Summary
- European hyperscale projects scheduled for 2026–28 sit an average 175km from major hubs, up from 46km for projects delivered in 2022–25.
- Greenfield developments account for 39% of the future pipeline as power availability increasingly overrides proximity to traditional demand centres.
- Powered land costs remain sharply higher in established markets, adding another incentive for developers to look towards secondary and tertiary locations.
Europe’s next generation of large data centres is moving further away from its traditional hubs as developers follow available power, cheaper land, and faster grid connections into new regions.
Research from JLL shows hyperscale projects scheduled for delivery between 2026 and 2028 are located an average of 175km from major data centre hubs. Projects completed between 2022 and 2025 averaged 46km from those markets.
The change is being driven heavily by AI training infrastructure, where campuses can require hundreds of megawatts and do not always need to sit close to large concentrations of end users. Greenfield locations now account for 39% of Europe’s future pipeline, compared with 8% of projects delivered during the previous period.
By contrast, the share of projects in inner-city locations is expected to fall to 5% from 13%.
The figures put measurable distance behind a shift that has been visible across the European market for several years. Frankfurt, London, Amsterdam, Paris, and Dublin remain the continent’s largest established data centre markets, but grid congestion, planning constraints, land availability, and rising site costs are making it harder to deliver large new campuses within or immediately around those hubs.
JLL estimates the FLAP-D markets have about 3.8GW of live capacity, with another 1.4GW under construction and 2GW planned. Vacancy across those markets stood at 6.4% in the second quarter.
At the same time, more than half of Europe’s expected AI-related growth is forecast to land in Nordic and Tier 2 markets, where developers have a better chance of assembling the combination of land and electrical capacity required for hyperscale projects.
That migration is also visible in the proposed gigawatt-scale end of the market. Data compiled by DC Byte and reported by Reuters identified nine proposed European projects of at least 1GW, with only one located near a major city — Paris. The remainder stretch from rural Spain to northern Sweden.
JLL says prime powered land across FLAP-D now averages about €2.26m per MW, 82% above its 2021 level. Primary markets carry a 2.3-times premium over secondary locations and roughly four times the cost of tertiary markets, adding a substantial land-price incentive to the search for power beyond established hubs.
The land itself is only one component of development cost, but the spread becomes material when a campus requires hundreds of megawatts and substantial space for substations, backup generation, cooling equipment, and future phases.
The geographic shift does not mean established hubs are being displaced. Their dense fibre networks, cloud on-ramps, customer ecosystems, and existing technical workforce continue to make them valuable for latency-sensitive, enterprise, and interconnection-heavy workloads.
Instead, the European market is becoming more functionally divided. Large AI training campuses can move towards regions where generation and grid capacity are more accessible, while connectivity-heavy and lower-latency deployments remain closer to major population and business centres.
That division also places more weight on fibre routes and transmission infrastructure between the new campuses and existing hubs. Moving compute towards available electricity only works if the network can move data reliably between those locations and the customers, cloud regions, and exchanges they serve.
DataCentral has previously examined how power availability is rewriting Europe’s data centre map. The latest JLL figures suggest that process is now becoming visible not only in market commentary but in the physical distance between planned capacity and established data centre clusters.
The change is likely to accelerate as AI infrastructure grows. JLL estimates the four largest hyperscale cloud providers will spend about $725bn in 2026, up 77% from $410bn in 2025, with AI and data centre infrastructure absorbing much of that expenditure.
Europe has demand for that capacity. The constraint is increasingly whether developers can find locations where land, planning consent, grid infrastructure, and delivery times align closely enough to turn demand into an operational facility.
