With the latest leak about Anthropic’s Dr Jeffrey Funk, I examine why the AI bubble may be even more fragile than its biggest advocates have been trying to dismiss. The core concern is that spending on infrastructure is growing faster than revenue, while signs of revenue growth slowing make it harder to justify the scale of investment.
I always feel I need to repeat the point that the question is not simply whether AI is useful. It is whether the business returns can support the enormous valuations and spending built around it, and from what we have seen in Anthropic’s leaked prospectus, the answer is categorically a big no!
We also examine how the tech sector uses the concept of Total Addressable Market (TAM) to justify massive capital injections into companies like Anthropic and OpenAI, despite a lack of significant ROI. Specifically, the risks created by massive, long-term compute commitments can be difficult to escape if expected demand does not arrive.
Jeff contrasts today’s “big market first” strategy with a more cautious approach: prove a product in a focused niche, show that it works, and expand from there. A large market forecast, on its own, is not proof of a sustainable business. This is literally startup 101, and the big AI labs seem to have missed the memo.
We then turn to responsibility and oversight. We argue that incidents described as “rogue AI agents” should not obscure the role of the people and companies that build, deploy, and monitor these systems.
When tools are poorly supervised or difficult to audit, that points to human decisions and engineering practices, not an independent machine “entity” choosing to act. The same accountability question applies to companies making ambitious claims, investors funding risky bets, and policymakers considering public support.
We cannot say exactly when or how an AI market correction might happen, but the potential consequences extend beyond technology companies. Bankruptcies, acquisitions, lawsuits, or losses affecting investors and funds could all be part of the fallout.
Our practical takeaway is to look past the hype and look at how investors are being pushed to bet heavily on an out-of-this-world, lucrative AI future. You need to remember that the current AI investment landscape is fraught with risks, particularly for pension funds and university endowments now being asked to back these firms.
Not to mention Trump’s latest antics, including renaming AI as “Super Intelligence” and floating the idea of public ownership of OpenAI and Anthropic.
Join the conversation and tell us whether you think the AI bubble has now entered the Ponzi scheme territory and whether it should be seen as a scam.
