01 · SCALE
Your story
Chips marked ● D are Damodaran's choices. Drag any slider to make it yours.
02 · PROFITABILITY
Target operating margin
03 · REINVESTMENT
Revenue per $1 of compute capital
04 · RISK
Cost of capital
05 · SURVIVAL
Chance it fails
06 · GOVERNANCE
Discount for weak shareholder rights
Five share classes, founders holding 50.1% of votes however many shares are issued, a benefit trust picking the board, and a public benefit charter. Damodaran says this needs "a much larger discount" on his value but doesn't put a number on it, so pick yours. It changes the price you'd pay, not the value above.
Your value of Anthropic's equity
—
vs Damodaran:
vs IPO low end $1.5T:
You
Damodaran base
Worst
Best
IPO price range
The revenue path
How your story scales, 2027–2036
Revenue in $ billions. Hover a bar for the year's margin.
The cash story
Free cash flow after reinvestment
$ billions. Amber years burn cash on compute and growth; blue years throw it off.
What the price implies
The 2036 revenue each IPO price needs
Turn it around: keep your margin, reinvestment, risk and governance choices, and solve for the revenue that justifies the price. This is pricing, not valuing.
What moves the number
Sensitivity around your story
Each bar swings one input and holds the rest. The line is your current value. Damodaran's point holds: revenue and margins swamp the cost of capital.
What the prospectus says
Glass half full, glass half empty
The same S-1 numbers, read by a bull and by a bear. Bulls and bears will both find support here, which is why the story you pick above matters more than the filing.
Growth→ input 01
Half fullFirst-half 2026 revenue was about 14× the first half of 2025. Around 1,900 enterprises now spend over $1M a year on Claude, up from under 100 a year earlier.
Half emptyGrowth slowed in Q3 as OpenAI won share, compute ran short and some clients moved to cheaper open-source models. The top five customers are 25% of revenue.
Profitability→ input 02
Half fullThe company says it made an operating profit in the latest quarter. Compute cost fell from over 250% of quarterly revenue to 48%, and every expense line is shrinking as a share of revenue.
Half emptyThe first-half operating loss was over $8B once you strip out a $72B non-cash charge (the reported loss was $80B). The cost of committed compute is still to hit future income statements.
Reinvestment→ input 03
Half full$600B+ of compute commitments lock in the capacity to grow and close the gap with OpenAI. Turning compute into a fixed cost is also why gross margins look high.
Half emptyLittle of this shows in capex or on the balance sheet yet. Chips age fast and need replacing, and community pushback is making data centres dearer.
Risk→ inputs 04–05
Half full$114B of cash, plus about $100B from the IPO, covers commitments for the next few years. Clients who build Claude into their operations may find it hard to rip out.
Half emptyThe bonds are rated BB+. Fixed compute costs raise operating leverage, AI budgets may prove cyclical, and the company needs growth to meet its obligations.
Governance→ input 06
Half fullAs a public benefit corporation, it can put safety and its stated purpose ahead of short-term profit.
Half emptyFive share classes. Founders keep 50.1% of votes however many shares are issued, and a benefit trust picks the board. Insiders can sell 10% on day one; founders wait 365 days.
The 3P test
Be honest about your story
Possible?
Could this happen at all, without breaking the laws of economics?
Plausible?
Is there a believable business path from today to your 2036?
Probable?
Would you bet your own money that it's the most likely outcome?
Tap your answers.
Under the hood
How the number is built
A 10-year discounted cash flow, built the way Damodaran structures his IPO valuations. Revenue grows from today's base to your 2036 target, with growth fading to the risk-free rate in years 6–10. Margins glide from break-even to your target by year 5. Each year's reinvestment is the extra revenue divided by your sales-to-capital ratio. A terminal value captures everything after 2036. The total is then haircut for the chance of failure, debt (including the present value of compute commitments) comes off, and cash (including IPO proceeds) goes on.
| Fixed input | Value | Source |
|---|---|---|
| Base revenue (trailing 12 months to Jun-26) | $19.7B | S-1 quarterly data |
| Year-1 revenue growth (2027) | 100% | Calibrated |
| Starting operating margin | 0% | Calibrated |
| Years to reach target margin | 5 | Calibrated |
| Tax rate (marginal) | 25% | Standard US assumption |
| Tax losses carried forward | $60B | Estimate |
| Risk-free rate / terminal growth | 3.8% | Calibrated |
| Mature cost of capital (cap) | 7.7% | Calibrated |
| Return on capital after 2036 | 12% | Calibrated |
| Debt incl. PV of compute commitments | ~$500B | Damodaran post |
| Cash ($114B) + IPO proceeds (~$100B) | $214B | Damodaran post |
| Premium / mass-market margins | 40% / 10% | Damodaran post |
The governance discount is applied last, to the equity value, as the most you would pay for a share. It is not part of Damodaran's value; it is the extra discount he says Anthropic's structure requires.
Calibration check. With these settings the engine lands close to Damodaran's published outputs: base vs his $1.475T, worst vs −$373B, best vs $4.04T. "Calibrated" inputs were tuned to the base case first because his full spreadsheet isn't reproduced here; treat them as estimates, not his exact numbers. For the real thing, download his spreadsheet.
Share your story
Send your exact valuation to someone
The link carries your inputs, so whoever opens it sees your scenario, not the default.
Get the next one
New Vatayan Labs dashboards in your inbox
IPO valuations, sector studies and data tools. No spam, one email when something new ships.
Framework and base-case assumptions from Aswath Damodaran's post "The Anthropic Prospectus Arrives: My (but not the) IPO valuation!" (Oct 28, 2026) on Musings on Markets, with his valuation spreadsheet and his shared sheet of reader valuations. Prospectus figures are from Anthropic's S-1 (filed Oct 26, 2026) as summarised in that post. This is an educational tool, not investment advice or a recommendation to buy or sell any security. Outputs are model estimates and will differ from Damodaran's own spreadsheet. Figures in $ billions unless stated.