Plans to put artificial intelligence data centres on sites tied to the UK's fossil-fuel past are raising a question that reaches well beyond the tech industry: who carries the cost, in electricity and in carbon, of the computing behind chatbots and image tools.
Reports say the UK's most polluting power plant could supply electricity to what would be the country's largest AI data centre. Separately, plans have been revealed for an AI data centre at a former power station in North Yorkshire. Both have emerged alongside warnings that AI data centres could undermine the UK's emissions goals.
Old power stations draw developers
Former power station sites are attractive to data centre developers for practical reasons. They tend to come with large plots of land and high-voltage connections to the grid, which are among the hardest things to secure for a new facility. Getting a grid connection is a long process, and sites that already have one can move faster.
A working power plant offers a further advantage. A data centre sitting next to its supply can draw electricity at scale with less new infrastructure. It also ties the facility to the carbon content of whatever that plant burns, and that is where the emissions argument begins.
The emissions question
Data centres run around the clock, and the carbon they produce depends on how the electricity is generated. Powered by wind, solar or nuclear, the footprint is far lower than with gas or biomass-heavy supply.
Developers often point to renewable purchasing agreements. Such deals buy clean power on paper over a year, while a data centre draws electricity every hour of the day.
The UK has a legal commitment to reach net zero by 2050, and the government has also set an aim of clean power. A large new source of demand that leans on the most polluting generation would pull against both. Even if a facility is matched with clean supply on paper, the extra load still has to be met by the system on hours when wind and sun are scarce, and that gap is often filled by gas.
Who is affected
The most direct effects fall on the communities around the proposed sites. Residents near the former power station in North Yorkshire face the prospect of a new industrial neighbour, with the building work, traffic and visual change that come with a large facility. Supporters of such projects tend to argue they bring investment and skilled work to places that lost jobs when coal and gas stations closed. Whether that promise holds depends on how many permanent roles a data centre actually creates, which is usually far fewer than a power station once employed.
Electricity users across the country have a stake too. The grid is a shared system, and a very large new customer takes up capacity that other projects might otherwise have used. Housing developments, factories, charging networks and new renewable generators all compete for connections. If a single AI facility absorbs a big share of what is available in a region, others may wait longer or face higher costs to be linked up. How network upgrades are paid for, and how much of that falls on bill payers rather than on the developers who cause the demand, is a live concern.
Climate targets are shared as well. Emissions from a data centre in Yorkshire or at a gas-fired or biomass-fired site count towards a national total that every sector is expected to shrink. If one industry grows its footprint, others have to cut faster to compensate, and that burden reaches households, businesses and public bodies far from any server hall.
A tension inside government policy
The proposals sit awkwardly between two official priorities. Ministers want the UK to be a strong location for AI development and see computing capacity as part of that ambition. They also want a rapid shift to clean electricity and a path to net zero. Siting the biggest facilities beside polluting generation puts those goals in direct contact, and the warnings about emissions suggest the two will not always sit comfortably together.
Pressure is likely to grow on developers to show where their power will come from and how much of it will be low-carbon at the times it is used, rather than relying on annual accounting. Campaigners and climate advisers can be expected to press for firmer rules on that point.
What happens next
The plans are at an early stage, and neither project has cleared every hurdle. The North Yorkshire proposal will need to go through the planning system, where local authorities and residents can weigh in on land use, local impact and the energy supply. The reported link between the country's most polluting power plant and the largest planned AI data centre would depend on commercial agreements and on grid arrangements being settled.
Decisions by planners, the grid operator and the government will show how far the clean power aim shapes where data centres are allowed to go. If approvals come without firm conditions on the carbon content of supply, the emissions concerns are likely to sharpen. If conditions are attached, developers may have to look at sites and power sources that fit more closely with the UK's climate commitments, even if that costs them time.
Photo: Unsplash
