Original Coverage & Source Attribution: www.techtimes.com
Anthropic’s quarter-three results — disclosed for the first time in an amended SEC registration statement filed Monday — show revenue still doubling year-over-year at a scale no AI company has previously reached, but also the first concrete evidence that the explosive pace of growth is moderating.
That evidence has already landed on investors: Reuters and the Wall Street Journal reported that prospective IPO buyers have pulled the valuation target from roughly $2 trillion down to $1.5 trillion to $1.6 trillion, a step-down of several hundred billion dollars triggered by the numbers in Monday’s filing. The Wall Street Journal separately reported institutional investor pressure toward the same $1.5 trillion to $1.6 trillion range.
Revenue Still Extraordinary — but the Trajectory Is Shifting
Anthropic generated $17 billion in revenue in the three months ended September 30, up 1,127% from the same quarter a year earlier and up approximately 47% from the $11.6 billion it posted in the second quarter. That 47% sequential gain sounds remarkable — and at this scale it is. It also represents a sharp break from the prior trend: the company had posted 132% sequential growth in the first quarter, 144% in the second quarter, and 48% in the fourth quarter of last year. Three consecutive quarters of acceleration had become Anthropic’s most compelling IPO narrative; the third quarter interrupted it.
For the nine months ended September 30, Anthropic reported $33.24 billion in total revenue, a roughly twelvefold increase from the comparable period in 2025.
The company, its underwriters, and its prospectus attribute the slowdown to at least three factors: capacity constraints during a period when demand outpaced available compute; seasonal softness in business activity during the summer months; and restrictions imposed on customers found to have violated Anthropic’s terms of service. What the company does not quantify is how much each factor contributed, leaving investors to weigh those explanations against an alternative hypothesis: that AI model pricing is under structural pressure and that Anthropic’s revenue ceiling may be lower than a $2 trillion valuation implies.
The pricing pressure is visible in the filing. Anthropic disclosed that price reductions and contractual discounts reduced Q3 revenue by approximately $600 million compared to prior-period rates. Adding that back would put sequential growth closer to 52% — still well below the 144% posted in the prior quarter.
Computing Costs Jumped from 48% to 58% of Revenue in One Quarter
The operating picture beneath the revenue headline is equally material for anyone considering the IPO. Anthropic posted an operating loss of $567 million in the third quarter, reversing the $974 million operating profit it had reported in the second quarter. The company attributes the reversal to two items: more expensive short-term compute, and a $2 billion non-cash charge tied to an employee donation-matching program.
Computing and infrastructure costs reached $10 billion in the third quarter alone, rising from 48% of revenue in Q2 to 58% of revenue in Q3. That movement represents the core structural tension Anthropic is navigating on the way to a public listing: its revenue is growing because usage is growing, but usage is growing precisely because its AI models consume enormous compute resources to answer every query, run every agentic workflow, and process every enterprise task. The more Claude is used, the more it costs to run — and the more it costs to run, the harder it becomes to convert revenue growth into profit.
This dynamic — the inverse of traditional software economics, where the marginal cost of an additional user approaches zero — is well documented in enterprise AI research. Vista Equity Partners’ AI inference economics white paper found that inference costs average 23% of total revenue at scaling-stage AI companies, with AI gross margins typically running 50% to 60% against the 70% to 90% of mature software businesses. Anthropic’s 58% compute cost ratio in the third quarter sits well above that industry average.
On a reported GAAP basis, Anthropic’s $23.29 billion quarterly loss — a figure that will produce alarming headlines and demands significant context. Essentially all of it reflects non-cash write-downs tied to the rising fair value of Anthropic’s convertible notes: as the company’s valuation has risen, its accounting obligations on complex financial instruments from prior fundraising rounds have increased correspondingly. That write-down does not represent cash leaving the business, and it largely disappears upon the IPO. Adjusted operating margins stood at 15% in Q3, down from 19% in Q2 but still positive. As of September 30, Anthropic held $146.6 billion in total liquidity, including cash, short-term investments, and an available credit line.
The $895 Billion Infrastructure Commitment — and What It Actually Means
The amended filing’s single most striking number is not the revenue figure or the net loss: it is the total infrastructure commitment, disclosed at $895.4 billion with roughly $447.5 billion expected to be incurred before the end of 2030. Substantially all of those obligations cover compute and data center infrastructure, and they extend through 2045.
The scale of the step-up deserves specific attention. The original S-1, filed October 26, disclosed approximately $265.6 billion in compute commitments as of June 30. The $895.4 billion figure in the amended filing reflects aggressive deal-making in the intervening months. To put the pace in context: Anthropic’s quarterly revenue grew from $11.6 billion to $17 billion between Q2 and Q3 — a $5.4 billion increase. In roughly the same span, its contractual compute obligations grew by more than $600 billion. The company’s financial obligations are scaling at a rate dozens of times faster than its actual revenue.
One nuance the headline figure obscures: Cape Fear Advisors, analyzing the S-1/A directly, identified $213.8 billion of the total $895.4 billion as conditional or terminable — meaning not all commitments carry the same certainty as fixed-pay obligations. Still, even the confirmed, unconditional portion represents an infrastructure commitment at a scale no technology company has previously made.
The strategy behind those commitments is named in the filing. Anthropic is building out what it calls Anthropic Direct — an asset-light model in which partners secure land, power, and facilities while Anthropic leases the computing capacity inside. The goal is to replace expensive short-term compute capacity — which the filing states currently costs significantly more than long-term locked-in deals — with cheaper dedicated infrastructure as new facilities come online. Short-term compute cost Anthropic roughly $5 billion in the third quarter alone. Under Anthropic Direct, partners finance and build the data centers; Anthropic commits to long-term leases in exchange for lower per-token costs. Prior agreements include arrangements with SpaceX, Amazon Web Services, Google, Microsoft, Broadcom, and Nscale — with SpaceX providing up to $120.5 billion in Nvidia GPU compute through 2029.
