Inference Cost Emphasis

Cameron Rohn argues that inference compute, not training, represents the majority of AI investment and will drive massive usage as intelligence layers embed across devices.