AI data centers seek space outside European hotspots

AI data centers seek space outside European hotspots

The rapid growth of AI is reshaping the European data center landscape. New hyperscale projects are increasingly being built far outside traditional data center regions. According to figures from JLL, published by Reuters, power availability is a more important factor than proximity to major cities.

That difference is significant. Data centers completed between 2022 and 2025 are located, on average, 46 kilometers from a major urban hub. For projects scheduled for completion between 2026 and 2028, that distance increases to 175 kilometers.

The well-known FLAP-D markets, Frankfurt, London, Amsterdam, Paris, and Dublin, remain important for business workloads. However, they are becoming less attractive for very large AI facilities. Available land is scarce, grid connections take a long time to establish, and zoning restrictions make expansion more difficult.

Power determines location

AI training, in particular, places different demands on data centers. Such environments combine large amounts of computing power with very high electricity consumption. Because these workloads do not always need to run close to end users, operators can move to areas where sufficient energy and land are available.

According to Assad Noori, head of data centers for EMEA at JLL, location choice is therefore becoming less and less about where demand is. The availability of sufficient power is becoming a decisive factor.

This trend is also evident in the development pipeline. Greenfield sites account for 39 percent of planned European projects. Among data centers already completed, that share is only 8 percent. At the same time, the share of projects in inner-city areas is expected to drop from 13 to 5 percent.

Even the largest projects are primarily located outside the well-known data center hubs. DC Byte is tracking nine European projects with a planned capacity of more than one gigawatt. Only one of these is located near a major city, Paris. Other locations range from rural Spain to northern Sweden.

Amsterdam is the most expensive

In addition to availability, location price plays an important role. JLL calculates the cost of land where the necessary power capacity is also available per megawatt of IT load. In the major European data center markets, that price averages 2.36 million euros per MW.

In secondary markets such as Copenhagen, Warsaw, and Milan, that figure drops to an average of 978,000 euros. In smaller markets, including Bordeaux, the average is 512,000 euros. In some locations, suitable land is available starting at around 200,000 euros per MW.

Amsterdam tops the European rankings at approximately 2.7 million euros per MW. London follows with 2.6 million euros, and Frankfurt with 2.5 million euros.

AI further drives up investments

The search for new locations coincides with a sharp rise in investments in AI infrastructure. JLL expects the four largest hyperscale cloud providers to collectively spend $725 billion this year. In 2025, that figure was $410 billion. A large portion of the money is going toward AI computing power and the data centers required to support it.

JLL also expects that by 2030, AI workloads could account for about half of global data center capacity. This increases the pressure to make new areas suitable for large-scale data center development.

This could bring investment and jobs to regions that have played virtually no role in the data center sector until now. At the same time, new locations may face debates regarding land use, nature, electricity, and water. The shift therefore does not resolve the physical constraints around traditional data center hubs, but instead relocates some of them to new regions.