I do not need to believe the technology is a bubble to worry about the financing. A useful product can still support a disappointing investment if too much money is spent to deliver it.
The mechanism is straightforward. An operator commits capital now, expects cash later and needs the difference to earn an adequate return. If funding becomes more expensive, either the future cash must increase, the project cost must decline or the investor’s expected return must deteriorate. Enthusiasm does not provide a fourth option.
The exposure differs across the chain. A software company with established cash generation may absorb the pressure more easily than a developer financing equipment and buildings before full utilization. A hardware supplier may look insulated until its customers reconsider the next orders.
My question after a hot inflation release is therefore not “Do I still believe in AI?” It is which assumption moved in the model. How much of the return came from cheap financing? What happens if commissioning is delayed? Can customers support a higher price? Wednesday made those questions more urgent without answering them. I would rather carry that uncertainty explicitly than hide it inside a technological conviction.