Summary
- DC Blueprint analysed hourly electricity generation across 15 European markets and ranked 12 whose data passed its checks.
- October carbon intensity ranged from 25g CO₂e/kWh in Norway's NO2 zone to 595g in Poland.
- Production-based calculations, electricity imports, and differences in emissions factors limit comparisons with other reporting methodologies.
Electricity generated across European power markets carried substantially different estimated carbon intensities in October 2026, with analysis placing southwestern Norway, central Sweden, and France well below Germany, northern Italy, and Poland.
The latest European Grid Index from DC Blueprint calculates hourly carbon intensity using electricity generation data and assigned lifecycle emissions factors, producing comparable estimates for a continuous electrical load in 12 markets.
Norway’s NO2 electricity zone recorded the lowest monthly average at 25g of carbon dioxide equivalent per kilowatt-hour, followed by Sweden’s SE3 zone at 38g and France at 40g.
Finland averaged 56g, while Spain and Great Britain recorded 158g and 159g respectively. Germany and northern Italy were substantially higher at 356g and 355g, with Poland recording 595g.
The difference between Norway and Poland was approximately 24-fold under the study’s calculation method, reflecting the contrasting generation mixes supplying those markets.
DC Blueprint examined hourly data for 15 markets but excluded the Netherlands, Switzerland, and western Denmark from its final ranking because of identified data-quality concerns.
The figures describe calculated emissions associated with electricity production, not emissions measured at individual data centres. The study also excludes imported electricity from its main production-based calculations.
Those methodological boundaries are important when interpreting the differences between countries. A facility consuming electricity in one market may receive power through an interconnected network, while the study calculates intensity from generation recorded within the specified market or zone.
The underlying electricity mix explains much of the variation. Norway’s NO2 zone relied heavily on hydropower during October, while Sweden’s SE3 zone and France benefited from substantial nuclear generation.
Poland remained more dependent on fossil fuels. Hard coal and lignite together accounted for approximately 57% of its electricity production in the dataset.
Germany recorded a comparatively high carbon intensity despite obtaining a substantial proportion of electricity from renewables because coal and gas continued to contribute to its generation mix.
The comparison also reveals how electricity conditions change between months. Austria’s calculated average rose from 77g in September to 170g in October as gas generation increased and hydro and solar output declined.
Belgium moved in the opposite direction, falling from 250g to 220g. The study records renewed nuclear generation during October alongside higher wind output.
These changes describe what the generation data recorded rather than establishing a single cause for every movement. Weather, plant availability, system operation, and electricity demand can all influence monthly results.
For Great Britain, the study’s October estimate was 159g CO₂e/kWh, compared with 115g reported through the national system operator’s own methodology.
The difference arises from calculation methods, including the choice of emissions factors and the treatment of electricity imports. Neither value should be substituted for the other without explaining the basis on which it was calculated.
Annual trends also require caution. A change between two individual months can reflect seasonal electricity conditions rather than a permanent reduction or increase in the carbon intensity of a national electricity system.
The index extends beyond monthly averages by examining whether electricity consumption could be concentrated into hours when the generation mix is less carbon intensive.
Under a hypothetical scenario in which a flexible load uses the same monthly electricity but operates only during the cleanest quarter of hours, Ireland’s calculated intensity falls from 234g to 98g.
Germany falls from 356g to 156g, while Belgium falls from 220g to 98g. These are modelled outcomes based on knowledge of which hours were cleanest after the month had finished.
The scenario assumes that a workload can be concentrated into one-quarter of the available hours, requiring four times its normal average electrical capacity while operating.
That assumption limits its application to real data centres. Time-sensitive inference, transaction processing, and customer services cannot generally be stopped for prolonged periods simply to follow changes in grid emissions.
Some batch computing, rendering, and other deferrable workloads have greater scheduling flexibility, although service commitments, available computing capacity, and forecasting accuracy still constrain how far consumption can move.
Operating only in selected hours would also require sufficient electrical and cooling capacity to accommodate the resulting concentration of demand.
The difference between geographical locations remained substantial even under the model’s idealised scheduling assumptions. Germany’s flexible-load result of 156g was still higher than the continuous-load averages for Norway, Sweden, France, and Finland.
Electricity procurement introduces a further distinction. Contractual renewable electricity arrangements, including power purchase agreements and certificates, are assessed through different accounting approaches from the study’s location-based generation estimates.
Consequently, the monthly index provides a comparison of electricity production characteristics rather than a complete assessment of any particular facility’s reported environmental performance.
Site-specific emissions calculations must also consider measured electricity consumption, the relevant accounting framework, and the operating conditions of the data centre over the reporting period.
The October findings establish the extent of variation between European electricity systems while demonstrating how methodology and operating flexibility influence the figures used to assess computing infrastructure.
