The European data centre market is currently in a state of profound transformation. Record demand for digital infrastructure meets extensive investments in Artificial Intelligence (AI) and, simultaneously, limited power capacities in traditional data centre hubs. This constellation is causing a significant shift in the criteria for selecting sites for new facilities, according to JLL's latest EMEA Data Center Report.
The analysis shows that hyperscaler projects slated for completion between 2026 and 2028 are located, on average, 175 kilometres from major data centre hubs. This contrasts with projects realised between 2022 and 2025, where the average distance was only 46 kilometres. Martina Williams, Head of Work Dynamics EMEA at JLL, highlights that the planning logic has fundamentally changed. While operators historically sought proximity to population centres, ensuring sufficient power capacities has become the decisive factor for the growing AI training infrastructure, outweighing mere proximity to demand.
The composition of the European supply is also shifting from industrially zoned areas to so-called greenfield sites. Between 2022 and 2025, almost 80 percent of newly created capacities were realised on existing industrial sites or at urban fringes. For the 2026 to 2028 pipeline, greenfield sites already account for 39 percent, representing an almost fivefold increase. The four largest hyperscale cloud providers worldwide plan investments of an anticipated US$725 billion for 2026, an increase of 77 percent compared to US$410 billion in 2025. A large portion of these funds is being directed towards AI computing and the associated data centre infrastructure. The German market is strategically gaining importance; for instance, AWS opened its European Sovereign Cloud in Brandenburg in January 2026 with an investment of EUR 7.8 billion.
Within Continental Europe, Frankfurt remains the largest data centre market with an installed capacity of 970 megawatts (MW). Paris follows with 614 MW, Amsterdam with 611 MW, and Milan with 230 MW. Across Europe, London leads with 1,277 MW. However, Frankfurt boasts the largest European pipeline, with 311 MW under construction and a further 687 MW planned. Nevertheless, realisation times are significantly extended by the necessity of grid connections and lengthy planning decisions; grid operators project capacities for current applications only from mid-2030s onwards. Utilisation in Frankfurt has intensified further with a vacancy rate of merely 3.1 percent, which is less than half the average of the five largest data centre markets (FLAP-D). In the first half of 2026, 36 MW of new capacity was commissioned, following a record year in 2025 with 128 MW.
Berlin has established itself as an attractive alternative location. Capacity has grown to 148 MW, and with a vacancy rate of only 0.9 percent, the market is considered almost fully utilised. The pipeline in Berlin has expanded significantly: 143 MW are under construction, and a further 379 MW are planned, including large grid-connected campus projects such as the Virtus site in Wustermark. Frank Weber, Senior Director Client Solutions Datacenter & New Energies at JLL Germany, sees this as an opportunity for decentralised locations in Germany. He points out that northern and eastern Germany produce massive electricity surpluses that often remain unused. This situation offers the data centre industry an immensely advantageous starting position.
Weber emphasises that the frequently cited lack of connectivity is becoming less significant, as latency differences between Frankfurt and, for example, Schleswig-Holstein are in the single-digit millisecond range and are irrelevant for most applications. Modern cloud architectures are designed for geographical distribution. A data centre in northern Germany, partly powered directly by wind energy, has a significantly improved environmental footprint. Ultimately, access to the power grid and local acceptance will determine the realisation of projects. Developers who involve communities early in planning processes will make faster progress. Access to energy will therefore become a decisive competitive advantage for developers, investors, and regions seeking to attract AI infrastructure.
