The great electrification: Can the EU power its AI ambitions?

The great electrification is reshaping Europe’s energy landscape, with electricity becoming the backbone for a clean, affordable, and secure energy system. At the same time, growing digitalization and the rapid rise of artificial intelligence (AI) are driving a significant expansion of data center demand in the EU.

In this article, we show how these developments intersect. The rise of AI boosts demand for data centers, while also increasing the electricity demand per data center due to their high workload. This results in an increase in electricity demand on top of the electrification trend of existing energy demand. As pressure on electricity grids and renewable generation grows, the pace of the great electrification will also partly determine the growth of the EU data center sector.

Data centers and AI computing capacity are scaling up across Europe as the demand for cloud services, low‑latency processing, and AI training workloads grows. On top of the growing usage of AI, the European Commission’s (EC) push for digital sovereignty demands an accelerated growth of data center capacity. As the European Union’s (EU) energy demand has already embarked on a path of electrification to enhance energy security and independence, the additional AI-driven electricity demand adds extra pressure to the EU’s electricity system, especially in the years to come.[1]

The European Commission’s push for AI infrastructure and digital sovereignty

The EU aims to strengthen its strategic digital position through a combination of policy initiatives and investment programs, aimed at expanding domestic AI computing capacity and accelerating the use of AI across the economy. Policymakers have raised concerns about Europe’s reliance on non-European cloud providers and digital infrastructure. Draghi's competitiveness report identified this dependence on foreign digital technologies as a structural weakness in advanced sectors such as AI and cloud computing. As the report stresses, “the EU is falling behind in providing the state-of-the-art infrastructure necessary to enable the digitalization of the economy.”

To achieve digital sovereignty, the EC is pursuing a dual strategy: First, the AI Act (adopted in 2024 and entering phased enforcement through 2026) establishes a regulatory framework for the safe and trustworthy development and deployment of AI systems. Second, a set of investment and industrial policy initiatives aims to expand Europe’s domestic AI computing capacity.

The AI Continent Action Plan (launched in 2025) forms a key part of this approach, supporting the development of AI factories. These facilities, to be co‑funded by the EU and member states under a EUR 2.6bn joint investment program, are intended to expand the EU’s high‑performance computing backbone and reduce reliance on non‑European cloud providers. The InvestAI program (2025) complements this strategy by aiming to mobilize up to EUR 200bn in private investment, including dedicated funding streams to support the build‑out of these compute centers and accelerate AI adoption across the economy.

The overall strategy to achieve digital sovereignty is supported by the EC’s aim to triple EU data center capacity by 2030-2032. Beyond this milestone, however, a longer-term strategy has yet to be defined. Achieving this target would also require expanding data center capacity from the equivalent of 8GW in 2025 to around 22GW in 2032 – with the additional capacity alone matching Hungary’s current installed electricity production capacity. Reaching this scale will require a rapid scale‑up in large, energy‑intensive hyperscale facilities capable of supporting AI workloads, in addition to the new EU‑level AI factories (see figure 1). As a result, data centers are becoming more directly embedded in energy planning. Reflecting this shift, the EC is preparing a 2026 data center energy efficiency package, including an EU‑wide label with requirements on energy use, water use, and renewable sourcing, alongside new reporting obligations for large facilities.

[1] While traditional workloads accounted for roughly three‑quarters of total workloads in 2025, future growth is expected to be increasingly driven by emerging technologies. In particular, the share of AI‑related workloads is projected to expand significantly within the overall mix – JLL 2026 Global Data Center Outlook.

Different national approaches to AI and data center development

Across the EU, differences in national policy, market dynamics, and electricity system constraints shape where and how data center capacity is developed.

In core hubs, such as Germany and France, data center development builds on already well-established digital ecosystems. Governments in these member states therefore focus on scaling existing capacity and strengthening AI capabilities.

This is reflected in national policy. Germany, for example, aims to double data center capacity and quadruple AI computing power by 2030, supported by around EUR 5bn in national programs and the hosting of an EU-supported AI factory. France combines public funding of EUR 2bn to EUR 3bn with the facilitation of approximately EUR 109bn in announced private and international investment commitments, alongside an AI factory under the same EU framework.

In contrast, in countries such as The Netherlands and Ireland, data center development is less driven by new government expansion strategies and more by market dynamics under increasing system constraints. In these countries, private investment remains the primary driver of new capacity, while grid congestion and permitting restrictions increasingly shape where and how data centers can be developed. In the Netherlands, this is complemented by approximately EUR 200m in national co-funding for participation in an EU-supported AI factory.

Outside the core hubs, several emerging markets focus more explicitly on attracting investment and scaling digital and AI infrastructure. In Poland, the government supports this development through approximately EUR 2.5bn in public AI-related funding, combined with participation in an EU-supported AI factory. Italy follows a more private‑led model, with around EUR 10bn in expected private investments between 2025 and 2026, while also participating in the EU AI factory framework.

