Summary
- Schneider Electric chief digital officer Peter Weckesser has warned that Europe's electricity costs and grid limitations could discourage AI infrastructure investment.
- European Commission figures show EU data centres consumed 68TWh in 2024, with demand projected to reach 114TWh by 2030.
- Investment decisions depend on power prices, reliable connections, infrastructure costs, and regulatory certainty; no specific project cancellation has been established by the warning.
Europe risks losing artificial intelligence infrastructure investment to regions with cheaper electricity and more accessible power networks, according to Schneider Electric chief digital officer Peter Weckesser, who has warned that energy costs are becoming an increasingly important factor in where computing capacity is developed.
Speaking on 19 November, Weckesser identified high European electricity prices, insufficient grid investment and fragmented regulation as obstacles to expanding AI infrastructure. He said the United States remained ahead in data centre construction, followed by China, while other regions including the Middle East were expanding more rapidly than Europe.
The warning concerns potential future investment decisions rather than confirmed cancellations of individual European projects. Data centre operators compare locations according to the cost and availability of electricity, construction requirements and the time required to connect facilities to essential infrastructure.
Electricity is a central operating cost for AI computing because advanced processors, networking equipment and cooling systems require substantial power. The scale of that requirement is increasing as technology companies and other organisations invest in more computing capacity to train models and operate AI applications.
Power costs influence the economics of computing
A data centre’s commercial performance depends on several connected factors, including the cost of constructing the facility, the computing equipment installed and the expenses incurred throughout its operation. Electricity can account for a substantial proportion of those recurring costs, particularly in facilities supporting power-intensive workloads.
Higher prices do not automatically make a development unviable, but they can influence where operators locate new computing capacity. The effect depends on the expected utilisation of the equipment, the amount of electricity consumed and the prices customers are prepared to pay for computing services.
AI workloads create particular challenges because they can require dense concentrations of processing equipment. The power needed to operate that equipment must be accompanied by adequate cooling, electrical distribution and network connectivity, with each system contributing to the facility’s capital and operating requirements.
Operators can improve efficiency through hardware selection, cooling technology and the way computing workloads are managed. However, the effect of those improvements depends on whether reduced electricity consumption per unit of computing is offset by increased demand for the services.
A more efficient facility may support a larger volume of work without proportional increases in energy demand, but overall electricity consumption can still grow as additional processors are installed. Infrastructure planning must therefore consider both the efficiency of individual systems and the expected scale of computing activity.
Long-term electricity purchasing arrangements may help operators manage exposure to changing prices. Nevertheless, the availability and commercial terms of those arrangements vary across locations, and electricity contracts cannot overcome every limitation associated with grid connections or the capacity of local infrastructure.
Grid availability can be more restrictive than price
A developer may secure land and financing for a data centre while still lacking the electricity connection required to operate it. Network capacity can constrain a project even where the buildings and computing equipment could otherwise be completed relatively quickly.
Electricity grids require investment in transmission and distribution infrastructure, with major projects involving planning, regulatory approval and construction work. These processes can extend beyond the development timetable of an individual computing facility.
Connection availability also depends on competing demands. Industrial electrification, transport infrastructure and changing patterns of energy generation can all increase pressure on electricity networks, while different regions experience those demands unevenly.
Data centre developers consequently need to assess the capacity available at specific locations rather than rely on national electricity prices alone. An area with attractive energy costs may have limited spare network capacity, while another may offer faster access to power at a higher recurring price.
Expansion creates further complications because facilities are frequently developed in phases. An operator might connect initial computing capacity but require additional infrastructure before installing further equipment, making the availability of future electricity supply part of the original investment assessment.
The European Commission and national authorities are pursuing measures intended to improve electricity network planning and accelerate infrastructure development. However, reducing administrative delays cannot immediately create physical capacity where new substations, transmission lines or generation resources are required.
EU electricity demand is projected to rise sharply
A European Commission report published in October 2026 estimated that data centres in the EU consumed 68 terawatt-hours of electricity during 2024. Consumption is projected to reach 114TWh by 2030, equivalent to approximately 3.2% of the Union’s electricity demand.
The projected increase illustrates the growing relationship between Europe’s digital infrastructure ambitions and energy planning. Computing facilities are becoming more substantial electricity consumers at the same time that other industries are seeking to replace fossil fuel processes with electrical technologies.
The Commission’s assessment also examines the efficiency and sustainability of data centres, including electricity consumption, water use and the recovery of waste heat. These issues can influence the environmental performance of facilities as well as their operating economics.
Cooling accounts for part of a data centre’s electricity and resource requirements, making more efficient approaches potentially valuable where computing hardware produces considerable heat. The choice of cooling technology depends on facility design, equipment density and local environmental conditions.
Waste heat recovery can provide opportunities where facilities are located near heating networks or other suitable users of recovered thermal energy. However, the practical benefit depends on infrastructure connections, the temperature of available heat and whether there is sufficient demand at appropriate times.
These efficiency measures can improve how resources are used, but they do not eliminate the requirement for reliable electrical supply. A facility with advanced cooling still needs sufficient power to operate its computing equipment, and a connection delay can prevent the project from becoming operational.
European computing investment faces competing pressures
European policymakers want to increase regional AI computing capacity and reduce certain dependencies on infrastructure located outside the continent. Greater domestic capacity could support research, commercial AI deployment and organisations with particular requirements concerning data control.
Those ambitions depend on attracting investors willing to finance expensive facilities and maintain them over many years. Operators must consider the expected demand for their services alongside hardware costs, energy prices and the availability of network capacity.
International competition can influence those calculations. Some computing workloads can be located in different regions with relatively limited consequences, while others need to remain close to users or within particular jurisdictions because of latency, contractual or regulatory requirements.
Locations with cheaper electricity or more readily available infrastructure may appeal particularly to operators developing internationally mobile computing services. Conversely, European facilities may retain advantages where customers need regional processing capacity, specialised network connections or tighter control over data.
Regulatory fragmentation can make project planning more complicated when operators expand across several countries. Different permitting procedures, electricity market arrangements and planning requirements can affect the time needed to secure approval and begin construction.
These differences do not necessarily indicate a single appropriate European policy response. Electricity networks and energy markets vary substantially across the continent, meaning infrastructure constraints and suitable investments must be assessed according to national and local conditions.
Investment decisions will provide the evidence
Schneider Electric supplies electrical infrastructure, energy management and automation technology used in data centres, placing the company close to developments in computing infrastructure and electricity systems. Weckesser’s warning reflects the commercial pressure created when facilities require large quantities of dependable power.
The concerns are consistent with the European Commission’s assessment of growing electricity demand and the need for more effective coordination between energy systems and digital infrastructure development. However, projected demand and higher costs do not establish that a particular investment has already moved away from Europe.
The outcome will depend on how electricity prices develop, whether grid improvements are delivered and how quickly data centre operators can obtain the power required for new facilities. Changes in AI hardware efficiency and the types of workloads customers purchase will also influence demand.
European investment plans must therefore be assessed alongside the energy infrastructure required to support them. Commitments to expand AI computing capacity become commercially meaningful when sites have reliable connections, equipment can be installed and facilities can operate at costs competitive with available alternatives.
Weckesser’s warning places the economics of electricity alongside processors and software as a constraint on European AI development. The next evidence of its effect will come from actual infrastructure investments, project delivery and the location choices made by companies financing the next generation of computing facilities.












