Pricing the Prodigy: An Independent Valuation of Anthropic’s IPO
Story to Numbers: Valuing Anthropic
My answer: Anthropic is worth about $1.47
trillion. That is roughly $447 a share on today's
expected share count. The reported $1.4 trillion IPO headline is about 5% below
it, so I see the IPO as fairly priced, with early talk of $1.5 to $2
trillion looking too high.
Inspiration and credit. My method is Prof. Aswath Damodaran's story-to-numbers approach,
and I used his open spreadsheet as the calculator. The reading of the
prospectus, my two changed inputs and the reasoning behind every input are my
own work. This is a study, not investment advice.
1. The answer at a glance
|
What |
My result |
|
Value of the company (equity) |
$1.473 trillion |
|
Value per share (tentative share count) |
$447 |
|
Damodaran's value for comparison |
$1.451 trillion ($439 a share) |
|
Reported IPO headline ($70 to $78 a
share) |
about $1.4 trillion |
|
My verdict |
Fairly priced: headline about 5% below my
value |
Sources: my model (AnthropicIPO.xlsx with
my inputs); funding rounds and IPO headline per press reports
2. My four levers, in plain English
|
Lever |
My number |
In simple words |
|
Size of the prize |
$1 trillion revenue in 2036 |
Anthropic wins the premium end of a $5
trillion plus AI work market. About 80% of its revenue is from businesses. |
|
Profit on each dollar |
27.5%
operating margin |
75% premium work at about 40% and 25%
mass market at about 10% gives 32.5%. Regulation takes off 5 points. |
|
Cost of growing |
$2.50 of sales per $1 of capital |
Computing power stays costly in the early
years, then gets cheaper. |
|
Risk |
9.34%
discount rate, 10% failure chance |
A typical large tech business, with the
compute-bill worry priced into the failure chance and debt. |
3.Two Inputs I Changed and Why
I changed two of his inputs. Together they lift my value by about $22 billion (1.6%), from $1.451 trillion to $1.473 trillion.
|
Input |
Damodaran |
Mine |
Effect on value |
My reasoning |
|
Starting cost of capital |
9.74% (tech risk plus a BB+ debt rating) |
9.34% (US
tech median) |
+$43.7 billion |
Anthropic sells to many industries, so I
treat it as a typical large tech firm. I price the compute-bill worry through
the 10% failure chance and by counting commitments as debt. Putting it in the
discount rate as well would count it twice. The rate still falls to 9.13% by
year 10. |
|
Sales per $1 of capital (years 1 to 5) |
3.0 |
2.5 |
−$21.2 billion |
The model's own label for 2.5 is compute
costly near term, easing later. Chips go out of date quickly, new data
centres face local opposition, and my notes say reinvestment never stops. I
return to 3.0 in years 6 to 10 as costs ease. |
Source: my model; Damodaran's published
$1.451 trillion update
The two changes pull in opposite
directions: a calmer view of risk raises value, and a more cautious view of
compute spending lowers it. Applying 2.5 to all ten years would put my value at
about $1.457 trillion.
How sensitive is my value? Company value in $ trillions. Green is $1.5 trillion or more,
yellow is $1.40 to $1.49 trillion, red is below $1.4 trillion.
|
Cost of capital \ Sales per $1 of
capital |
2.0 |
2.5 |
3.0 |
3.5 |
|
9.00% |
1.48 |
1.51 |
1.53 |
1.55 |
|
9.34% |
1.44 |
1.47 |
1.49 |
1.51 |
|
9.74% |
1.40 |
1.43 |
1.45 |
1.47 |
|
10.00% |
1.37 |
1.40 |
1.42 |
1.44 |
My cell is 9.34% with 2.5. Damodaran's is
9.74% with 3.0. No cell reaches the early $2 trillion talk: these two inputs
move value by tens of billions, while the revenue target moves it by trillions.
What I kept the same as Damodaran, and
why
|
Driver |
Both of us |
Why I agree |
|
Revenue in 2036 |
about $1T (42% a year, years 2 to 5) |
1,900 customers now spend over $1 million
a year and many come back. It is plausible, but it is the weakest link. |
|
Operating margin |
32.5% less 5 points = 27.5% |
Compute has become a fixed cost, so
margins widen as sales grow. My notes show operating margin moving from −249%
to +8% in six quarters. |
|
Chance of failure |
10% |
Cash of about $114 billion plus about
$100 billion from the IPO and easy access to capital keep it low. The BB+
rating shows it is not zero. |
|
Regulation |
−5 points of margin |
Rules add cost and cut some revenue, but
they also keep out cheaper new rivals such as Chinese models. |
|
Base-year data |
Includes Q3 2026 |
Q3 growth slowed, so the starting point
should show it. |
Where I add my own judgement
•
Compute promises: my notes record about $650 billion of commitments. I treat them
like debt, which is why debt in my bridge is large.
