Global data centre electricity consumption is projected to more than double between 2024 and 2030, reaching approximately 945 terawatt-hours annually, according to data compiled by the International Energy Agency. The United States hosts the largest share of this capacity, accounting for roughly one-third of global data centre electricity demand.

The IEA attributes much of the projected growth to AI workloads, which are significantly more energy-intensive than traditional cloud computing tasks. Training large language models and running inference at scale requires compute infrastructure that consumes orders of magnitude more power per unit of output compared to conventional server workloads.

In 2023, data centres globally consumed an estimated 415 terawatt-hours of electricity, or approximately 1.6 percent of total global electricity demand. The IEA's central scenario projects that figure rising to between 800 and 1,050 terawatt-hours by 2030, depending on the pace of AI deployment and improvements in hardware efficiency.

United States data centres consumed approximately 200 terawatt-hours in 2023. New hyperscale campuses under construction across the Southeast, Mid-Atlantic, and Midwest are expected to significantly increase that figure over the next five years.

The IEA notes that efficiency improvements in chips, cooling systems, and facility design will partially offset raw demand growth, but are unlikely to prevent a net increase in absolute electricity consumption across the sector.

Source: International Energy Agency -- https://www.iea.org/energy-system/buildings/data-centres