The energy demands of AI data centres become a global concern

The surging energy demands of the artificial-intelligence buildout emerged this year as a major global concern, as the enormous electricity consumption of the data centres powering AI strained power grids, drove up demand and raised difficult questions about the technology's environmental footprint and its implications for climate goals.

The rapid growth of AI had spurred the construction of vast data centres packed with power-hungry chips, and the computing required to train and run large AI models consumed electricity on a scale that was beginning to rival that of entire industries or cities, placing new pressures on energy systems.

In many regions, the demand from data centres contributed to rising electricity consumption after years of relative stability, challenging utilities and grid operators to supply the needed power and raising concerns about reliability, costs and the sources of the additional energy.

The trend posed a dilemma for efforts to combat climate change, as the additional electricity demand risked increasing reliance on fossil fuels or slowing the transition to clean energy, even as technology companies pledged to power their operations with renewable and low-carbon sources.

In response, companies and governments explored a range of solutions, from investing in renewable energy and advanced nuclear power to improving the efficiency of chips and data centres and managing demand, seeking to reconcile the growth of AI with energy and environmental constraints.

The issue underscored the material reality underlying the digital transformation: that the seemingly intangible world of AI rested on vast physical infrastructure with substantial resource demands. As AI continued to expand, its energy appetite became a defining challenge, forcing a reckoning with the costs, as well as the benefits, of the technology, and with the question of how to power an increasingly intelligent world sustainably.