Summary
- Dutch data centres consumed 5,664GWh in 2025, equivalent to 5% of national electricity consumption, compared with 4.6% in 2024.
- Approximately 50 facilities consuming at least 10GWh annually accounted for more than 90% of data centre electricity use.
- The provisional figures exclude an estimated 142GWh consumed by internally operated data centres at organisations including hospitals and universities.
Data centres consumed 5,664 gigawatt-hours (GWh) of electricity in the Netherlands during 2025, accounting for 5% of national electricity consumption, according to new figures from Statistics Netherlands (CBS).
The sector’s share increased from 4.6% in 2024 and has more than tripled from the 1.5% recorded in 2017. Electricity supplied to Dutch data centres rose by approximately 12% during 2025, with the increase concentrated among larger installations rather than smaller facilities.
Approximately 50 data centres consuming at least 10GWh annually accounted for more than 90% of the sector’s electricity consumption. CBS estimates that the country has around 200 data centres, meaning roughly one quarter of the facilities covered by the statistics account for the overwhelming majority of measured demand.
The figures, published on 18 November, are provisional and measure electricity supplied to existing installations. They do not represent proposed development capacity, reserved grid connections or forecasts of demand from facilities that have yet to enter service.
The Netherlands has one of the highest data centre electricity consumption shares in Europe. In the latest comparable EU figures, covering 2024, only Ireland recorded a higher share of national electricity demand attributable to data centres.
Large facilities drive the increase
CBS divides data centres into two groups according to annual consumption. Facilities using at least 10GWh accounted for approximately 5,289GWh in 2025, compared with 4,662GWh in 2024.
Smaller facilities moved in the opposite direction. Their consumption declined from approximately 398GWh to 374GWh over the same period, indicating that growth was not distributed evenly across the sector.
The larger group therefore accounted for essentially all of the net increase. Its electricity consumption grew even though the number of facilities in that category has remained relatively stable over recent years.
A stable facility count does not imply stable operating demand. An existing data centre can increase electricity consumption by installing additional computing equipment, expanding customer occupancy or operating hardware at higher utilisation without constructing another building.
Changes in equipment configuration can also alter demand. Higher-powered processors and accelerator systems may increase electricity use within an existing electrical connection, although greater efficiency can reduce the energy required for a particular amount of computing work.
The CBS statistics do not identify which individual technologies or workloads caused the increase. They therefore cannot establish how much additional consumption resulted specifically from AI processing or from changes in other computing services.
Annual electricity consumption also differs from maximum electrical capacity. A facility with a large grid connection may operate below its maximum load because computing equipment is installed gradually or customers have not occupied all available space.
Conversely, an established data centre may increase annual electricity use without increasing its contracted maximum capacity if its average demand rises while remaining within the existing connection limit.
These distinctions affect the way the figures should be compared with development pipelines. A planned 100MW data centre does not consume 100MW continuously simply because that figure appears in its application documents or investment plans.
Electricity consumption is not an efficiency measure
The national statistics describe the amount of electricity delivered to the sector, not the energy efficiency of individual facilities. A rise in consumption may reflect greater computing activity, but it can also result from changes in equipment, cooling loads or operating practices.
Power usage effectiveness, commonly expressed as PUE, compares total facility electricity consumption with the amount used by IT equipment. Calculating it requires suitable measurements of both quantities within consistent system boundaries.
The CBS series does not provide the facility-level information needed to calculate PUE for each data centre. It cannot therefore establish whether individual sites have become more or less efficient during the period.
Energy consumption also differs from carbon emissions. The emissions associated with electricity depend on the generation supplying the network, the time at which electricity is consumed and the accounting method used to assess purchased energy.
Location introduces another variable. Large electricity users can create substantial demand for transmission and distribution infrastructure, but the consequences depend on local network capacity rather than national consumption alone.
Data centres concentrated around particular industrial or metropolitan areas may require additional substations, cables and network reinforcement. Those requirements can influence connection times, investment costs and the availability of capacity for other electricity users.
The published data provide a measured indication of national demand, but do not show the hourly consumption profiles or grid connection conditions of individual facilities. Such information would be needed to assess their contribution to specific network constraints.
CBS has also excluded a separate category of computing infrastructure operated internally by organisations whose primary business is not data centre services. Hospitals, universities and other organisations with substantial computing facilities fall within that distinction.
Estimated electricity consumption for those internal installations was 142GWh in 2025. Adding the estimate to the published commercial data centre figure would increase the total amount associated with computing facilities, but it would also change the statistical scope and require a corresponding adjustment to any comparison.
International figures require similar care. The latest comparable data cited by CBS cover 2024, when Ireland recorded data centre consumption equivalent to 21.3% of national electricity demand, compared with 4.6% in the Netherlands and 4.2% in Germany.
Those percentages measure the sector’s share of each country’s electricity consumption, not absolute demand. Germany used considerably more electricity in data centres than the Netherlands in 2024, despite recording a smaller percentage of national consumption.
The provisional 2025 Dutch figures establish that existing data centres are drawing a growing share of the electricity supplied within the country. Further revisions and subsequent annual measurements will show whether demand continues increasing at a similar rate as existing sites expand and new facilities become operational.