Across all markets, EU-supported AI factories contribute to strengthening regional AI capabilities. However, large-scale data center expansion remains predominantly driven by private investment decisions, which are increasingly shaped by local electricity system constraints, permitting conditions, and access to suitable sites.

Data centers are highly dependent on access to the electricity grid and to reliable and affordable energy, not all of which must necessarily come from renewable sources. To assess how energy constraints may affect AI expansion within the EU, let us dive deeper into the role of data centers and AI computing in the context of the great electrification. Figure 2 illustrates the key dynamics.

Data centers consume significant amounts of electricity

Data centers are currently responsible for 2% of electricity consumption in the EU.[2] With the growing use of AI, this share is expected to increase to 5% by 2030, [3] which is equivalent to Poland’s yearly electricity consumption. Although data centers remain a fraction of total power consumption, they are set to account for a significant share of demand growth this decade. To put this into perspective: Between now and 2030, expected demand growth from data centers is comparable to the demand growth of electrified transport or electrified industry. As an illustration, a single AI training data center can consume as much electricity annually as around 100,000 households.

… that can support the business case of new solar and wind capacity

As clean and affordable power is an ever more important factor in the data center business case, operators increasingly contract electricity directly from wind and solar generators. The total renewable energy requirements are substantial, especially as the rollout of data centers intensifies.

However, rather than merely adding to the challenge, data centers can play an important role in the development of new large-scale solar PV and wind electricity generation capacity. To secure their clean energy needs and hedge against price risks, many data centers now procure their future electricity consumption through the means of power purchase agreements (PPAs) with solar PV or wind farms. These PPAs support the business case and bankability of renewable energy projects by providing long-term contracted revenue streams to the generation asset. In that regard, data centers can indirectly contribute to the expansion of renewable electricity supply needed to power the great electrification.

In the EU, data centers are the largest offtakers of PPAs that can subsequently have a big influence on the bankability of new renewable generation capacity. Data centers were the offtakers for more than a third of PPAs in the EU over the last ten years, and have supported the business case of a total 29GW of combined new solar and wind capacity (see figure 3).

[2] Source: BloombergNEF (2025). Numbers based on Global IT Capacity datacenters.

[3] JLL (2026) – 2026 Global Data Center Outlook.

Data centers add to the queue for scarce grid connections

Data center deployment faces its primary constraint in grid availability across Europe’s traditional data center hubs. Data centers are large, location-specific, and often continuous electricity consumers, which makes them challenging to integrate into existing infrastructure at the connection stage.

In most member states, grid connections are already scarce. As a result, new planned data centers increasingly compete with other electrification priorities, such as new housing projects, industrial electrification, EV charging networks, and the electrification of heating in existing buildings. In several member states, grid access is still allocated on a first‑come‑first‑served basis, contributing to long and often inefficient grid connection queues.

To address these challenges, several countries are revising their grid connection frameworks. Denmark, for example, is moving forward to a “first-ready-first-served” model that prioritizes projects that are ready for construction. The Netherlands is introducing priority groups for specific types of electricity demand in congested areas, while Spain is introducing auctions to allocate scarce connection capacity when demand exceeds availability.

As a result, grid connection capacity has become a decisive factor shaping data center planning and siting decisions across the EU, increasingly determining where new facilities can realistically be developed.

… but can also offer flexibility through on-site generation, storage, and load shifting

However, despite these constraints, data centers can provide flexibility to the electricity system.

As grid queues grow, data center developers increasingly consider on-site energy solutions to secure reliable power. In the US, this already includes dedicated gas-fired generation to bypass grid connection delays. More broadly, options such as solar PV and wind turbines co-located with batteries, fuel cells, and even small modular nuclear reactors are being explored, often as part of integrated microgrid solutions. We expect this trend to emerge in the EU as well, perhaps even driven by legislation, with Ireland providing the first example. In Ireland, new data centers must provide behind‑the‑meter generation equal to 100% of the grid connection and be sited in unconstrained locations. In addition, they are required to match 80% of annual demand with renewable energy investments.

Once connected, data centers can also provide operational flexibility. Through demand response, including load shifting and the use of on-site storage, operators can adjust electricity consumption in response to system conditions. Research from Ember suggests this can help relieve grid congestion and, in some cases, enable more flexible connection agreements, potentially shortening waiting times.

The distinction is important: While data centers are difficult to accommodate in grid connection planning due to their size and location needs, they can still contribute to system flexibility during operation. This flexibility can help to better utilize existing grid capacity and, in some cases, support more flexible grid connection agreements.

Beyond factors such as electricity costs, permitting issues, and developments within the electricity system, data centers in the EU are shaped by local socioeconomic conditions. These conditions influence how new projects are received by local communities and policymakers.

Across the EU, local debates typically center around the allocation of scarce resources, as demonstrated in the examples in the text box. Data centers, and hyperscalers in particular, also require substantial amounts of land and access to water. At the same time, hyperscale facilities generate relatively few long‑term local jobs, shaping perceptions of their value relative to their physical footprint and electricity demand.

