Anthropic’s total customer spend fell by about 20% in the two weeks ended 2 January 2026, compared with the two weeks ended 19 December 2025, according to its amended IPO prospectus filed with the SEC on 10 November. The company ties the drop to “the code freezes that many companies schedule over holidays”.
The same filing says its run-rate measure annualises a trailing 28-day window of consumption fees. By our arithmetic, a 20% dip across half of such a window averages to 10% across all of it, about $8.1 billion on the $81 billion run-rate the filing gives for 31 October. That is an illustration of the mechanism, not a forecast.
Business Pill · Usage-Based Pricing
An evergreen Business Engineer explainer on usage-based pricing, the model in which revenue moves with how much customers use. It teaches the general idea only and does not discuss Anthropic’s filing.
The key insight: Anthropic’s amended prospectus says total customer spend fell about 20% in the last two weeks of 2025, and that its run-rate annualises a trailing 28-day window. By our arithmetic, a dip that size across half the window is about $8.1 billion on the $81 billion October run-rate, an illustration rather than a forecast.
The prospectus says customer usage has “typically exhibited seasonality, particularly around holiday and vacation periods in the summer and winter, when certain workflows slow down and overall customer activity declines”.
It gives one measured example: the last two weeks of 2025. It adds that “our historical growth rates may mask the impact of seasonal trends on our results”, and calls these dynamics “inherent aspects of a pay-for-intelligence revenuemodel”.
Anthropic quarter-on-quarter revenuegrowth, Q2 2025 to Q3 2026, from the quarterly revenue in its S-1/A of 10 November 2026. The fourth quarter of 2025 includes the two weeks in which the filing says total customer spend fell about 20%. Growth rates are our arithmetic.
Why the Run-Rate Is Exposed
The filing says run-rate “annualizes our consumption-based fees for a trailing 28-day period and our subscription-based fees for a single month”. It warns that a figure measured during seasonally lower usage “may differ significantly from Run-Rate Revenue measured at other times”.
The S-1/A puts the run-rate at about $81 billion as of 31 October, up $10 billion from 30 September, as covered in our piece on the $81 billion figure. Our illustration treats the whole figure as 28-day consumption fees; subscriptions are annualised from a single month.
What the Quarterly Revenue Shows
Quarterly revenue in the filing rose from $1,382 million in the third quarter of 2025 to $2,040 million in the fourth, 47.6% growth by our arithmetic. The quarter before grew 75.6% and the quarter after 131.9%.
That fourth quarter was the slowest quarter-on-quarter growth in the series until the third quarter of 2026, which grew 46.7% to $16,956 million, by our arithmetic. The filing also says price reductions offset about $70 million of revenue from the third to the fourth quarter of 2025.
The Other Filed Drags
The filing names two more forces on revenue. Price reductions offset about $600 million from the second to the third quarter of 2026. It also estimates that customers breaking its usage policy or commercial terms made up about 5% to 7% of revenue in the first nine months of 2026, covered in our piece on that estimate.
Posts on X on 11 November relayed a fourth-quarter guidance remark below the October run-rate, attributed by one account to Barclays; as our piece on those posts sets out, they are unconfirmed and we could not read any Barclays note.
The Structural Read
As we read the filing, the holiday effect is a feature of how Anthropic charges: the prospectus says revenue is primarily consumption-based, so when customers freeze code over the holidays, spend falls with the work.
As we read the quarterly table, the effect has so far been hidden inside growth: the fourth quarter of 2025 still grew 47.6% by our arithmetic, and the filing says seasonal effects are expected to become more apparent as growth slows.
Anthropic S-1/A, 10 November 2026
“These dynamics are inherent aspects of a pay-for-intelligence revenuemodel and may be important determinants of our financial results in future periods.”
Three Implications
FOR THE RUN-RATE A run-rate measured over a 28-day window that includes the holidays can sit below one measured in October; a 10% lower window is about $8.1 billion on $81 billion, by our arithmetic.
FOR THE QUARTER The filing’s 20% figure covers two weeks of total spend, not a quarter; the fourth quarter of 2025 still grew 47.6%, by our arithmetic.
FOR THE OTHER DRAGS The filing also names price reductions of about $600 million in the third quarter of 2026 and an estimated 5% to 7% of nine-month revenue from customers breaking its usage policy or commercial terms.
The analysis describes the token as “the unit of cognition the model produces, the unit of compute the data center serves, the unit of price the lab charges, and the unit of value the enterprise extracts”.
As we read it, the holiday dip is that identity showing up in a prospectus: when the token is the unit of price, revenue moves with the work customers do, and code freezes cut both. The analysis predates the filing and does not discuss Anthropic’s seasonality.
What Is Not Established
The filing measures one holiday period, the end of 2025, when revenue was a fraction of today’s. It gives no figure for the 2026 holidays, and a 20% dip in total spend over two weeks is not the same as a 20% drop in a quarter’s revenue.
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 own prospectus says total customer spend fell about 20% over the last two weeks of 2025, and that its run-rate, built on a 28-day window, can move with such dips. By our arithmetic, a dip that size across half the window is about $8.1 billion on an $81 billion run-rate.
A note on sourcing. Every filed figure and quotation here is from Anthropic’s Amendment No. 2 to its Form S-1, filed with the SEC on 10 November 2026, which we read in full on EDGAR. Growth rates and the 28-day window illustration are our arithmetic. The fourth-quarter guidance remarks on X are unconfirmed and no figure from them is used here. 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.