•
Pay in shares: options and restricted stock dilute owners. I allow for them in the
value per share.
•
Gross versus net revenue: Anthropic reports gross revenue and OpenAI reports net, so I do not
compare their sizes directly.
•
Few big customers: the top five customers are about 25% of revenue. Losing one would
hurt.
•
Margins are not a straight
line: gross margin fell to 23.6% in Q4 2025 before
reaching 77.5% in Q2 2026.
•
Governance: founder voting control is not in the number. I would want a discount
for it.
4. Can Revenue Really Reach $1 Trillion ?
Source: AnthropicIPO.xlsx, Valuation output
sheet
Revenue has to grow about ten times in nine
years. Most of that growth comes in years 3 to 8.
5. Where the value comes from: mostly after year 10
Source: my model; IPO headline per press reports
About three-quarters of the value comes
from what happens after year 10. That is why I watch the $1 trillion revenue
target more closely than any other number.
6. Bear, base, bull: from a failed company to $4 trillion
|
|
Bear |
Base (mine) |
Bull |
|
2036 revenue |
$200 billion |
about $1 trillion |
$1.5 trillion |
|
Operating margin |
10% |
27.5% |
40% |
|
Sales per $1 of capital |
1.0 |
2.5 |
4.0 |
|
Cost of capital |
9.88% |
9.34% |
7.79% |
|
Company value |
−$373 billion |
$1.473 trillion |
$4.04 trillion |
The bear and bull columns are Damodaran's
published extremes. I think both are possible but unlikely. The wide gap is the
real message: this company is priced on a story, not on today's profits.
7. The market in simple terms
|
If AI is… |
Market size |
Who wins |
|
A helper tool |
about $250 billion |
Small prize for everyone |
|
A replacement for some highly paid
white-collar jobs |
$5 trillion or more |
Premium players such as Anthropic |
|
A replacement across all industries,
worldwide |
up to about $10 trillion |
Very large prize, very high bar |
•
OpenAI is the main rival and is gaining share.
•
Open-source and cheaper
models squeeze the low-price end.
•
Funding so far: $30 billion at $380 billion (Feb 2026), then $65 billion at $965
billion (May 2026). The IPO range of $70 to $78 a share opened on 10 November,
with pricing expected 18 to 19 November.
8. Main risks
|
Risk |
What could go wrong |
How I handle it |
|
Revenue falls short |
Everything else weakens with it |
Bear case and the 10% failure chance |
|
Compute bills |
Fixed costs bite if growth stalls |
Counted as debt, and 2.5 sales per $1 of
capital |
|
Competition |
Cheaper models cut prices |
Margin held at 27.5% |
|
Regulation |
Heavier rules after an election |
−5 points of margin |
|
Governance |
Shareholders have little say |
A discount in my own judgement |
9. What I will watch
1.
Quarterly revenue growth
against the 42% path.
2.
Operating margin: about 14% by
year 2 and 27.5% by year 5.
3.
Compute commitments and what
each unit of capacity costs.
4.
Whether big customers stay and
whether OpenAI keeps gaining share.
5.
Any rule that cuts revenue, not
just margin.
10. Sources and credit
1.
Damodaran, A. (28 Oct 2026). The
Anthropic Prospectus Arrives: My (but not the) IPO valuation, Musings on
Markets.
2.
Damodaran, A. (3 Nov 2026). The
Anthropic Feedback Loop, Musings on Markets.
3.
Damodaran, A. Anthropic
IPO valuation spreadsheet and YouTube walkthrough.
4.
Anthropic, PBC. Form
S-1, 26 Oct 2026.
5.
IPO terms and funding history: INDmoney,
FourWeekMBA, The
Paypers.
6.
My own reading notes on the
prospectus.
My value comes from rebuilding Damodaran's
model with my two inputs; it reproduces his $1.451 trillion exactly when his
inputs are used. Press reports of the IPO value disagree (about $1.4 trillion,
$968 billion and $1.02 trillion), probably because of different share counts.
Check the final prospectus after pricing.




Comments
Post a Comment