Data-centre electricity demand
node
The electricity drawn by data centres, which is where the compute trend becomes a physical claim on a shared system. The International Energy Agency's Energy and AI report of April 2025 put global data-centre consumption at about 415 terawatt-hours in 2024, roughly 1.5% of world electricity, and projected it to more than double to around 945 TWh by 2030. The IEA's subsequent update reports that demand rose about 17% in 2025 to roughly 485 TWh — against about 3% growth in global electricity demand overall — with AI-focused facilities growing far faster, and projects roughly 950 TWh and about 3% of world electricity by 2030. Two cautions belong with those numbers. Data centres are not only AI, and the AI share is estimated rather than metered. And the constraint that actually bites is local and temporal rather than global: a grid connection in one county, at one time of day, is what determines whether a facility can be built, which is why the global percentage is a poor guide to where the difficulty is.
A node is not a place. Drawing it on a map would assert something about the world that no stored fact supports.
Read in · 1
Evidence · 2
Timeline
No dated observations are stored for this object. Atlas shows what was observed and when — it does not infer a history.
Connections · 1
- Growth in frontier training computecaused_by
Assembled narrative · 1
Assembled from 20 blocks · 2 evidence · 12 related
- Story
- The electricity drawn by data centres, which is where the compute trend becomes a physical claim on a shared system. The International Energy Agency's Energy and AI report of April 2025 put global data-centre consumption at about 415 terawatt-hours in 2024, roughly 1.5% of world electricity, and projected it to more than double to around 945 TWh by 2030. The IEA's subsequent update reports that demand rose about 17% in 2025 to roughly 485 TWh — against about 3% growth in global electricity demand overall — with AI-focused facilities growing far faster, and projects roughly 950 TWh and about 3% of world electricity by 2030. Two cautions belong with those numbers. Data centres are not only AI, and the AI share is estimated rather than metered. And the constraint that actually bites is local and temporal rather than global: a grid connection in one county, at one time of day, is what determines whether a facility can be built, which is why the global percentage is a poor guide to where the difficulty is.
- Knowledge
- Growth in frontier training compute
- Data-centre electricity demand
- Connections
- International Energy Agency
- Growth in frontier training compute
- Northern Virginia data-centre corridor
- The IEA publishes its first global analysis of energy and AI
- The IEA reports data-centre electricity demand up about 17% in 2025
- The transformer architecture
- Training compute
- Data-centre electricity demand
- Northern Virginia data-centre corridor
- Epoch AI
- The IEA publishes its first global analysis of energy and AI
- Evidence
- Data centres accounted for about 415 TWh, or roughly 1.5% of world electricity, in 2024, projected by the IEA to more than double to about 945 TWh by 2030; the IEA's subsequent update reports demand rising about 17% in 2025 to roughly 485 TWh against about 3% growth in global electricity demand, with AI-focused data centres growing considerably faster, and projects roughly 950 TWh and about 3% of world electricity by 2030. V55 verification basis: iea.org was not fetched. The 2024 figures and the 945 TWh projection were carried in search results attributing them explicitly to the April 2025 Energy and AI report, and are taken as verified. The 2025 update figures were carried with attribution to IEA news and to Key Questions on Energy and AI, but the publication date of that update was NOT resolved in this run and the AI-specific share of data-centre demand is an estimate in every source retrieved.
- Loudoun County, and the Ashburn–Sterling corridor within it, hosts the densest concentration of data-centre capacity in the world by the county's own account, and county material addresses growth constraints arising from electricity supply. V55 verification basis: the county documents were NOT fetched; only their titles and general subject were returned by search. No capacity figure is carried into this pack because retrieved third-party figures for the corridor differ by factors of several — installed IT load, contracted capacity and planned capacity are conflated across sources — and the widely repeated claim that a large share of global internet traffic passes through the county could not be traced to any authoritative source. This record names where a curator should check, not what was checked.
Observed changes · 0
No public Signals are attached to this object. Signals show what changed and when it was observed — never a direction or a rank.
Actions
/atlas?object=RESOURCE_DATA_CENTRE_ENERGY&experience=RESOURCE_DATA_CENTRE_ENERGY