Skip to content
AI supply chain term

Training (pretraining)

The large, compute-heavy run that creates a base model from scratch.

What it means

Training is the large, compute-heavy process that creates an AI model in the first place. During pretraining, a model is fed enormous amounts of data and its internal parameters are adjusted over and over until it learns useful patterns — a one-off run that can consume thousands of accelerators working together for weeks. It is bursty and capital-intensive: a frontier training run is one of the most demanding computing tasks on earth, and it stresses not just chips but the networking that ties them together and the power feeding the building.

Why it matters to investors

Training is the spike in the AI build-out — the part that drives the giant, lumpy purchases of accelerators, networking and power. It explains the scale of capital spending headlines, but it is front-loaded; the recurring revenue from AI comes later, from inference. Distinguishing the two helps investors judge whether spending is a one-time build or a durable run-rate.

See Training in the AI value chainIts live model score, why it matters, and every company exposed to it.

Companies on this part of the chain

Named to show where the term sits in the AI supply chain — research, not advice, and never a recommendation to buy or sell.

Related terms

See Training in the live AI chain.

THE ENTITY maps every constraint onto one live model — which part is tight now, who owns it, and who gets squeezed when it moves. Plain-English reads you can check.

THE ENTITY is an educational read on the AI supply chain — research, not investment advice. It explains how the chain works and who sits where, never price targets or buy/sell calls.