Data Centre

Power Usage Effectiveness (PUE)

Definition

PUE is the ratio of total facility energy to IT equipment energy, measuring how efficiently a data centre delivers power to computing equipment. A PUE of 1.0 would mean all energy goes to IT; a PUE of 1.25 means 25% of total energy is consumed by cooling, lighting, and other non-IT infrastructure. Industry average PUE is approximately 1.55. Modern hyperscale and purpose-built GPU facilities achieve 1.1-1.3. Liquid-cooled GPU facilities can approach 1.08-1.15, as liquid cooling is more energy-efficient than air cooling.

Technical Context

PUE varies by season, load, and cooling methodology. A facility may achieve 1.15 in winter with economiser cooling but 1.35 in summer when mechanical chillers are required. Annualised PUE provides a more accurate picture. For GPU deployments, a key nuance is that PUE affects the effective GPUs-per-MW calculation: a facility with PUE 1.25 delivers approximately 80% of its total power to IT equipment, compared to 90%+ at PUE 1.1. This difference scales linearly with facility size and directly impacts revenue per MW.

Advisory Relevance

PUE assumptions directly affect GPU fleet economics. In our due diligence work, we have found that management teams using PUE 1.1 in their business plans are often housed in facilities that actually operate at 1.25-1.4. The resulting overstatement of deployable GPU capacity flows directly into revenue projections.

This glossary is maintained by Disintermediate as a reference for GPU infrastructure professionals, investors, and operators. Each entry reflects terminology as used in active advisory engagements and market intelligence work.

View all termsDiscuss this topic