This may be a really dumb question but what are all these data-centers really for? I can't imagine after all the billions already poured into AI data-centers that there needs to be more. Data-storage and processing from AI ingesting everyone and everything via Flock, Axiom, Genetec, Renkor, Motorola, Milestone, Verkada, GE Current, Ubicquia, CityIQ cameras perhaps?
Social Credit Monitoring?If the real answer is that cloud AI just requires that much to function I have to wonder if there could be some efficiency and energy savings in putting ~90% of work into local models and that last ~10% into frontier models or am I just being naive?