Skip to content
Briefings are running a touch slower this week while we rebuild the foundations.See roadmap
Lawrence Berkeley National Lab
OrganisationUS

Lawrence Berkeley National Lab

US Department of Energy laboratory at UC Berkeley; publishes authoritative US data centre electricity consumption reports.

Last refreshed: 26 April 2026 · Appears in 1 active topic

Key Question

What does Lawrence Berkeley's latest data say about how much electricity AI is actually using?

Timeline for Lawrence Berkeley National Lab

#116 Apr

IEA: 17% growth, $700B capex run-rate

Data Centres: Boom and Backlash
View full timeline →
Common Questions
How much electricity do US data centres use?
Lawrence Berkeley National Laboratory's 2024 estimates put US data centre electricity consumption at approximately 200 TWh in 2023, around 4% of total national consumption. AI workloads are the fastest-growing component.Source: LBNL
Does the US government track data centre energy use?
Yes. Lawrence Berkeley National Laboratory publishes the main US government-backed estimates of data centre electricity consumption, used by the IEA, EPA, and Congress. There is no mandatory disclosure requirement; LBNL's estimates rely on facility surveys and hardware shipment data.Source: LBNL

Background

Lawrence Berkeley National Laboratory (LBNL) is the US Department of Energy research laboratory at the University of California, Berkeley. Its Data Center Technologies team publishes the authoritative tracking reports on US data centre electricity consumption, used by the IEA, EPA, and Congress as the primary evidence base for energy policy affecting the sector. LBNL's 2024 update found US data centres consumed approximately 200 TWh of electricity in 2023, roughly 4% of national consumption, with AI workloads identified as the fastest-growing sub-category.

The laboratory's reports are directly cited in the IEA's April 2026 global data centre electricity analysis, which restated the 17% global demand growth figure and projected a further 75% rise in hyperscaler capex in 2026. LBNL's methodology — combining facility surveys, utility data, and hardware shipment tracking — is considered the gold standard for consumption estimation in the absence of mandatory disclosure.

LBNL is managed by the University of California for the US Department of Energy. Its researchers have testified before Congress on data centre energy policy and contributed to the EPA's ENERGY STAR programme for data centre equipment.

Source Material