Look at a map of major data-centre developments and a pattern quickly emerges. Facilities cluster around a relatively small number of places: major internet exchanges, telecommunications hubs, power infrastructure and regions able to offer suitable land and planning conditions.
- Electricity is the first requirement
- Why connectivity still matters
- Does a cold climate make the perfect location?
- The Eemshaven example
- The grid connection queue becomes a battleground
- Local benefits—and local costs
- Why not put every data centre in the countryside?
- The next map of the digital economy
- Sources and further reading
It can seem odd that buildings holding the world’s digital information are concentrated in particular industrial corridors. Surely the internet means data can be stored anywhere? In reality, a data centre’s location is a strategic decision shaped by physics, economics and infrastructure.
Electricity is the first requirement
A data centre needs a continuous and reliable supply of electricity. Servers run around the clock, and a power interruption can disrupt services unless backup systems take over. Large facilities therefore require substantial grid connections, transformers, switchgear and backup arrangements.
The rapid growth of AI has made power availability even more important. High-performance AI systems can concentrate enormous demand into relatively small buildings. The International Energy Agency’s 2026 update estimated that global data-centre electricity consumption could rise from about 485 terawatt-hours in 2025 to around 950 TWh in 2030, with AI-focused facilities growing faster.
That demand is not evenly distributed. A data centre can be a major new customer for a local electricity network, even if the sector’s share of global consumption looks comparatively modest. Local grid capacity, the availability of new generation and the time required to build substations and transmission lines can determine whether a proposed site is viable.
Why connectivity still matters
Data centres need fast, reliable connections to the rest of the internet. Fibre-optic cables carry data between facilities, businesses, homes and international networks. Sites near major telecommunications routes and internet exchanges can reduce latency and make it easier to connect customers to cloud services.
Some workloads are highly sensitive to delay. Financial trading, video calls, online gaming and interactive applications may benefit from locating computing capacity close to users. Other tasks, such as training a large AI model or storing backups, can be carried out farther away if sufficient bandwidth is available.
This creates different kinds of data-centre location. A large campus may be built where power and land are available, while smaller facilities are placed close to cities and network interconnections.
Does a cold climate make the perfect location?
Cooler air can reduce the amount of mechanical cooling a facility needs. In suitable conditions, operators can use outside air or other forms of free cooling for much of the year. This is one reason Nordic countries have attracted data-centre investment.
But temperature is only one factor. A site also needs reliable electricity, fibre connectivity, suitable land, skilled staff, access to equipment and a stable regulatory environment. Water availability matters too, particularly where cooling systems rely on evaporation.
Renewable electricity is another attraction, but the presence of wind turbines nearby does not prove that a facility runs entirely on renewable power at every hour. Operators may use power-purchase agreements, certificates or grid electricity, and the physical mix of generation varies over time. Understanding the real energy footprint requires more than looking at a photograph of turbines beside a server building.
The Eemshaven example
An aerial photograph of the Google data centre at Eemshaven in the Netherlands shows the facility surrounded by wind turbines, with energy infrastructure nearby. It is a useful illustration of the way data-centre development intersects with regional energy systems.
It also highlights a wider point: data centres are not just computer buildings. They are industrial-scale electricity customers that depend on networks, cooling systems, construction materials, security and transport. Their environmental and economic impact is therefore connected to the region around them.
The grid connection queue becomes a battleground
Britain offers a particularly clear example of the tension between digital investment and limited infrastructure. In July 2026, Ofgem said demand-connection applications had risen from 41 gigawatts to 125 GW in less than a year, with data-centre projects accounting for a large share. The regulator proposed measures to tackle speculative applications and help viable projects connect sooner.
The numbers describe applications and queue capacity, not electricity already being consumed. A proposed data centre may never be built, and a project’s place in a queue does not mean it has a completed power connection. Nevertheless, speculative requests can complicate planning and make it harder to identify which developments are genuinely ready to proceed.
The International Energy Agency has also warned that data-centre projects can be built faster than the electricity infrastructure needed to supply them. Transformers, grid equipment, generation and planning approvals may take years, creating a mismatch between technology investment and energy-system delivery.
Local benefits—and local costs
Data centres can bring investment, construction work, tax revenue and some permanent technical jobs. They can support digital services, research and businesses that depend on reliable computing. In some locations, waste heat can be reused in district-heating systems or industrial processes.
But the benefits and costs are not always distributed evenly. A facility may employ relatively few people once construction is complete, while requiring a large electricity connection and significant land. Residents may be concerned about noise from cooling equipment and backup generators, water consumption, landscape changes and the opportunity cost of grid capacity.
These concerns are not automatically proof that a project is harmful or unnecessary. They are reasons to examine proposals with transparent information: expected electricity demand, water sources, backup generation, noise, heat-reuse plans, employment and the effect on local infrastructure.
Why not put every data centre in the countryside?
Remote locations can offer land and sometimes access to energy, but distance creates other problems. Long-distance fibre links add cost and may increase latency. Construction teams, replacement equipment and specialist staff need to reach the site. A remote facility may also require new power lines, roads and telecommunications infrastructure.
For some workloads, proximity to users is essential. For others, remote campuses are entirely practical. The industry is therefore likely to remain a mix of large regional campuses and smaller facilities near population and network centres.
The next map of the digital economy
Data-centre geography is becoming a question of public policy as much as commercial strategy. Governments want investment in AI and cloud computing, but they must also plan electricity networks, water resources, land use and the needs of existing communities.
The most attractive location on a company’s spreadsheet may not be the best location for the wider energy system. A more responsible approach would assess grid capacity before granting approval, account for water stress, require clear reporting and encourage facilities to make productive use of waste heat and flexible power demand.
The internet may be global, but the buildings that sustain it are local. Their location determines who gets the investment, who uses the electricity, which water resources are affected and which communities live beside the infrastructure of the digital age.
Sources and further reading
- International Energy Agency (2026), Key Questions on Energy and AI, executive summary.
- International Energy Agency (2025), Energy and AI.
- Ofgem (29 July 2026) – Proposals to tackle speculative data-centre grid connection projects.
- House of Commons Library (2026), “Data centres: planning policy, sustainability, and resilience”.
- Image: Wvdp (2023), “Google data center and wind turbines in Eemshaven”, Wikimedia Commons, CC0 1.0.
