Data center growth in Europe expands to emerging markets

Data centers are playing a growing role in Europe’s electricity demand and currently account for an estimated 2% to 3% of total European electricity consumption, or roughly 60 to 100TWh per year. [1] Driven by cloud adoption, artificial intelligence (AI), high‑performance computing, and broader digitalization, demand is set to rise substantially over the coming decade. By 2030, electricity consumption from data centers may reach around 150 to 170TWh, close to 5% of total projected European electricity demand, making them one of the fastest‑growing demand segments in Europe’s electrification landscape. [2]

Data center development remains uneven across Europe (see figure 1). While initial growth was concentrated in Frankfurt, London, Amsterdam, Paris, and Dublin (FLAP-D) due to their high digital connectivity, these markets are becoming saturated. Increasingly, new capacity is built in Nordic markets,[3] Spain, Belgium, Italy, Poland, and Portugal. In this report, we analyze the key drivers that determine the location of new data centers, how they interact to influence location decisions, and how they are reshaping the geography of data center growth within Europe’s electricity system.

[1] Source: BloombergNEF Global IT Capacity datacenters 2025

[2] Source: JLL 2026 Global Data Center Outlook

[3] Nordic countries include Denmark, Finland, Norway, and Sweden.

While multiple drivers influence data center development across Europe (see table 1), power system conditions play an increasingly relevant role. Developers search for locations with adequate digital infrastructure and regulatory conditions, as well as access to affordable and, preferably, clean power. In its 2026 outlook, the European Data Center Association (EUDCA) reports that access to power is consistently cited among the main challenges. Understanding how the different drivers vary across Europe helps to identify the most attractive markets for new data center capacity and thus where additional electricity can grow fastest.

The importance of each driver depends on the type of data center and in particular on the workload characteristics of an individual project. Three main types of workloads are cloud computing, AI inference, and AI training, and these workloads have different priorities. Cloud computing and AI inference prioritize proximity and connectivity, while AI training prioritizes scalable, reliable power and cost certainty. Therefore, different European countries are likely to attract different mixes and types of data center workloads.

Access to the grid

Data centers require a substantial grid connection, ranging from 1MW to 100+MW, depending on the type of data center and workload. As grid connections become increasingly scarce across Europe,[4] grid access is becoming a leading driver for data center growth. Limited grid connection availability in a market poses risks of delays or cancelations.

Indicators for assessing grid connection availability include local grid connection queues, connection allocation policies, and planned grid and network reinforcement projects. Transmission system operators across Europe have reported expanding backlogs for large new connections, as electrification makes end users compete for limited network capacity. Moreover, some countries are moving away from the first-come-first-serve principle for grid connections and prioritizing projects that serve public interests, which might result in a disadvantage for data centers. Some have even implemented moratoriums (Ireland[5] and the Netherlands) on new data center capacity.

Although a large data center can typically be constructed within roughly two years, actual delivery timelines now average close to four, largely influenced by the availability and timing of grid connection capacity. According to Ember, connecting a new data center in traditionally congested markets (FLAP-D) can take up to 10 years or more when significant upgrades are needed, while in other countries with more proactive planning (Belgium, Portugal, Spain, Nordic countries, and Italy) the process can be significantly shorter.[6] Where reinforcement of substations or transmission infrastructure is required, connection lead times can extend significantly.

Connection queues are emerging where local grid capacity has not kept pace with rising demand, reflecting broader electrification trends alongside rapid digital expansion. These extended timelines increasingly shape developers’ location decisions. New projects can be realized more rapidly in areas where grids can deliver connection capacity within a predictable timeframe. In areas where queues are long and reinforcement uncertain, capacity may be delayed, downsized, or reconsidered. Analysis by EUDCA underscores that time-to-connection has become a central driver, in some cases outweighing traditional factors such as land cost or regulatory incentives.

Countries that undertake anticipatory grid planning, advance reinforcement projects, and clarify connection pathways improve their competitive position for attracting data center and AI-related investments.

Access to clean, cost-predictable, and contractable power

Electricity is a major operating expense for data centers. Predictable and low power prices are therefore crucial, particularly for power-intensive workloads such as AI training. As electricity prices are subject to both short-term volatility and long-term uncertainty, data centers typically hedge price risks by signing a power purchase agreement (PPA) for multiyear periods.

