Data centers could raise EU electricity demand by up to 14%, Czech outlook uncertain

According to a study by US firm Goldman Sachs, electricity consumption in Europe could rise by more than 40% by 2033. Alongside the electrification of transport and industry, the development of artificial intelligence and the related construction of data centers could also be a key factor. Their capacity is currently growing primarily in Germany, the United Kingdom and France. Future development in the Czech Republic remains uncertain for now.
In recent years, European electricity demand has been hit by several shocks, including the Covid-19 pandemic and the energy crisis. According to a study by US bank Goldman Sachs, this has slowed the electrification of individual sectors compared with original expectations. As a result, electricity consumption on the continent has fallen cumulatively by approximately 10% since 2008.
This trend is likely to be reversed in the coming years by the electrification of transport, industry and heating. Rapid development of artificial intelligence and data centers should also make a significant contribution to increased demand. Data centers provide the essential infrastructure for the computing power that artificial intelligence requires to function.
According to the scenarios presented in the study, electricity demand in Europe is set to rise by 34 to 48% by 2033. Scenarios assuming substantial growth in data center consumption estimate a cumulative increase in demand of 42% or more. In these scenarios, data centers account for 8 to 14% of newly demanded electricity. In other words, consumption driven by artificial intelligence applications and data center construction is expected to generate additional demand of 220 TWh in Europe. This is roughly four times the Czech Republic’s annual electricity consumption in 2023.
The materialisation of such development scenarios currently appears realistic. Significant data center expansion is already evident today, for example as a result of higher demand for cloud storage and streaming services. Major technology companies such as Google and Amazon are also contributing, as they need to secure computing capacity for their products.
According to a report prepared for the European Commission, data centers in the EU consumed an estimated 45 to 65 TWh of electricity in 2022 (1.8 to 2.6% of total EU electricity consumption).
AI data centers consume up to 10 times more electricity
A data center is a physical room, building or facility housing computing infrastructure, including servers, storage systems, network equipment and cooling mechanisms. Servers are responsible for carrying out computing tasks, storage systems provide capacity for storing data and applications, and network equipment enables communication and data transfer.
Given the nature of computing equipment concentrated in a small area and the generation of large amounts of heat, efficient cooling systems are an essential part of the entire facility to prevent overheating. Alongside servers, these systems represent the largest item in the overall electricity consumption balance. Depending on the type of server and cooling technology, they account for 35 to 45% of total consumption.
Traditional cooling equipment has a capacity of 5 to 15 kilowatts. By contrast, equipment used for artificial intelligence has a capacity of 50 to 100 kilowatts. According to the findings of academic studies, total data center consumption could increase tenfold as the use of artificial intelligence grows.
Development is concentrated mainly in countries with cheap electricity and financial centers
The high consumption requirements of data centers suggest they cannot be built at scale just anywhere in Europe. According to the authors of the above study, demand for data centers will be particularly strong in two areas. First, in countries with cheap baseload electricity, such as Spain, France and Northern European countries. The second group will be regions and countries with a high concentration of financial and technology companies. These include the United Kingdom, Germany and Ireland.
These assumptions are already reflected in current trends. The largest numbers of data centers are located in Germany (522), the United Kingdom (517), France (315) and the Netherlands (300). The largest facility currently built is in London, one of Europe’s financial centers, with a power capacity of 1,053 megawatts. It is followed in size by data centers in Frankfurt, Amsterdam, Paris and Dublin.
The Czech Republic is not yet on the list of attractive locations
The possibility of building a data center was recently discussed in the Czech media after Minister Josef Síkela held talks with technology company Microsoft about its construction. At the beginning of the year before last, the company bought a nearly six-hectare plot in Prague 9 for CZK 1.5 billion, where a new data center was to be built. The future of the project remains uncertain, however.
According to public statements, the main reason construction has not gone ahead is the high bureaucratic burden associated with building it. Taking into account the assumptions in the Goldman Sachs study, uncertainty over future baseload price developments or insufficient development of renewable energy sources may also be behind the uncertain outlook for data centers in the Czech Republic. Available renewable electricity is currently one of the key requirements of global technology companies, which often have very ambitious plans to reduce their carbon footprint.
Significant data center development in the Czech Republic is also not envisaged by the Resource adequacy assessment published in 2023. It is published annually by the transmission system operator, ČEPS, and includes, among other things, electricity consumption development scenarios and an assessment of the main factors affecting consumption in the Czech Republic over the coming decades.
Translation disclaimer
This article is a machine translation of the Czech original and has not yet been fully reviewed. In case of any doubt, please refer to the Czech version.




