Every major cloud provider is pouring billions into data centres, and the market keeps asking when the revenue will show up. But that question misses the point. The spending is not a bet on today's revenue. It is a bet on tomorrow's cost curve. The hardware being installed now is being bought at prices that will look absurd in three years, but only if you wait three years. The companies buying it are pricing in the fact that compute gets cheaper, not more expensive. They are not overpaying for the future. They are paying a premium to own the future.
The real risk is not the spending itself. It is the concentration. Every major player is building the same thing, in the same places, for the same customers. If demand stalls, the correction will be brutal. But demand is not stalling. It is being created. Every new model release, every new API, every new agent creates more demand for compute, not less. The companies that are spending are not doing it because they are optimistic. They are doing it because they have no choice. The cost of not building is higher than the cost of building.
The bubble talk is lazy. It assumes bubbles are about excess. This is about necessity. The infrastructure is not a luxury. It is a utility. And utilities do not pop. They get built.
Paul