Nate B Jones separates demand for AI services from the question of whether investments in infrastructure and companies will earn their expected returns. He argues that genuine usage can coexist with overpriced assets and overbuilding.
Jones explains why agent workflows can require repeated model calls, increasing inference demandAI inference economics examines the revenue, cost, utilization, pricing, and margins involved in serving trained models. and the importance of power, chipsAI compute infrastructure is the hardware, facilities, networks, power, cooling, storage, and software used to train and run AI models. and efficient routingAI model routing sends each request to a model chosen for that request's complexity, cost, speed, privacy, or other requirements.. He compares potential payback across workloads rather than treating every use case alike.
He recommends examining paid usage, production adoptionAI adoption is the process by which people and organizations begin using AI systems as a sustained part of real products, decisions, and workflows., financing and workflow value. His market outlook and spending estimates are his analysis at the time of recording, not a prediction of investment performance.
Watch on YouTube
