Capacity shortages at the leading cloud providers began reshaping large companies' AI strategies, with diversification into custom chips, alternative clouds and efficiency measures, and a direct effect on smaller companies' access to frontier models.
For anyone starting out, the effect is about timing before it's about price. Capacity reserved under long contracts leaves the open market, and those without consumption history queue behind those who have it.
That changes what a new company can promise a customer. Commitments on response time and volume depend on contracted capacity, and contracted capacity depends on availability that isn't guaranteed.
The practical answer that emerged in the period was the same at every scale: a fallback chain with more than one supplier, routing by cost, and open models on your own hardware for the predictable part of the load.
It's the software version of the lesson industry learned with components: depending on a single source isn't a choice, it's a risk.