The national number and the neighbourhood question
How much electricity does the data centre near you use? Official statistics can help explain the scale of demand, but answering that local question requires more than finding a large number beside a map pin. This evergreen explainer uses previously published evidence, checked on 7 October 2026; it is not a report of a new outage or statistical release.
DESNZ estimates that the data centres within its study consumed 4.5 terawatt-hours of grid electricity in Great Britain in 2024, around 2% of the total. In Slough, the corresponding figures were 1.3 TWh and 65% of the area's grid electricity consumption. The report was published on 30 June 2026 and is classed as Official Statistics in Development. [1]
Those percentages answer different geographic questions. A modest national share can coexist with concentrated local demand. Neither percentage identifies which installation is drawing electricity during an interruption today. Nor does either establish that a data centre caused a particular household's power cut.
Check what is included
The study matches identified facilities to electricity meters and excludes enterprise data centres used for an organisation's own internal needs. It also describes uncertainty from missing facilities, unmatched meters and buildings shared with other activities. The estimates cover Great Britain, not Northern Ireland. [1]
ONS's methodology, revised on 24 August 2026, explains another difficulty: data centres lack a unique industrial classification. Their activity can sit within different business categories. A broad industry total should therefore not be presented as a complete count or meter reading for the facilities visible on a local map. [2]
Give every figure four labels
Our practical approach is to ask for the place, period, measure and boundary. Does the figure describe one building or a whole campus? Is it measured over a stated year, or announced for a future phase? Does it describe electricity consumed, or capacity? Does the boundary include the whole facility or only its computing equipment?
The Commons Library explains that capacity is usually expressed as the maximum power a data centre can draw. [3] A megawatt capacity announcement alone cannot supply the missing time and utilisation needed to calculate actual energy use.
A deliberately hypothetical example shows the distinction: a constant measured demand of 1 MW sustained for 24 hours would use 24 MWh, or 24,000 kWh. That is arithmetic, not an estimate for any named facility. Substituting an advertised maximum into the calculation silently assumes that demand stays at that maximum all day.
The same care applies to household comparisons. Divide a documented annual electricity total by a stated annual household electricity benchmark, with matching units. The result compares quantities of electricity; it does not identify homes deprived of supply. A scenario based on assumed loading should display those assumptions beside the result.
What would answer the question during a power cut?
A useful facility record would show its documented development stage, the source and date of any capacity claim, and separately any disclosed consumption period. Current grid supply and backup operation would need their own dated evidence. An operating business, a planned expansion and a verified live electricity reading are different observations.
Use a facility map to investigate infrastructure around you. To assess a particular interruption, look for evidence tied to that incident and address. Putting two markers close together is a starting point for a question, not an answer about electrical connections, priority or cause.