Europe Promotes Energy Use in Data Centers While Millions Go Cold

William Burns is a fellow at Tech Policy Press.

Understanding where data centers fit within Europe’s broader energy landscape is challenging. Part of the difficulty lies in the limited data on their energy and resource use, but it also reflects wider uncertainty about the future of Europe’s energy systems. The issue sits awkwardly between policy communities in energy, environment and information technology, fields that have historically had little interaction and where complex systems, both natural and artificial, face off against one another. But one thread that runs through the issue is fairness in a continent where energy and other resources are perpetually scarce.

As we speak, many low-income households cannot afford to heat their homes adequately. For some, staying alive through the winter is difficult, as cold weakens their immune systems and leaves them vulnerable to serious respiratory infections. The numbers of cold-related deaths, which sum to hundreds of thousands every winter, mainly in eastern Europe, are predicted to rise due to climate change. It begs the question as to why energy is expended on data centers when it is needed so desperately elsewhere, and, as lethal as the current policy failure seems, it could worsen.

The energy system is complex and therefore the effect of significant new demands upon it is unpredictable. After decades of deindustrialization and falling power demand, engineers and officials in many regions of Europe have not experienced the opposite. Furthermore, integration of data centers into local electricity grids presents a significant technical challenge. The recent Iberian power blackout, although not linked with data centers, illustrated just how unpredictable electricity systems have become, with explanations of why it happened still unavailable even to experts.

When European governments, such as Portugal, propose to build data centers on the basis of spurious claims of economic growth, they are therefore proposing a giant experiment. If it goes wrong, it could lead to the poorest in society falling into debt to cover energy bills, skimping on heating and even shivering to death, as the additional energy demand exacerbates already high energy prices. Systemic impacts on the energy market that undermine industrial users, such as steelworks, cannot be ruled out either, with broader economic indices taking a hit, thereby reducing wages and employment levels. While the system-wide effects of data center expansion are obviously still untold, potential risks, which have often manifested themselves in large industrial projects on other continents, are not being addressed as openly and democratically as they ought to be.

World of scarcity

In a tiny continent like Europe, with its arid south and cold, dark north, energy and resources have always been precious commodities. There is a limited amount of feedstocks such as gas (and even the sun only shines for a certain number of hours each day), a limited amount of generation and transmission equipment, a limited number of qualified personnel to build and maintain it, and a limited capacity of the Earth to act as its sewer for carbon dioxide and other pollutants. The EU’s legally binding 2030 target for primary energy consumption, 1210 million tonnes of oil equivalent (Mtoe), is an acknowledgement of a limit – although the bloc is not yet on track to meet it. In recent years, Europe has cut itself off from major fossil fuel sources like Russia, Libya, and Iran, while failing to build the renewables capacity needed to substitute them. By July 2022, the International Energy Agency (IEA) had declared the “first truly global energy crisis in history,” a “red alert for the EU” due to Russia's weaponization of gas. With less feedstock, rationing was on the cards, such as setting heating thermostats lower and enforcing demand reduction. This would actually have been fairer, but in the end, officials stuck to “market” mechanisms, shifting the burden of the war to the poorest in society as the gas price rose, albeit with mitigations such as cash subsidization.

It was in the context of this wider energy crisis that the European Commission received 62 written responses to its ”consultation” on a proposed energy rating scheme for data centers. Only a small number of critical voices from academia and civil society engaged with it, and of these, only two groups got to the crux of the matter: measuring the value data centers deliver for society. Slovakia’s “Elektrosmog a zdravie,” for example, called for “a distinction to be made between entertainment or other purposes. This means setting criteria to prioritize the use of technologies: give priority to those that are essential for the modern and efficient development of society and to penalise unjustified…uses…such as small daily tasks…AI should be used selectively for research that ensures some scientific progress.” The group is associated with a campaign to regulate high-intensity electromagnetic radiation from wireless networks, as it poses a health hazard. The campaign has so far been unsuccessful, and although its ideas are scientifically sound, they remain outside the political mainstream. In this case, its comments were the most valuable submitted and should have prompted a rethink.

Energy distribution in a world of scarcity raises ethical questions. As Marie Donnelly of Ireland’s Climate Change Advisory Council noted, “access to the grid…for the data centre that supports an Garda Síochána [police] would or could be considered a priority area. That would be a decision for the government as to how it would rank the prioritization.”

