AI company Anthropic reportedly told investors at a private meeting that its revenue totaled $16.8 billion in the third quarter of 2026. That was up about 45% from the previous quarter and more than 3.6 times its full-year 2025 revenue. But growth has cooled from the more than 140% increase in the second quarter, putting the spotlight on how the slowdown could affect the valuation of its IPO, planned for November.
Revenue keeps growing, but growth rates cool
At $16.8 billion, revenue for a single quarter is already enormous. The fact that it is more than 3.6 times the total for full-year 2025 shows how the business has moved up a level in just over a year.
But investors are looking at more than just scale. In the second quarter, demand for Claude Code, Anthropic’s developer tool, surged, driving growth of more than 140% from the previous quarter. That has now slowed to 45%. Put the figures side by side, and the slowdown is unmistakable.
One explanation is that workloads tend to decline over the summer, which can also reduce consumption of “tokens,” the units used to measure the volume of text processed by AI. In addition, OpenAI launched the powerful competing product Astra in September, while Anthropic reportedly did not strike back with its rival model Opus 5.5 until late September. It appears to have been a quarter shaped by both seasonal factors and competition.
Revenue scale and growth momentum are measured on separate gauges. Even as the speed drops, the distance already covered is on a completely different scale. At an IPO, investors will scrutinize the needle on that “speedometer.”
The economics of inference emerge as AI’s next business focus
One notable aspect of the report is the profit figures. Anthropic’s adjusted operating margin was 11% from the start of the year through the third quarter. It reportedly posted adjusted profits for a second consecutive quarter and also generated positive cash flow.
In the third quarter, the company reportedly generated $1.70 in revenue for every dollar it spent on computing resources, such as servers and GPUs. The figure suggests it can make money from inference—the process of sending a question to an AI model and getting an answer. Even as price competition continues, models that have completed training can generate profits. That would be encouraging news for the AI industry.
However, this is an “adjusted” measure that excludes major expenses such as stock-based compensation. Model training and labor costs remain substantial, and open-weight models—whose weights are publicly available for anyone to use—could still replace paid models such as Claude. A reasonable assessment is that a path to profitability is now coming into view.
Anti-distillation measures and price competition could shake up revenue
Another issue is “distillation,” a technique in which other companies collect large volumes of output from existing AI systems, train on their behavior and use that to build their own models. Anthropic reportedly estimates that 5% to 7% of its revenue is related to distillation.
The company has repeatedly raised concerns about imitation by Chinese AI companies and others, but recently introduced an effective countermeasure. It reportedly told investors that the move is expected to weigh on revenue in the coming quarter.
It is a decision that will come at a cost in the short term. Over the longer term, however, it could make it harder for competitors to match Claude’s quality. Giving up a few percentage points now to protect the asset of model superiority is a bold move just ahead of an IPO, but it may also reflect a focus on the business’s long-term resilience.
Pressure from price competition is also significant. Price-cutting battles with OpenAI, Google and others have continued for months, while open models are closing the performance gap. When revenue growth slows, this competitive landscape has a greater impact on valuation.
IPO valuation hinges on the durability of the business beyond revenue
The IPO is reportedly planned for November, before Thanksgiving. It is expected to be the biggest test yet of investor enthusiasm for the AI boom.
The company’s valuation in private markets reached $965 billion in May, up from less than $20 billion two years ago. Some banking insiders reportedly floated a $2 trillion valuation at the time of the IPO.
But some investors were reportedly disappointed by the latest figures. The company also did not provide an annualized revenue figure for September—the latest revenue run rate projected over a full year—or a full-year outlook, apparently reinforcing cautious views. Its annualized revenue was reported at $65 billion as of July. Several investors believe the lead banks could lower their valuation expectations to below $1.5 trillion.
The market environment is not particularly favorable, either. Inflation remains elevated and borrowing costs are rising. Public opinion on AI is more negative than positive, and unresolved cybersecurity risks remain.
Even so, as models improve, AI is becoming more deeply embedded in everyday office work, and usage is still trending upward rather than declining. The question is whether revenue growth can sustain the profits and competitive strength needed to support a massive valuation, even as growth slows. The November IPO will be the moment of truth. Valuation estimates and investor reactions are based on media reports and are not confirmed information.
Comments