Europe’s AI Data Centers Are Moving Toward Power, Not Cities
Frankfurt, London, Amsterdam, Paris and Dublin remain vital for enterprise demand, but the next hyperscale sites are being selected around power availability, land and grid access.
Listen to this story
The audio brief
Story brief
3 key pointsEurope’s planned hyperscale AI capacity is becoming a power-and-land allocation story rather than a city-proximity story. JLL finds projects expected online in 2026–2028 average 175 kilometers from major hubs, versus 46 kilometers for 2022–2025 deliveries. Greenfield sites now make up 39% of the future pipeline, while powered land costs fall from €2.36 million per megawatt in core markets to €512,000 in tertiary...
- 01
JLL’s 2026–2028 hyperscale pipeline averages 175 kilometers from major hubs, nearly four times farther than recent deliveries.
- 02
Greenfield developments comprise 39% of the future pipeline, versus 8% of projects delivered between 2022 and 2025.
- 03
Powered-land costs average €2.36 million per megawatt in core markets, compared with €978,000 in secondary cities.
Europe’s largest data-center markets remain critical for enterprise demand. But the evidence in the next hyperscale AI pipeline points elsewhere: developers are moving farther from major cities as they seek sites with enough power, land and faster grid connections.
JLL’s figures put scale on that shift. Hyperscale data centers due online between 2026 and 2028 average 175 kilometers from a major hub, versus 46 kilometers for projects delivered from 2022 through 2025. The coming cohort is therefore almost four times as distant on average.
The pipeline is moving beyond urban infill
The change is visible in the type of projects being planned. Greenfield developments account for 39% of Europe’s future data-center pipeline, compared with 8% of delivered projects. Inner-city sites are expected to represent 5% of the pipeline, down from 13%; most of the rest are planned for industrial or edge-of-city locations.
AI campuses turn the grid into the first constraint
JLL attributes the shift largely to AI training campuses, which need large amounts of electricity and cooling water. Established European markets face land shortages, planning restrictions and lengthy grid-connection waits. That changes site selection: the decisive question is increasingly where developers can secure sufficient power, rather than where demand already exists.
JLL reported an average powered-land cost of €2.36 million per megawatt of IT load in Europe’s core markets.
JLL reported average powered-land costs of €978,000 per megawatt in secondary cities.
JLL reported average powered-land costs of €512,000 per megawatt in tertiary areas.
The core markets carry their own price differences. JLL puts Amsterdam at roughly €2.7 million per megawatt, followed by London at €2.6 million and Frankfurt at €2.5 million. Cheaper powered land can make locations outside those hubs more attractive when a project needs a large electricity allocation.
The city hubs retain a different role
This is not a retreat from Frankfurt, London, Amsterdam, Paris or Dublin. JLL says those markets remain critical because enterprise demand continues there, even as hyperscale AI infrastructure requires a different scale of power and land. DC Byte’s early-stage data supports the split: among nine proposed gigawatt-plus European data centers, only one is planned near a major city, Paris; the others run from rural Spain to northern Sweden.
The new geography is still a pipeline, not completed capacity. But the direction is clear: Europe’s established hubs continue to serve enterprise customers while planned large AI sites increasingly follow available power and buildable land.
Sources
- calcalistech.comEurope’s AI data centers are moving far from the cities | CTech
