The enthusiasm around the listing of Anthropicscheduled for November, has encountered its first concrete obstacle. As reported by Reuters, during a meeting with investors at the San Francisco headquarters, on Wednesday 15 October some participants began discussing a valuation around 1,500 billion dollarsa figure lower than the 2,000 billion previously circulated.
The CEO was present at the meeting Dario Amodei and CFO Krishna Rao, who shared third-quarter numbers with potential investors. Revenues rose to 16.8 billion dollarsup 45% compared to the previous quarter. The data, which at first glance was solid, however disappointed some of the audience: in the second quarter revenues had jumped to 11.6 billion, an increase of 147% on the 4.7 billion in the first quarter of the year.
The problem, according to those who were present, is not growth itself but its deceleration. Going from a quarterly growth rate above 100% to one of 45% is seen by some funds as a sign of faster-than-expected maturation for a company that until a few months ago was valued almost exclusively on its exponential trajectories.
Profits are coming, but doubts about costs remain
Not all the news that emerged from the meeting was negative. Anthropic said it had achieved the profitability on an adjusted basis in the third quarter, a milestone that few players in the generative AI sector can boast at this stage. Investor Gavin Baker of Atreides Management spoke of positive free cash flow and accelerating growth seen as early as October, noting that about 75% of the company’s compute spend goes toward model training, not inference.
The question of the financial commitments undertaken to support the expansion remains in the background: Anthropic has signed agreements for more than that 518 billion dollars in ten years intended for cloud services, computing capacity and infrastructure. An enormous figure, which investors themselves are looking at with growing attention at a time when the market is starting to question the use of debt to finance the race for artificial intelligence.
Competitive pressure from Chinese laboratories also weighs heavily, pushing Anthropic to cut prices and reduce revenues linked to model distillation, the practice by which third-party companies train smaller systems starting from the outputs of larger models. It should be said that the revenue figures released during the meeting were not officially confirmed by an Anthropic spokesperson, e.g come from sources who attended the meeting.
The downsizing, if confirmed, does not erase the scope of the operation: even at 1,500 billion dollars, Anthropic’s listing would remain among the largest ever attempted by an artificial intelligence company. But the margin between early autumn expectations and current ones tells a good story of how much the market is starting to demand more rigorous accounts even from the hottest names in the sector.