Data centers are one of the largest offtakers of PPAs in Europe. Contracts are typically signed with solar or wind project developers, who need the PPA to hedge their own risks. Through such agreements, renewable energy projects facilitate the construction of data centers by ensuring their power supply, and data centers enable the materialization of new renewable generation capacity by ensuring revenues. Data centers in Europe have committed to becoming climate‑neutral by 2030, including operating on 100% carbon‑free energy and meeting strict efficiency targets under the Climate Neutral Data Center Pact. Beyond sector‑wide commitments, individual hyperscale firms have announced their own net‑zero pathways, often aiming for complete decarbonization well before 2050.

Multiple factors influence PPA prices and volumes, including solar and wind generation capacity and local wholesale electricity prices. Figure 2 shows that PPAs signed with offtakers in the information, communication, and technology (ICT) sector are most common in countries with high shares of solar and wind energy in the generation mix, such as Spain, Nordic countries, Germany, the Netherlands, the UK, and Ireland. The figure also reveals a correlation between average wholesale power prices and the share of hyperscale data centers. Because the power demand of data centers is so high, electricity prices play a decisive role in where they are located.

However, PPAs do not have to be limited to one country. Through nonphysical PPAs, data centers can, on paper, procure energy from a country with excess cheap renewables and consume it in a country with higher electricity prices. Typically, the negotiated price on such a PPA sits between the wholesale electricity prices of the two markets.

[4] Source: JLL 2026 Global Data Center Outlook

[5] Although no longer a moratorium as of December 2025, Ireland’s policy has not lowered the bar for new data centers. Instead, strict, measurable conditions have replaced the former de facto freeze, including requirements for on‑site or proximate generation equal to 100% of demand and an 80% renewable‑matching obligation.

[6] Source: JLL 2026 Global Data Center Outlook

Access to digital infrastructure and connectivity

Digital connectivity refers to access to high-capacity fiber networks, major internet exchange points (IXPs), international subsea cable routes, and integration into dense digital ecosystems. It enables data centers to operate within distributed, cross-border architectures by reducing latency, increasing redundancy, and facilitating efficient interconnection for cloud services, AI inference, and content delivery.

Europe’s established digital hubs are anchored by some of the world’s largest IXPs, including DE-CIX, AMS-IX, and LINX, which support internet traffic volumes and provide low-latency connectivity across western Europe. Dense metropolitan fiber networks in these regions reinforce their role as interconnection centers for hyperscale and colocation facilities.

Access to predictable permitting and connection frameworks

Regulatory frameworks shape how quickly and predictably data center projects can move from planning to operation. In the current context, this primarily concerns how grid connection applications are assessed and prioritized, how reinforcement costs are allocated, and how permitting processes are coordinated across national and regional authorities in Europe. These elements directly influence time-to-power and overall project risk.

In congested hubs, queue access and connection sequencing are central. In large continental systems, coordination across administrative layers shapes execution. In diversification markets, regulatory frameworks are adapting to rising volumes of large-load applications. Differences in cost-allocation approaches and in the balance between transmission- and distribution-level constraints further influence project economics and connection timelines.

Besides regulatory environments, market track records are also an indicator of how likely regulatory hurdles are in particular regions. Markets with experience in hosting large infrastructure projects often benefit from accumulated expertise among grid operators, utilities, contractors, and permitting authorities. In markets with less historical concentration of hyperscale facilities, institutional capacity may still be adapting to the scale and complexity of large data center loads.

Taken together, these drivers show clear patterns in how data center development concentrates and shifts across Europe. These clusters reflect how power system conditions, grid readiness, spatial limits, regulation, and workload needs interact.

Differences in workload characteristics play an important role in shaping these patterns. Latency‑sensitive workloads such as cloud computing and AI inference tend to favor locations with strong connectivity and proximity to end users, while power‑intensive AI training workloads are more sensitive to power availability, grid access, and access to clean and reliable electricity.

Core hubs under constraint: The Netherlands and Ireland

The Netherlands and Ireland remain central to Europe’s data center landscape and form part of the traditional FLAP-D core. These markets benefit from strong digital connectivity, proximity to major demand centers, and mature data center ecosystems. Dense fiber networks, major IXPs, and established supply chains support a wide range of workloads, particularly cloud computing and AI inference.

