Anthropic’s third-quarter numbers look impressive as revenue jumped 45% from the previous quarter to $16.8 billion, while the Claude developer told prospective investors it was profitable on an adjusted basis.
But the numbers also expose the question likely to dominate its coming IPO.
Revenue had surged from about $4.7 billion in the first quarter to $11.6 billion in the second, meaning growth slowed sharply in Q3 even as expectations around Anthropic’s valuation kept rising.
With Anthropic targeting a valuation above $2 trillion and analysts discussing a year-end revenue run rate near $110 billion, merely growing fast may no longer be enough.
Anthropic’s 45% growth suddenly looks slower
At almost any valuation, 45% sequential revenue growth would look extraordinary.
Anthropic’s problem is the comparison, as revenue rose from $4.7 billion in Q1 to $11.6 billion in Q2, an increase of well over 100%, before climbing to $16.8 billion in Q3.
PitchBook analyst Harrison Rolfes told Business Insider that the business was “still growing, but not growing as fast,” describing Anthropic as reaching an “inflection point.”
He said the slower growth from Q2 to Q3 could shave the valuation the company commands when it goes public.
That is the expectations problem, as Anthropic does not need revenue to decline for investors to become cautious. It only needs expansion to normalise faster than the valuation assumes.
There is also uncertainty around the headline numbers, as analysts noted it is unclear whether Anthropic’s revenue tally is gross or net, while the company has historically counted gross cloud-partner sales even though it does not retain the full amount.
AI usage may rise faster than customer spending
The next question is whether exploding model usage can keep producing equally explosive revenue.
Wells Fargo strategist Ohsung Kwon has highlighted a widening gap between token consumption and customer spending across OpenAI and Anthropic. The analyst noted that usage continues rising while token spending has plateaued.
“Plateauing spend is concerning for frontier labs,” Kwon wrote.
The mechanism is simple, as models are getting cheaper, competition is intensifying and developers can shift workloads to lower-cost alternatives. Customers may therefore consume more AI while paying less for each unit of compute.
For Anthropic, that creates an uncomfortable possibility. Claude adoption can keep expanding while revenue growth slows.
The same dynamic matters across AI. If hyperscalers stop raising capital-spending forecasts aggressively, suppliers that benefited from the infrastructure boom could also lose momentum.
A $2 trillion valuation leaves little room for normalisation
Anthropic is trying to convince investors that growth can outrun the enormous cost of building and operating frontier AI models.
The company is targeting a valuation above $2 trillion and has committed $518 billion over a decade to cloud services, computing and infrastructure. Analysts who met Anthropic executives estimate its annual revenue run rate could reach about $110 billion by year-end.
The bullish case is that adjusted profitability shows operating leverage is emerging.
The bearish case is that investors may pay an enormous valuation at IPO before anyone knows how durable those margins are.
Jefferies analyst Desh Peramunetilleke said AI’s investment appeal is fading as earnings growth slows, funding risks rise and valuations remain elevated. He expects AI earnings growth to decelerate materially through 2028.
The post Anthropic revenue jumps 45% to nearly $17B, but its $100B growth story is cracking appeared first on Invezz