Summary
- CITP estimates the UK had 348 live data centre sites and about 2.0GW of operational electrical capacity by the end of 2025.
- The analysis identifies 0.87GW under construction and 10.88GW announced or permitted, implying a possible 5.8-fold capacity increase.
- Its recommendations include environmental assessments above 10MW, disclosure, hourly matched renewable procurement, heat recovery, and a national strategy.
Critical Infrastructure Technology Policy has mapped the UK’s data centre build-out, estimating that the country hosted 348 live sites and about 2.0GW of operational electrical capacity by the end of 2025.
The analysis identifies a further 0.87GW under construction and 10.88GW in the announced or permitted pipeline. If that whole pipeline were commissioned, CITP says live UK capacity would grow by a factor of about 5.8 within a few years. The work uses the Baxtel database and treats data centres as infrastructure assets with consequences for energy policy, industrial strategy, foreign direct investment, and digital sovereignty.
The geography is heavily concentrated. CITP says England hosts 313 of the 348 live sites, or about 90% of the UK stock, while Scotland has 14, Wales 12, and Northern Ireland nine. London alone accounts for 218 sites, nearly two-thirds of the national total.
The ownership picture is also more concentrated than the raw site count suggests. CITP says there are 117 distinct operators, but the top ten groups account for around three-fifths of operational megawatt capacity. Carrier-neutral colocation dominates site numbers, while only 16 sites are classed as hyperscale.
The cloud estate becomes infrastructure policy
The report treats data centres as a physical infrastructure market rather than as an invisible backend to cloud services. Once capacity is measured in gigawatts, the policy questions move beyond digital competitiveness and into energy allocation, planning, emissions, heat, ownership, tax, and local consent.
The scale of the pipeline is the sharpest figure. A move from 2.0GW of operational capacity to a much larger stock would place new pressure on grid connections, substations, transmission reinforcement, distribution networks, construction labour, cooling design, and equipment supply. Not all announced projects will be built, but the scale of the pipeline shows how much speculative and strategic activity is now gathering around UK data centre land and power.
CITP argues that the sector is capital intensive and that scale, grid access, and engineering capability create durable advantage. A market can look fragmented by site count while being concentrated by effective capacity. If the largest platforms and hyperscale owner-operators control the bulk of megawatts, policymakers will need to look at data centres as concentrated infrastructure rather than a loose collection of server rooms.
The London and south-east concentration remains a constraint. Data centres cluster around interconnection nodes, financial services demand, cloud regions, and enterprise customers, but those same locations face grid and land pressure. Government efforts to push AI growth zones and digital infrastructure beyond established hubs will have to deal with latency requirements, fibre topology, customer demand, labour availability, and energy delivery.
Disclosure, heat, and hourly power
CITP’s policy recommendations are unusually specific. It calls for mandatory environmental impact assessments for data centres above 10MW, environmental disclosure statements, public support linked to 24/7 hourly matched renewable procurement, mandatory waste-heat recovery, and a National Data Centre Strategy.
Those proposals go beyond a generic call for greener data centres. Hourly matched renewable procurement is much stricter than annual renewable energy matching. It addresses the difference between buying enough renewable energy certificates over a year and matching demand with clean supply in the hours and locations where electricity is consumed.
Mandatory heat recovery is more difficult. Data centres produce large amounts of rejected heat, but useful reuse depends on temperature, distance, district heating infrastructure, anchor customers, commercial agreements, and planning. Requiring every project to consider heat reuse is one thing; requiring effective heat export is harder. Even so, putting heat into planning policy would force developers to account for the thermal side of digital infrastructure earlier.
The proposed 10MW threshold for environmental impact assessment would pull many modern data centre projects into a more formal review process. That may slow some schemes, but it would also give planners, network operators, and communities a clearer basis for comparing claims on jobs, power, water, cooling, emissions, and local benefit. As projects grow larger, the old distinction between digital development and industrial infrastructure weakens.
CITP’s analysis also brings digital sovereignty into the infrastructure layer. Sovereignty is often discussed as data location, cloud provider control, or software capability. The report points to a harder foundation: who owns, powers, and operates the facilities that host the compute. If the UK relies heavily on foreign-owned platforms while offering land, grid capacity, incentives, or relaxed environmental standards, the public value case becomes more contested.
The report is best read as a policy intervention rather than an official market forecast. Some recommendations would be resisted by developers, and some pipeline capacity may never materialise. Even so, the underlying pressure is clear enough: the UK’s data centre build-out is now large enough to require national infrastructure policy, not just local planning decisions and investor appetite.
