Every AI system runs on electricity. The servers that power large language models, the data centres that host cloud infrastructure, the grids that keep it all running — all of it consumes energy, and that consumption is growing fast. The International Energy Agency projects a dramatic increase in energy demand from data centres globally in the years ahead. For grids, for climate commitments, and for countries that want to participate in the AI economy, this is not a background issue. It is a central one.
Yet in most conversations about AI — about its governance, its risks, its potential — the crucial energy question is often absent.
Institutions