The expansion of artificial intelligence is creating growing infrastructure requirements for electricity, cooling, water and waste management, making access to physical resources an increasingly important factor in data-centre investment decisions.
Behind AI services are semiconductor manufacturing facilities, server halls, cooling installations, substations, transmission networks and water systems. The scale of this infrastructure is reflected in figures cited by MAT, which indicate that global data centres currently consume approximately 1.5% of the world’s electricity.
That share could rise to around 3% by 2030, when annual data-centre electricity demand may reach approximately 945 TWh. AI-accelerated servers are expected to account for a substantial part of the increase, with their electricity consumption projected to grow by around 30% annually under the report’s cited base-case scenario.
Water Consumption and Electronic Waste Increase
Electricity demand is only one part of the resource requirements associated with AI infrastructure. Data centres need water for cooling, while the power generation supplying their electricity also creates additional water demand.
Research cited by MAT estimates that the global water footprint of AI could reach between 4.2 billion and 6.6 billion cubic metres annually by 2027. Under an electricity-demand scenario of 945 TWh, the wider water footprint associated with data-centre power consumption could amount to approximately 9.3 billion cubic metres by 2030. Electronic waste is another growing concern. AI infrastructure could produce around 2.5 million tonnes of electronic waste annually by the end of the decade.
Energy Mix and Cooling Systems Shape Site Selection
The environmental impact of data centres varies according to the electricity sources available at each location and the cooling technologies used. Identical computing workloads can have different carbon and water footprints depending on the infrastructure supporting them.
Facilities connected to coal-intensive electricity systems have a different emissions profile from those supplied by nuclear power and renewable energy. Cooling technologies also involve trade-offs: systems that reduce electricity consumption may require more water, while alternatives that save water can have different energy requirements. Consequently, electricity use and water consumption cannot be assessed independently when evaluating the sustainability of data-centre projects.
Grid Capacity Becomes a Constraint for New Projects
Electricity network capacity is emerging as a major factor in determining where new facilities can be developed. MAT cites an International Energy Agency (IEA) estimate that approximately 20% of planned data-centre projects worldwide could face delays if electricity grid bottlenecks remain unresolved.
Europe faces a particular challenge because its relatively low-carbon electricity mix is accompanied by congested networks and lengthy permitting procedures. As a result, access to reliable grid connections, available electricity capacity, water resources and credible low-carbon power supplies is becoming increasingly important in determining the viability and location of future data-centre investments, alongside factors such as land costs and tax incentives.
