What does the AI boom mean for Europe's climate goals?

The rapid growth of AI data centers raises questions about the impact of this technology on the climate. From a European perspective, the boom in energy-intensive AI applications poses the risk of carbon leakage, among other things. This refers to the effect that companies, due to CO₂ pricing or other climate regulations - and potentially higher electricity prices - relocate their activities to countries with less stringent climate policies. The Federal Environment Agency (UBA) has examined this risk together with the research institute Infras and the Roegen Centre for Sustainability in the context of the European emissions trading system.

The study shows that AI-related energy consumption in data centers will increase to approximately 300 terawatt-hours by 2028, which corresponds to about one percent of global electricity consumption. A large part of this consumption is fundamentally geographically flexible. This means that these calculations can theoretically be executed from anywhere. As a result, the data center industry gains enormous leeway in terms of location choice.

European AI requests cause emissions abroad

The study's results also show that current AI computing capacities and the projected growth are primarily concentrated in the USA and China - two jurisdictions with less stringent climate policies than the EU. This means that part of the European AI requests is processed in other countries and causes emissions there.

One reason for a smaller number of data centers in Europe could be carbon leakage. However, the current extent seems to be moderate. This is mainly due to the current expansion of the AI industry: The focus is more on developing new computing capacities rather than minimizing costs. Instead, historically grown cluster structures of the IT industry, as well as a stable energy supply and network infrastructure, seem to play the biggest role in site selection.

Carbon leakage could increase in the future

However, the incentives for carbon leakage could change in the future - for example, it is assumed that the cost pressure for the data center industry will increase. Furthermore, increasing cross-border data flows could increase imported emissions, even if such data flows do not intentionally circumvent CO₂ pricing.

For this reason, the expansion of renewable energies in AI-strong regions and compliance with the climate targets announced by AI companies are particularly important. Both reduce the weakening of European climate targets through carbon leakage and the import of emissions into the EU.