The old scaling recipe turned more high-quality human text into lower loss. That trade is still real, but the easy part is ending: future gains depend less on raw token count and more on data quality, synthetic-data mixtures, multimodal transfer, and memory. A modeled look at the regime change.

The AI buildout is not a simple bubble story. It is a race between capability, demand breadth, and the delayed arrival of compute supply.

The first measurable AI productivity signal is not where task substitution predicted. High-exposure sectors are not breaking out, and the early evidence points toward large operational labor pools — though causality is unproven.