EU officials issued what they called “the first phase of the establishment of a common Union rating scheme for data centres.” While the scheme gestured toward basic reporting standards, what is needed is an effort to unpack “efficiency.” Few would contest useful work for light bulbs or washing machines, but it is not so simple for data centers, where the debate lies in defining what constitutes “useful work.” Tech Policy Press has already brought to light aspects of the data center story in the Americas, citing air pollution from the gas turbine rigs powering them. In Europe, data center build-out is confined to a handful of points on the map, such as Germany and Ireland, yet it raises wider questions of electricity, water, and land consumption, as well as the political economy of resources needed for essentials like food, heat, and shelter.

What happens next cannot be easily predicted because the levels of capital are practically unprecedented. The Big Tech firms need more computers—and more electricity—immediately, yet the pace of expansion and its endpoint remain uncertain. Policy, therefore, remains reactive rather than strategic. Nevertheless, EU positions on data centers serve as a canary in the mine for the credibility of social democratic and environmentalist movements. Bigger increases in demand could arise from electric vehicles and air conditioning, requiring even closer attention.

An EU report predicted 98.5-160 TWh of extra demand from data centers in Europe by 2030. In principle, the extra demand would be accommodated by using renewable energy, which, due to its efficiency, is better at converting primary energy into useful work. Unfortunately, at present, the EU relies overwhelmingly on fossil fuels and nuclear, the most expensive and polluting sources of energy, dependent on deeply exploitative, polluting supply chains. The political impetus to shift away from them seems to lessen by the day, with nuclear power and a softening on internal combustion engines actively promoted by influential European politicians like Mario Draghi. It therefore seems likely that increased demand will come from within the existing mix of energy technologies. This could, logically, mean a cut in other areas to stay within the limit while sustaining data center expansion (the binding 2030 limit of 1210 Mtoe is equivalent to about 14000 TWh, based on an IEA conversion factor).

Policy blindspots and lessons from history

No one has yet reconciled these issues. France’s policy to build more nuclear reactors, drawing explicitly on the rhetoric of Big Tech firms and potentially costing trillions of Euros, illustrates the influence of factors like hubris, prestige and conceptions of “national security,” rather than social democratic concerns such as public accountability and environmental protection. Decent energy policies involve commitment to probity, receptivity to accountability mechanisms such as environmental NGOs and public interest journalism, and public service-oriented science and research — all currently insufficient. Historical experience shows that solutions often emerge only after trial and error, and time is limited.

In Ireland, data centers account for a stunning 22% of total electricity demand and 50% of electricity demand in Dublin. In June, leftwing deputies in the Dáil attacked policies favoring AI and data centers over the environment and housing.“Is it credible for Ireland to keep allowing so many data centres to locate here? We, and others, have called for moratoriums until we can see how to square the circle,” Ivana Bacik, of the moderate left Labour Party, added.

The Irish government has made assumptions about the trajectory of energy demand over the next five years, which may not, of course, play out as predicted. But the problems it faces are in the present. The country has so far “squared the circle” by importing electricity from Britain through submarine cables known as “interconnectors” (a further “Celtic” interconnector to Brittany in France is currently under construction). It also installed a so-called “temporary emergency power station” fueled by natural gas in Dublin (built by Greek and American firms). Another part of Irish policy is tied up with completing the privatization of the electricity system through the Private Wires Bill. This could give private firms almost unfettered ability to obtain electricity by building their own power system, breaking the monopoly the government held on regulating electricity transmission.

Discussion in the run-up to the publication of the Irish government policy statement in July suggested there would be no absolute requirement for renewable electricity in these developments. The Climate Change Advisory Council – an Irish government body that offers commentary on energy policy – made a clear-cut point, namely, that current policy “leaves an open door for industry to ramp up its use of fossil fuels without capacity limitations.”

The latest government paper implies a commitment to renewables based on the claim that “the objective of any change is to accelerate the deployment of renewable generation and storage.” However, Irish labor unions still expressed skepticism with the statement that “this approach fundamentally misallocates Ireland's access to scarce infrastructure resources at a critical time for our climate commitments. We're essentially allowing data centres to try to jump the queue while ordinary customers and decarbonisation efforts are left waiting.” Friends of the Earth were also critical and called for a moratorium on data center expansion in Ireland; it seems there is general agreement that progressives should continue to lobby against the policy change, with debates continuing into 2026.