As a result, local backlash has become a significant element in the development of new facilities. It influences permitting processes and timelines, site selection, and regional planning discussions. As these pressures Intensify, they are likely to play a growing role in shaping the pace and geographic distribution of future data center expansion in the EU.

Current developments and outlook to 2030

The EU’s ambitions for the data center sector remain high, with the EC’s strategy referring to a potential tripling of capacity by 2030–2032. Achieving this would require a rapid scaling of electricity supply and grid connections. However, current conditions point to a widening gap between ambition and reality. Limited grid capacity, increasing competition for connection points, and stricter permitting conditions are already slowing development across several member states.

Against this backdrop, the International Energy Agency expects European data center demand to grow by around 70% by 2030. Assuming a similar growth rate for the EU, this would imply a demand of roughly 17GW by 2030, compared to around 10GW today. This indicates that grid availability, rather than capital or demand, is likely to be the binding constraint in the near term.

Data center developers are increasingly exploring alternative approaches to secure power and accelerate project deployment, including on-site power generation, co‑location with energy assets, and emerging microgrid solutions. These methods can help individual projects move forward, but they are unlikely to unlock the scale of additional capacity implied by current policy ambitions within this decade.

Overall, RaboResearch considers the goal of tripling EU data center capacity by 2030–2032 as highly ambitious and not within reach, as grid limitations remain the primary bottleneck, and alternative solutions are unlikely to close the gap within this timeframe. In this context, the IEA’s projection of around 70% growth by 2030 appears more plausible, as it reflects a trajectory that is more consistent with the broader limitations shaping Europe’s data center expansion. Such an outcome would correspond to an installed capacity of 17GW. If realized, it would also mitigate some of the (future) pressure on grids and electricity demand.

Outlook to 2040

Looking beyond 2030, as grid capacity and supporting infrastructure gradually expand, data center expansion is expected to accelerate again, driven by continued growth in AI-related computing demand. This implies a two-phase trajectory, with constrained growth toward 2030, followed by a renewed expansion phase as grid capacity and supporting infrastructure scale up. Projections for EU data center capacity toward 2040 vary widely. However, RaboResearch considers a continuation of recent growth trends to be the most likely outcome. In an upside scenario, advances in model architecture and systems design could reduce the energy intensity of AI workloads, although the durability of such gains remains uncertain. Under this trajectory, EU data center capacity could reach around 75GW by 2040 (see figure 4).

At this capacity level, a 75GW data center fleet would consume roughly 510TWh of power per year. In comparison, projected electricity demand in 2040 amounts to about 1,050TWh in industry, 500TWh in transport, and 1,750TWh in the residential and built environment sectors. Even under a conservative outlook, data center electricity use might thus exceed electricity demand in the transport sector and equal about half of the electricity used by industry. Several higher‑growth outlooks from other forecasters show that data center consumption could eventually surpass the electricity demand of any major individual sector later in the 2040s if expansion conditions remain favorable.

Growing adoption of AI applications, in combination with strong demand and sustained investments, supports further expansion. However, severe grid and wider energy‑system constraints remain a major obstacle to scaling the sector. The accelerating shift toward on‑site generation and other decentralized energy solutions is becoming essential to overcome these bottlenecks and enable additional growth over time. This shift reflects how underlying demand continues to rise, and while the grid cannot currently accommodate it, the market is already responding by pursuing these workarounds.

Figure 5 summarizes the projected EU electricity demand across sectors in 2030 and 2040. Industry and residential power demand increases only moderately, while transport is expected to grow more strongly. Data centers show the steepest demand growth, rising from about an estimated 115TWh in 2030 to more than 510TWh in 2040.[4] Even under conservative assumptions, data centers would become one of the fastest‑growing off-takers of electricity during the outlook period.

[4] The estimate of 510TWh is based on a load factor of 0.6 and a Power Usage Effectiveness (PUE) of 1.3.

Looking ahead, the key question is under what conditions additional data center capacity can be delivered. As demand for AI accelerates, developers are likely to adopt new approaches – such as on-site generation, co-location with energy infrastructure, and microgrids – to bypass grid constraints and reduce reliance on traditional connections. While these solutions may accelerate development in areas where the grid is a limiting factor, they are unlikely to unlock the required scale at system level.

If the EU aims to remain competitive in AI, data centers will require clearer space to grow. Policy adjustments at both the EC and member state levels may support this by aligning energy planning, permitting, and infrastructure development with broader AI ambitions. Such changes could create conditions for growth that exceed the current RaboResearch projection of 75GW by 2040, particularly if market momentum and technological innovation continue to strengthen while constraints in grid capacity and access to renewable electricity ease.

One conclusion is clear: Data center capacity across the EU that will continue to expand, driving demand for substantial volumes of electricity. Within the broader context of the great electrification, this will place increasing pressure on grid infrastructure, while also creating opportunities to accelerate the financing and deployment of renewable generation. Grid capacity and access to renewable electricity through PPAs will therefore be decisive in determining where, and how quickly, data center capacity can be deployed, with on-site generation and related solutions playing a complementary role in enabling individual projects to move forward.