However, structural constraints increasingly bind both countries. Grid congestion and extended connection timelines have become decisive barriers to new large-scale developments. In Ireland, connection limits and grid stability concerns have directly affected the pace of new approvals, while in the Netherlands, regional congestion and land scarcity have slowed expansion. As a result, growth is shifting from unconstrained greenfield development toward optimization, densification, and delivery of select projects. While demand remains strong, infrastructure constraints increasingly determine the type, scale, and timing of new capacity.

Large continental systems: Germany, France, and the UK

Germany, France, and the United Kingdom combine large domestic demand with electricity systems capable of absorbing substantial additional load at the national level. Their scale reduces the likelihood that data centers will become an immediate system-wide concern, although local grid conditions remain critical. Proximity to major industrial and enterprise users continues to support sustained demand.

France benefits from access to stable and relatively low-carbon electricity, supporting workloads with high and continuous power demand, including certain AI-related applications. Germany faces more complex grid congestion patterns and regional permitting dynamics, which shape where capacity can realistically be delivered. The United Kingdom combines strong digital connectivity with regional variation in grid headroom, leading to increasing development outside traditional London-centric clusters. In all three markets, feasibility depends heavily on regional infrastructure.

Emerging alternatives with scalable power: Spain and Portugal

Spain and Portugal are increasingly positioned as alternative growth markets for data center development. These countries benefit from comparatively greater land availability, expanding power system capacity, and rapid renewable energy deployment. The growing maturity of corporate PPA markets improves access to low-carbon electricity at scale, while new subsea cable landings strengthen global connectivity.

This makes Iberian markets attractive for AI training and other power-intensive workloads. However, local transmission constraints and permitting timelines still affect deliverability. As reinforcement and grid expansion progress, Spain and Portugal are well positioned to capture a larger share of incremental capacity.

Clean power and system efficiency: Denmark, Finland, Norway, and Sweden

Denmark, Finland, Norway, and Sweden stand out for their access to low-carbon electricity and, in several cases, abundant dispatchable power. Combined with cooler climates and highly efficient systems, these characteristics support lower operating costs and alignment with corporate sustainability objectives.

These advantages are particularly relevant for large-scale and energy-intensive workloads, including AI training. However, distance from major demand centers and thinner ecosystems limit Nordic countries’ ability to replace established digital hubs. As a result, the Nordic markets are more likely to attract specific workload types rather than a full spectrum of data center activity.

Diversification markets: Italy, Belgium, and Poland

Italy, Belgium, and Poland are increasingly positioned as diversification markets within Europe’s evolving data center geography. While not traditional hyperscale hubs, they are attracting spillover demand from more constrained regions, supported by selected areas of grid availability, expanding deployment of renewable energy, and strong integration with European electricity and digital networks.

Belgium benefits from its central location and cross-border interconnection, making it a logical recipient of demand shifting from neighboring congested markets. Italy combines growing domestic digital demand with regional variation in grid headroom, particularly in its northern regions. Poland adds scale and a rapidly expanding digital economy, supported by ongoing infrastructure investment and integration into the European power system. These markets complement established hubs by absorbing demand where grid access and permitting remain manageable.

Structural constraints in electricity systems and supporting infrastructure will increasingly shape data center growth in Europe. As facilities scale in size and power intensity, access to grid connection capacity, system headroom, and predictable reinforcement timelines will largely determine where new capacity can be delivered.

The rise of AI is accelerating these dynamics, as higher utilization rates and power density reduce operational flexibility and increase the system impact of each new megawatt. This will place greater emphasis on locations where reliable electricity can be secured at scale, whether through inherent system characteristics or mature contracting frameworks offering long‑term cost visibility. Data centers are also becoming major buyers of renewable PPAs in Europe, driven by sustainability goals and tightening power supply.

At the same time, growth will not shift uniformly away from established hubs. Latency‑sensitive workloads such as cloud computing and AI inference will remain anchored in core digital markets, while more power‑intensive, latency‑flexible workloads like AI training may move toward regions with fewer grid constraints. The interaction of these structural constraints with workload characteristics is increasingly determining which markets can support power‑intensive growth and which remain limited to latency‑driven activity.

Taken together, these dynamics point to a more regionally differentiated pattern of data center growth, shaped by grid reinforcement timelines, permitting outcomes, and national policy. Rather than a single new hub emerging, Europe is likely to develop a more distributed network of markets increasingly embedded within national electricity systems.