Are Investors Right to Cut the Price?
The investor response — Reuters and the Wall Street Journal both reported that discussions have settled around $1.5 trillion to $1.6 trillion, down from the $2 trillion expectation that framed the IPO for months — is a rational application of standard growth-rate discounting. A company whose sequential growth drops from 144% to 47% in one quarter, while its cost structure simultaneously deteriorates, warrants a lower growth multiple.
The question is whether the discounting goes far enough. A Motley Fool analysis published Monday raised the harder version of the concern, arguing that Anthropic may not have a durable moat: competing models only need to be good enough for the work customers actually need done, and once they reach that threshold, customers will optimize for cost rather than capability. The central point is that pricing pressure in AI does not require Anthropic to fall behind technologically — it requires only that alternatives become good enough for most enterprise use cases.
There are benign interpretations of the Q3 data. One investor told Fortune that Anthropic could absorb a period of slower growth and still represent substantial value. Several factors specific to the quarter — capacity constraints, summer seasonality, customer enforcement actions — are real and non-recurring. The company’s net revenue retention remains approximately 900%, meaning existing customers are spending roughly nine times more than they were a year ago, which is an unusually strong retention signal for an enterprise software company of any kind. And 79% of Q3 revenue came from customers outside the company’s top five, up from 64% in the first quarter — a sign that revenue is becoming less concentrated rather than more.
The adjusted operating margin of 15% also indicates the business is not structurally unprofitable on a cash basis, even in a quarter where revenue growth slowed and short-term compute costs were elevated. The long-term bet — that locking in dedicated infrastructure at scale will compress per-token costs and expand margins — is coherent, even if the $895.4 billion in commitments required to make that bet is unprecedented.
What Still Has No Answer: Valuation and Date
The amended S-1 that landed Monday does not contain a share price, a share count, or a target raise amount. Anthropic and its 22-bank underwriting syndicate — led by Goldman Sachs and Morgan Stanley — plan to begin formal investor marketing during the week of November 9. Trading could begin before Thanksgiving on November 26, pending market conditions and regulatory clearance, according to Bloomberg’s IPO timeline reporting.
The revised $1.5 trillion to $1.6 trillion range being discussed by investors would make this IPO one of the largest in US market history, though well below the $86.2 billion raised by SpaceX in its June debut — which achieved a valuation of approximately $1.77 trillion. Whether Anthropic’s offering lands at $1.5 trillion or some other figure will depend on how prospective investors weigh the deceleration signal against the company’s aggregate trajectory: a business that has grown revenue roughly twelvefold in nine months, that holds $146.6 billion in liquidity, and that has built infrastructure commitments extending nearly two decades into the future — all in pursuit of a bet that demand for AI will continue compounding long after the current quarter’s numbers have faded from memory.
Frequently Asked Questions
Why did Anthropic’s revenue growth slow in the third quarter?
Anthropic’s amended prospectus attributes the deceleration from 144% to roughly 47% sequential growth to three factors: capacity constraints when demand exceeded available compute supply, seasonal slowness in enterprise business activity during the summer, and restrictions placed on customers who violated the company’s usage policies. The filing also discloses that price reductions on new model versions and contractual discounts cut revenue by approximately $600 million compared to what it would have been at prior prices. Investors and analysts are debating whether these factors fully explain the slowdown or whether it reflects the beginning of broader pricing pressure in the AI market.
What does Anthropic’s $895 billion compute commitment actually mean for investors?
The $895.4 billion figure represents contractual purchase and lease obligations for computing infrastructure, not money already spent. Of that total, roughly $447.5 billion is expected to be incurred before the end of 2030. These commitments are how Anthropic is locking in dedicated compute capacity at long-term rates it believes will be cheaper than the short-term compute it currently pays. However, Cape Fear Advisors’ S-1/A analysis identified $213.8 billion as conditional or terminable — meaning a meaningful portion of the total is not identical to fixed-pay obligations. For investors, the relevant risk question is whether Anthropic’s revenue growth is durable enough to service these obligations if demand softens or if more efficient AI models arrive faster than expected.
Is the $23 billion net loss in Q3 a signal that Anthropic is in financial trouble?
No — the $23.29 billion GAAP net loss is primarily a non-cash accounting item, not a signal of cash-based financial distress. It reflects the rising fair value of convertible notes from prior fundraising rounds: as Anthropic’s valuation has increased, accounting rules require the company to mark up these instruments, producing enormous paper losses that will largely disappear upon the IPO. The cash-relevant measure is the Q3 adjusted operating loss of $567 million, which represents a reversal from the operating profit reported in Q2 but remains modest relative to the company’s $146.6 billion in total liquidity. The bigger investor question is not about the GAAP loss; it is about whether the adjusted operating margin trend — from 19% in Q2 to 15% in Q3 — will reverse when dedicated infrastructure comes online and reduces per-token costs.
When is Anthropic’s IPO and at what valuation?
Anthropic plans to begin formal IPO marketing during the week of November 9, 2026, with trading potentially starting before Thanksgiving on November 26. The company has not announced a share price, share count, or total raise amount. Investors and underwriters are currently discussing a valuation of approximately $1.5 trillion to $1.6 trillion, according to Reuters and the Wall Street Journal — a significant reduction from the $2 trillion valuation widely discussed a month ago. The final pricing will depend on investor demand, market conditions, and the roadshow that begins next week. No firm date or price has been confirmed by Anthropic or its underwriters.
