Summary
- Scientific advisers say future UK computing demand is difficult to forecast reliably.
- Siting decisions should reflect workload requirements, electricity access, connectivity, and resilience.
- The 12 November publication is expert advice, not adopted government policy.
The UK’s Government Office for Science has published expert advice on where future domestic data centres should be located, warning that uncertainty over computing demand makes it difficult to determine a single national requirement for additional capacity.
The paper, published on 12 November, draws on a roundtable held on 20 February and chaired by Government Chief Scientific Adviser Professor Dame Angela McLean.
Participants examined the factors that should influence the location of computing infrastructure as demand grows from artificial intelligence and other digital services.
The resulting document is scientific advice rather than an adopted government policy. It does not introduce new planning requirements or determine where individual facilities should be built.
The advisers identify the difficulty of forecasting domestic data centre demand as a central challenge.
Future requirements will depend on the adoption of artificial intelligence, the geographical distribution of computing workloads, decisions about data sovereignty, and the extent to which businesses and public organisations need to retain processing capacity in the UK.
Rather than attempting to calculate demand solely by economic sector, the paper recommends examining the tasks computing infrastructure must perform.
That approach allows the location of a facility to be assessed against the technical requirements of its workloads.
Latency is one consideration. Some activities can tolerate substantial distances between processing equipment and users, while others require shorter response times.
Large-scale AI model training may be less dependent on proximity to an individual user than certain inference workloads, financial applications, or systems requiring rapid responses.
The resulting location requirements are not uniform across the sector.
Sensitive data creates another constraint. Some government and regulated activities may require specific assurances about where information is stored and processed, who controls the infrastructure, and how the service can be maintained during disruption.
Domestic hosting can address certain jurisdictional and sovereignty requirements, but the physical location of equipment does not alone establish operational resilience.
The paper notes the importance of geographic separation and refers to the experience of Ukraine when examining the consequences of physical threats to digital infrastructure.
A concentration of facilities in one location may create shared exposure to electricity failures, extreme weather, or hostile action, even where individual buildings have resilient internal systems.
Electricity availability is among the most immediate physical constraints identified by the advisers.
Large computing developments require substantial and dependable power connections. A suitable location must therefore be assessed against existing grid infrastructure, potential reinforcement, connection costs, and the timing of supply.
The paper also discusses co-location with low-carbon generation and the role of economic incentives in directing development towards suitable sites.
Those arrangements involve trade-offs. A site with favourable electricity conditions may be further from users or telecommunications networks, while a location offering excellent connectivity may have limited network capacity or high land costs.
Water use is considered alongside cooling design.
The advisers suggest that the UK’s relatively cool climate offers opportunities to limit direct cooling-water consumption through appropriate equipment selection.
Lower water demand can, however, involve higher electricity consumption, depending on how heat is removed from the facility.
A particular cooling technology cannot therefore be assessed solely by its water requirements without considering electricity use, performance, and operating conditions.
Waste heat recovery introduces a further location-specific question. Recovered heat is most useful where suitable customers or heat distribution networks exist nearby.
Without that infrastructure, the presence of potentially reusable heat within a data centre does not establish a viable commercial or environmental benefit.
Planning and community confidence
The paper identifies the absence of national planning guidance specifically addressing data centre location.
It proposes an interactive map showing existing facilities alongside fibre routes, power lines, and other relevant infrastructure, providing a clearer basis for understanding potential development locations.
Public acceptance is also treated as a practical development constraint.
Communities need reliable information about local employment, infrastructure costs, environmental effects, and the benefits expected from hosting large facilities.
Claims about jobs or the protection of consumers from electricity price increases must be capable of scrutiny, rather than being treated as guaranteed outcomes of construction.
The advice arrives while England’s draft National Policy Statement for data centres is undergoing separate consultation.
That planning process will determine the formal assessment framework proposed for nationally significant applications. The scientific paper does not substitute for it.
The advisers’ recommendations instead identify the evidence needed to match computing demand with suitable infrastructure and locations.
The next policy decisions will determine whether that evidence is translated into more coordinated national siting guidance and the information available to developers, planning authorities, and communities.
