What it means
Data-center power is the electricity an AI facility can actually draw — increasingly the hardest physical limit on the whole build-out. AI clusters are extraordinarily power-dense, and a site is only as large as the power it can secure and cool. Efficiency is tracked with PUE (Power Usage Effectiveness), the ratio of total facility power to the power that reaches the computers; lower is better. But efficiency only goes so far: beyond a point the constraint is raw availability of electricity and the equipment to deliver it. That is why "AI is a power story" has become a common refrain — the limit moves from chips to megawatts.
Why it matters to investors
Power is the constraint that pushes the AI trade beyond semiconductors into electricity, equipment and infrastructure. When megawatts — not chips — gate the next data center, the beneficiaries widen to include power-equipment makers, generation and the broader grid build-out, reframing AI as an energy-infrastructure story as much as a silicon one.
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 Data-center power 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.