Digitalization is often presented as a tool for progress, but its uncontrolled expansion also has a dark side. The growing demand for data centers, fueled in particular by the popularity of artificial intelligence (AI), risks undermining the European Union's climate goals, the destabilization of energy systems and the deepening of environmental and social inequalities.

Growing energy demand and the impact of AI

AI infrastructure is emerging as one of the biggest environmental challenges of the digital age. Data centers, which serve as the physical backbone of digital services, require vast amounts of energy to power servers and cool them. International Energy Agency (IEA) explicitly identified AI-powered data centers as a major structural factor in the global increase in electricity demand by 4.3 % in 2024. In addition, data centers contributed to the growth of electricity consumption in the building sector by almost 60 % last year.

In Europe, data center consumption is expected to triple by 2030, rising from around 62 terawatt hours (TWh) to over 150 TWh, increasing their share of total European consumption from 2 terawatt hours (TWh) to around 5 terawatt hours (TWh). In comparison, in Ireland, data center energy demand already accounts for around 18 terawatt hours (TWh) of the country’s total electricity consumption.

AI algorithms—especially the training stages of complex models—require enormous computing power and specialized hardware, such as graphics processing units (GPUs) and tensor processing units (TPUs). This is leading to the massive expansion of hyperscale data centers, which are designed for scalability and high throughput.

Environmental and social consequences

The impact of data centers goes beyond electricity consumption. They also place a significant strain on water resources, as they require large volumes of clean water for cooling. For example, one Google data center uses about 450,000 gallons (about 1.7 million liters) of water per day. AI models like ChatGPT alone use about two liters of water for every 10 to 50 queries they make.

Furthermore, rapid hardware obsolescence (e.g. GPUs with a lifespan of 3–5 years) contributes to massive e-waste production. The construction of data center infrastructure requires extensive mining of raw materials, including copper, lithium, and cobalt, which exacerbates environmental degradation and social problems in supply chains.

Social inequalities are also exacerbated at the local level. In Virginia (USA) and Dublin (Ireland), local communities are complaining about excessive water use and strain on energy networks, which can lead to rising energy costs for low-income households.

Insufficient EU strategy and recommendations

The European Union currently lacks a coherent vision for regulating AI and its infrastructure. While initiatives such as the Energy Efficiency Directive (EED) require mandatory energy consumption monitoring (for data centers >500 kW), they do not impose binding targets for reducing energy and water consumption.

Current sustainability efforts are largely focused on efficiency. However, this “efficiency trap” (Jevons’ paradox) is only a temporary solution, as wider and cheaper use of technologies negates environmental gains. Furthermore, the use of Guarantees of Origin (GOs) or Renewable Energy Certificates (RECs) allows companies to claim that they are “100% renewable” even though their electricity comes from fossil fuel-dependent grids (so-called greenwashing).

Therefore, it is crucial to accept principles of sufficiencyThis means:

1. Introduction of legally binding, sector-specific targets to reduce energy consumption and emissions.
2. Optimization of existing infrastructure before permitting new construction.
3. Prioritizing AI applications with high societal value (e.g. healthcare or science) over non-essential applications such as personalized advertising or mass content generation.
4. Integrating data centers into national energy strategies and ensuring that renewables meet strict “additionality” criteria.

Without ambitious, enforceable and holistic targets that go beyond mere efficiency and transparency, the EU's climate goals are at risk. JRi