The race for data center power solutions is testing electricity grids, water supplies and local trust as AI facilities grow faster than infrastructure, leaving governments to decide who pays for the expansion, according to Geneva Solutions.
The United Nations Economic Commission for Europe (UNECE) laid a warning that data centers can be built within two to five years, while new transmission lines may take more than a decade. The International Energy Agency (IEA) says data center electricity demand rose 17% in 2025.
That gap between data center power solutions building computers and supplying them with power is becoming a problem for households, businesses, and the communities asked to host the sites.
AI Grows Faster Than Grids
A data center power solutions need electricity to run its computer chips, move and store information, and keep equipment cool. AI adds intensive computing work, training models, answering requests, and running agents that can carry out longer tasks. Those machines also produce heat, increasing the need for cooling and, in some facilities, water.
That is why powering AI takes more than plugging a new building into an existing network. Electricity must reach the site reliably, including when demand changes quickly, or other users need power at the same time.
According to the IEA, sharp swings in demand at AI facilities can even stretch the ability of onsite gas plants to serve them.
The scale of data center electricity consumption is rising despite better technology. The IEA says the power used for each AI task is falling rapidly. But more people are using AI, while energy intensive applications such as agents are expanding. Efficiency reduces the electricity needed for one task, it does not guarantee lower total use when the number of tasks grows faster.
UNECE projects that global data centers will use about 950 terawatt-hours of electricity in 2030, up from approximately 485 terawatt-hours in 2025. It estimates that this would equal about 3% of global electricity demand. Yet the local effect can be much greater when several large sites seek connections to the same grid.
Those concentrated data center power requirements can force utilities to expand generation, substations, and transmission lines.
If work falls behind, new connections may be delayed, and the local grid may have less room to handle surges in demand. The risk depends on the capacity of each network and how projects are planned, a data center does not automatically cause an outage.
As things stand right now, grid utility data center load demand is being measured early enough to dictate those decisions. UNECE revealed that large facilities are arriving faster than electricity infrastructure in many regions. Its warning puts pressure on governments to examine proposed sites alongside existing demand, planned housing and industry, rather than approve each project in isolation.
Who Pays When AI Comes to Town?
More power for data centers could raise electricity bills if new plants and grid upgrades are paid for by a utility’s wider customer base.
The IEA says higher demand does not necessarily mean higher prices when policies and infrastructure investment are handled well. It also warns that large facilities can trigger costly work because they use so much power in one place and grow quickly.
In parallel, UNECE says the allocation of those costs remains unresolved in many cases. Its proposed approach calls for clear rules that divide costs and risks fairly among developers, infrastructure providers and consumers.
That would help determine whether the companies creating new demand pay for the capacity they need, or whether households and smaller businesses help fund it through their bills.
The cost debate extends beyond electricity. Geneva Solutions reports growing opposition from communities concerned about water use and the limited number of permanent jobs at large facilities.
“Very few people work in data centers; it carries huge energy and water demand, so local communities are asking what’s in it for us,” according to Dario Liguti, UNECE Sustainable Energy Director.
Governments therefore need reliable public figures on data center electricity consumption, water needs and the upgrades each project requires. Without them, residents cannot judge the promised economic gains against the resources their area will provide. Clear reporting also gives utilities a stronger basis for forecasting demand and explaining who will pay.
Some data center power solutions could ease that pressure. The IEA says technology companies signed around 40% of corporate renewable power purchase agreements in 2025, while developers are exploring nuclear power, geothermal energy, onsite gas, and batteries. Each option has different costs, construction times, and effects on the grid.
Using renewable energy sources for data centers can add cleaner supply, but a purchase agreement alone does not show whether electricity will be available at the hour and place a facility needs it. Grid connections, storage, and the ability to reduce or shift some work still matters.
UNECE identifies flexibility, energy storage and clearer data as ways to manage large new loads, which makes data center power systems a planning issue for governments, as well as a design one for operators.
More efficient chips and cooling may reduce waste inside a building. But what they cannot, by themselves, is build transmission lines, settle water disputes, or even decide the cost of new infrastructure that appears on an electricity bill.
The IEA puts this challenge as a central one, plainly saying “there is no AI without energy.”
As powering AI expands, governments must decide how much supply to add, where facilities can be supported, and what operators should pay before construction moves ahead.
For developers, data center power requirements now shape where projects can go and how quickly they can open. For communities, the test is whether those projects bring enough local value to justify their demands on electricity, water, and land. Geneva Solutions reports that opposition has already delayed or cancelled proposed facilities in the US.
Future green power supply models for data centers will need to account for the full local burden, including backup supply and grid upgrades.
Expanding renewable energy sources for data centers may help meet demand more cleanly, but transparent costs and long-term planning will determine whether communities accept the growth.
The race for data center power solutions is therefore also a decision about who carries the risk. If governments require developers and utilities to explain demand, water use and costs before approving projects, they can give residents a clearer answer. Without that work, communities may see the bills long before they see the benefits.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Impact section to stay informed and up-to-date with our daily articles.
