The system developed by the same American company analyses the risks of a possible listing and advises investors to wait. It is an unusual episode that raises questions about the independence of artificial intelligence.
The story reported today by Alessandro Rigamonti in MilanoFinanza has something paradoxical about it. Anthropic, one of the major players in artificial intelligence, is aiming for the stock market, yet the one warning investors is Claude, the system the company developed and its flagship product. In effect, the AI is examining its own company’s accounts and finding more than one reason for caution.
Claude is said to have analysed the financial prospectus, first recognising extraordinary growth. Reported revenue rises from $4.6 billion in 2025 to $16.3 billion in the first half of 2026, with thousands of major customers and considerable financial resources. But behind those figures are billion-dollar commitments for computing capacity, contracts that far exceed revenue already guaranteed, and a corporate structure that would leave the founders with majority voting control. There are also disputes with the US government and a strong dependence on a small number of large clients.
According to the analysis attributed to Claude, the problem is fairly simple: Anthropic is betting that future growth will be fast enough to cover expenses already planned. At a hypothetical valuation of around one trillion dollars, the share price could already reflect very high expectations. Hence the suggestion not to buy immediately, but to wait for at least the first quarterly report as a listed company.
The part I find most interesting, however, is something else. Before giving its view, Claude declares a conflict of interest: it is an Anthropic product assessing Anthropic itself. Then it proceeds with the analysis, apparently without sparing the company that created it.
This is not, of course, an official rejection of the IPO. It is an assessment generated by a model, reported through a journalistic summary, and the figures require checking against the original prospectus. But the episode remains significant.
We are used to seeing companies invest astronomical sums to make their artificial intelligence systems more powerful. Here, by contrast, the company’s own product, when asked whether it makes sense to invest in the business, essentially says: wait.
I do not know whether Claude is right about Anthropic’s accounts. But I find it interesting that an artificial intelligence system can, at least on paper, question the financial expectations of the people who built it.
Source: MilanoFinanza, Alessandro Rigamonti, 28 October 2026. Figures and listing status should be checked against the original prospectus.
Contacts