Energy efficiency and technological limitations

One of the “solutions” dangled for the scarcity of resources has been “energy efficiency.” This could refer both to cuts in primary energy by powering with renewables but also to final (used) energy due to “more efficient” computation. The International Energy Agency, a supporter of data centers as a means to revive the market for electricity equipment and a supposed motivator for decarbonization, bandied around potential increases in “efficiency” in its recent commentary. Put simply, technological improvement and learning by doing could solve the problem of scarcity.

Yet, all is not as it seems. Public sector R&D expenditure on climate change research has cratered since the pandemic, as recent analysis byAbbas AbdulRafiu and Chux Daniels showed. If techsolutionism was ever a valid theory, it has ceased to exist as a reality. The academic Tessa Dunlop points out another problem: energy efficiency is a “motherhood issue” in policy. “While generally supported by everyone in the abstract, these motherhood issues often lead to contradictory interpretations when concretized,” she wrote.

In its analysis, the European Commission has used the data center industry’s definition of energy efficiency, namely, power usage effectiveness (PUE). This is a ratio of total energy consumption to total energy consumption of IT equipment. Yet even the Commission’s own report noted the slippery nature of the measure. “Although PUE can already serve as a simplistic yet measurable metric for energy efficiency, improving the PUE could make other indicators worse off. For instance, if the improvement is achieved through shifting the cooling method to a more water-intensive one, this can worsen the Water Usage Effectiveness (WUE).

Spain’s progressive government has proposed further reporting requirements on data center operators within Spanish borders that include contributions to employment and the local and national economy. This is a good start but the conversation about the societal uses of energy, not just how it is generated and how “efficiently” it is used, will ultimately need to be systematic and subject to democratic checks and balances beyond firms simply making statements to the authorities.

Earlier in the year, academics Jessica Commins and Kristina Irio published an analysis of what they described as the “puzzle” of “law applicable to data centres’ development and construction.” It was an effort to find a path through the thicket of often impenetrable detail and interlocking policies and try to gain purchase on what seems to be a global system of production and consumption, in which individual data centers serve as nodes.

Commins and Irio highlighted issues such as limiting frivolous digital uses (cryptocurrency mining), promoting second-hand equipment, and addressing “dark data,” estimated at 40–90% of global storage. They emphasized global dependencies, adding that “third [non-EU] countries where regulation of data centres’ sustainability is significantly lower than in the EU could otherwise bear the brunt of the EU’s outsourcing of computing needs.”

Most of the future electricity demand for data centers is not predicted for Europe, but for the USA, according to the IEA. Large European firms in the electricity sector, including utilities like Enel and Iberdrola, as well as gas turbine manufacturer Schneider, see the opportunity to increase sales in the USA, where electricity demand had previously been static. Spain’s Iberdrola, trading as “Avangrid”, is, for example, already highly active in the USA, such as announcing a “US$20 billion investment.” With European capital and machinery deeply embedded in the US tech-industrial complex, political tangles for progressive policymakers only multiply.

What can be done?

In a sense, however, the solution for Europe is already on the table. The European Green Deal and associated ambitions around the “just transition” captured in the phrase Clean Energy for All, while certainly imperfect, suggested an overarching logic to balance different pressures on the energy system in a way that was progressive and beneficial for society as a whole. However, these have largely fallen by the wayside in European rhetoric as part of a broader offensive by the right that erroneously portrayed green policies as an economic dead end. Requipping the European energy system for renewables also required Europe to collaborate closely with China on technology, like solar panels, batteries and other relevant devices and software, which was risky given American and Gulf interests based on oil and gas that turned out to be unassailable.

An approach driven by green and democratic goals that could have paid off has now been dropped in favor of more ad hoc policies with few solid measures of accountability. It is in this tainted political environment that data center policy now lands, potentially exposing the shaky foundations on which the policy is being built. As always, the costs of mistakes will reverberate through complex energy systems that few of us understand. The costs will be paid by the people least able to complain, not by those making the decisions.