Anthropic’s run-rate revenue was approximately $81 billion on 31 October 2026, up $10 billion from 30 September, according to the second amendment to its IPO prospectus, filed with the SEC on 10 November. The first amendment, filed on 2 November, had left that figure blank.
By our arithmetic, the $10 billion added in October alone is about 2.2 times Anthropic’s entire 2025 revenue of $4,585 million. The two are not the same measure: run-rate is an annualised snapshot, while the 2025 figure is revenue Anthropic recognised over the year.
Business Pill · Run-Rate Revenue
A one-minute explainer of run-rate revenue, the measure in this report. It teaches the general idea only and does not discuss Anthropic.
The key insight: Anthropic’s run-rate revenue was approximately $81 billion on 31 October, up $10 billion from 30 September, its amended prospectus says. The October gain alone is about 2.2 times its 2025 revenue of $4,585 million, by our arithmetic.
The filing says: “our Run-Rate Revenue was approximately $81 billion as of October 31, 2026, up $10 billion from September 30, 2026”. It does not give the 30 September figure on its own. Subtracting, it was about $71 billion, by our arithmetic, so the October gain was about 14%.
Anthropic defines run-rate as the sum of its consumption-based fees in the trailing 28-day period, “multiplied by 13”, and its aggregate monthly subscription fees, “multiplied by 12”. The filing does not split the $81 billion between the two.
The same filing puts third-quarter revenue at $16,956 million, up 1,127% from a year earlier. That is 47% more than the second quarter’s $11,555 million, and about 45 times the $376 million of the first quarter of 2025, by our arithmetic.
Anthropic quarterly revenue and compute and infrastructure expense, Q1 2025 to Q3 2026, from its Form S-1/A filed 10 Nov 2026. Percentages are compute as a share of revenue, as stated in the filing; 45x is our arithmetic.
What $81 Billion Means Per Day
Third-quarter revenue works out to about $184 million a day over 92 days, by our arithmetic. An $81 billion run-rate spread over 365 days is about $222 million a day, about 20% higher. Put another way, the run-rate is about 19% above third-quarter revenue times four ($67.8 billion).
The October gain also lines up with the third quarter’s pace. Revenue growth of 47% in a quarter compounds to about 13.7% a month, by our arithmetic, close to the run-rate’s roughly 14% rise in October. The two are different measures, so the comparison is a guide, not a match.
The Wall Street Journal had reported on 6 November, citing a person familiar with the matter, that the run rate topped $80 billion at the end of October. The filing now puts the figure at approximately $81 billion.
The Caveats in Anthropic’s Own Words
The prospectus says run-rate “is not a substitute for revenue prepared in accordance with GAAP or a forecast of revenue”. It warns that measuring during holiday periods can move the figure, and that “we expect our Run-Rate Revenue growth rate to decline over time”.
It also says run-rate “has included, and we expect will continue to include, amounts from customers that we subsequently determine are using our models, platform, and applications in violation of our usage policy or commercial terms”. It gives no dollar amount for that.
The Cost Side of the Same Quarter
Compute and infrastructure expense was $9,863 million in the third quarter, or 58% of revenue, up from 48% in the second quarter, according to the filing. That followed a fall from 148% a year earlier. The filing says the rise reflects “the delivery of additional compute”, including $5.0 billion of compute under short-term commitments that “currently costs significantly more” than compute bought under multi-year deals.
Between the second and third quarters, revenue rose $5,401 million and compute rose $4,305 million. That is about 80 cents of extra compute for each extra dollar of revenue, against about 32 cents a quarter earlier, by our arithmetic. Anthropic swung from operating income of $974 million in the second quarter to an operating loss of $567 million in the third. Our earlier piece looked at how R&D compares with the cost of serving Claude.
Anthropic says it has purchase commitments and leases with an aggregate committed spend of approximately $900.9 billion, of which approximately $70.7 billion is due in the 12 months after 30 September. That is about 87% of the $81 billion run-rate, by our arithmetic.
Customers Behind the Number
The filing counts more than 11,900 customers with $100,000 or more of revenue over the trailing twelve months, up from more than 950 a year earlier, and puts Net Revenue Retention at approximately 900%. Those customers produced 74% of revenue in the twelve months to 30 September. The other 26% came from approximately 40 million smaller customers.
As we read the filing, the run-rate and the third-quarter cost lines point in two directions at once: revenue kept compounding in October at roughly the third quarter’s monthly pace, by our arithmetic, while compute took a larger share of each new dollar in the quarter.
As we read the caveats, Anthropic itself flags two limits on the $81 billion: it is an annualised snapshot of a 28-day window, and it includes some revenue from customers later found to break its usage policy, an amount the filing does not give.
Anthropic, Form S-1/A, 10 November 2026
“we expect our Run-Rate Revenue growth rate to decline over time”
Three Implications
FOR IPO BUYERS The prospectus gives run-rate at one date, 31 October, and says it is not a forecast of revenue; there is no fourth-quarter guidance.
FOR THE COST LINE Third-quarter compute was 58% of revenue, up from 48%, the filing says, with $5.0 billion of short-term compute that costs more than multi-year capacity.
FOR COMMITMENTS The $70.7 billion of commitments due in the next 12 months is about 87% of the $81 billion run-rate, by our arithmetic.
The framework says: “Inference is a revenue engine: continuous, distributed across millions of applications and agents, incurring marginal cost with every query.”
As we read it, Anthropic’s third quarter shows that marginal cost in the accounts: about 80 cents of extra compute for each extra dollar of revenue, by our arithmetic, against about 32 cents a quarter earlier. The framework predates this filing and does not discuss Anthropic’s figures.
What Is Not Established
The filing gives no split of the $81 billion between consumption and subscriptions, no run-rate after 31 October, no October GAAP revenue and no fourth-quarter guidance. The 30 September run-rate of about $71 billion is our subtraction, not a stated figure. The share of run-rate from customers later found to break its usage policy is not quantified.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Anthropic’s second prospectus amendment puts run-rate revenue at approximately $81 billion on 31 October, up $10 billion in a month. That October gain is about 2.2 times its 2025 revenue, by our arithmetic, while its third-quarter compute bill rose to 58% of revenue from 48%.
A note on sourcing. Run-rate revenue, its definition and caveats, quarterly revenue and compute figures, customer counts and commitments are from Anthropic’s Form S-1/A filed with the SEC on 10 November 2026 (Amendment No. 2); Amendment No. 1 of 2 November left the run-rate blank. Ratios, per-day figures and the implied 30 September run-rate are our arithmetic. Anthropic was not contacted for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
