Skip to content

AI Energy & Data Centers

The power, cooling, and footprint behind the AI boom.

The energy and data-center story behind AI covers the power, cooling, and infrastructure required to run models at scale — and the debate over AI’s real environmental footprint as demand for compute accelerates.

Energy estimates need a boundary: one inference request, a training run, a facility or the whole electricity system. The IEA's Energy and AI report separates scenarios for AI adoption, infrastructure bottlenecks and efficiency improvements rather than treating a single forecast as certain. Its cases produce different paths for data-centre electricity demand (IEA, Energy and AI executive summary). When comparing an estimate with your own workload, record what it includes and whether it is measured or projected.

Andy Masley in episode 45 examines the sources behind environmental claims and how assumptions about cooling change the discussion. Jeetu Patel in episode 44 connects AI deployment to the infrastructure underneath it. Is AI bad for the environment? walks through the show's arguments and their limits. Use the AI infrastructure hub for serving and capacity questions, and the AI hardware hub for the chips doing the work.

Start here

  1. Debunking AI's Environmental Panic | Andy Masley 5:24 in transcript

    Compare a published water use claim with the estimate the episode says its source contains.

  2. The First Fully Autonomous AI Attack is Coming | Kris Lovejoy, Kyndryl 17:04 in transcript

    Hear how data center load forecasts and cooling choices enter the power discussion.

Go deeper

Power and data centers

7 episodes

Explainers on this topic

Terms on this topic

Essay on this topic

Guests on this topic

Kris LovejoyPaige BaileyAndy MasleyJeetu PatelAurimas GriciūnasDenny LeeSara HookerCraig WileyVikram Chatterji