6.6 – Energy case study – AI and Energy

The data centers that power AI have grown in energy consumption at about 12% annually since 2017, which is four times faster than total energy consumption over that period. Data centers currently consume approximately 25% of the power generated in the state of Virginia in the US (the state with the highest level of data-center energy use). And forecasts suggest the pace of growth will accelerate. At the same time, AI excels at optimization and can generate savings throughout the sector. The case study here looks at the current understanding of AI’s role in energy, including data gaps that are to be expected in a largely proprietary landscape: AI developments are largely in the private sector.

First, consider the Executive Summary of the International Energy Association’s 2025 World Energy Outlook Special Report on Energy and AI.^[1] The executive summary avoids more technical language and provides a good synthesis of the report findings. For specifics, read on, into the body of the report itself.

Second is an episode of Inside Climate News Sunday Morning, titled AI’s Massive Energy Demands, with a conversation between executive editor Vernon Loeb and energy reporter Dan Gearino that helps clarify some of the issues discussed in the report.