The AI Bill Is Too Damn High: Tokenomics 101 — Eugene Cheah, Featherless AI

Why watch

Eugene Cheah presents a four-step framework (measure, right-size, buy capacity, check processes) for reducing AI inference costs at scale, targeting teams spending heavily on LLMs. Covers practical tactics like splitting API keys per team, using classifiers to route to cheaper models, fine-tuning small models for specific tasks, and understanding the speed-vs-cost tradeoff in GPU purchasing. Ideal for platform engineers, DevOps, and technical leads managing AI infrastructure costs.