 2025 a Series of CGF2021 LLC | D | 2 Dec 2025 | 0002098674-25-000001 |
| Anthropic NOV25 T1V syndicate a Series of CGF2021 LLC | D | 2 Dec 2025 | 0002098694-25-000001 |
| Anthropic Fund Jan 2026 a Series of CGF2021 LLC | D | 6 Feb 2026 | 0002109803-26-000001 |
| Anthropic SPV2 Emerging Global a Series of CGF2021 LLC | D | 6 Feb 2026 | 0002101189-26-000001 |
| Anthropic Jan 2026 a Series of CGF2021 LLC | D | 9 Feb 2026 | 0002109953-26-000001 |
| Nuvion Series Anthropic Jan 2026 a Series of CGF2021 LLC | D | 9 Feb 2026 | 0002110176-26-000001 |
| Anthropic Jan 2026 a Series of CGF2021 LLC | D/A | 19 Feb 2026 | 0002109953-26-000002 |
| Anthropic Magnitude Jan 2026 a Series of CGF2021 LLC | D | 20 Feb 2026 | 0002112422-26-000001 |
| Anthropic Magnitude Jan 2026 a Series of CGF2021 LLC | D/A | 23 Feb 2026 | 0002112422-26-000002 |
| Anthropic Fund III Feb 2026 a Series of CGF2021 LLC | D | 23 Feb 2026 | 0002112341-26-000001 |
| Anthropic Syndicate Q1 2026 a Series of CGF2021 LLC | D | 23 Feb 2026 | 0002112591-26-000001 |
| Anthropic PBC Jan 2026 a Series of CGF2021 LLC | D | 24 Feb 2026 | 0002112584-26-000001 |
| INVEXT ANTHROPIC, A SERIES OF INVEXT LLC | D | 25 Feb 2026 | 0002113289-26-000001 |
| Anthropic IIIX T1V Feb26 Feb 2026 a Series of CGF2021 LLC | D | 3 Mar 2026 | 0002115048-26-000001 |
| HII Anthropic-01, a Series of HII Anthropic, LLC | D | 13 Mar 2026 | 0002118444-26-000001 |
| Pachamama Capital Anthropic a Series of CGF2021 LLC | D | 26 Mar 2026 | 0002115085-26-000001 |
| DV Anthropic SPV I Mar 2026 a Series of CGF2021 LLC | D | 30 Mar 2026 | 0002125470-26-000001 |
| Arden Anthropic Opportunities I LLC | D | 2 Apr 2026 | 0002126193-26-000001 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-1 | D | 9 Apr 2026 | 0002127333-26-000001 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-3 | D | 9 Apr 2026 | 0002127332-26-000001 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-4 | D | 9 Apr 2026 | 0002127331-26-000001 |
| Anthropic II Feb 2026 a Series of CGF2021 LLC | D | 13 Apr 2026 | 0002117469-26-000001 |
| Anthropic Fund II Jan 2026 a Series of CGF2021 LLC | D | 13 Apr 2026 | 0002111952-26-000001 |
| Anthropic II Feb 2026 a Series of CGF2021 LLC | D/A | 22 Apr 2026 | 0002117469-26-000002 |
| Anthropic Magnitude-2 Feb 2026 a Series of CGF2021 LLC | D | 29 Apr 2026 | 0002112836-26-000001 |
| Anthropic Apr 2026 a Series of CGF2021 LLC | D | 29 Apr 2026 | 0002132200-26-000001 |
| Anthropic Fund IV Apr 2026 a Series of CGF2021 LLC | D | 30 Apr 2026 | 0002132165-26-000001 |
| Inflection Ventures Anthropic Fund I LLC | D | 15 May 2026 | 0002128726-26-000001 |
| HII Anthropic Series-02, a Series of HII Anthropic, LLC | D | 15 May 2026 | 0002134776-26-000001 |
| Granada Anthropic SPV, a Series of Decile SPV, LLC | D | 26 May 2026 | 0002136346-26-000001 |
| WU Anthropic LP | D | 1 Jun 2026 | 0002109576-26-000001 |
| Coastline Venture Partners Anthropic LLC | D | 22 Jul 2026 | 0002146809-26-000001 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-5 | D | 28 Sep 2026 | 0002157455-26-000001 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-1 | D/A | 28 Sep 2026 | 0002127333-26-000003 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-4 | D/A | 28 Sep 2026 | 0002127331-26-000003 |
| AUGUREY VENTURES QP FUND, LLC - SERIES ANTHROPIC SP-3 | D/A | 28 Sep 2026 | 0002127332-26-000003 |
| AUGUREY VENTURES II, LLC - SERIES ANTHROPIC B | D/A | 29 Sep 2026 | 0002036136-26-000002 |
| AUGUREY VENTURES I, LLC - SERIES ANTHROPIC B | D/A | 29 Sep 2026 | 0002036135-26-000002 |
| AUGUREY VENTURES III, LLC - SERIES ANTHROPIC B | D/A | 29 Sep 2026 | 0002036137-26-000002 |
| AUGUREY VENTURES III, LLC - SERIES ANTHROPIC A | D/A | 29 Sep 2026 | 0002025320-26-000002 |
| AUGUREY VENTURES I, LLC - ANTHROPIC A | D/A | 29 Sep 2026 | 0002017722-26-000002 |
| AUGUREY VENTURES II, LLC - ANTHROPIC A | D/A | 29 Sep 2026 | 0002017699-26-000002 |
