Contracted offers in the UK electricity demand connection queue rose from 41 GW in November 2024 to 125 GW by June 2025, against Great Britain’s peak electricity demand of 45 GW on 11 February 2026.
Approximately 50 GW of that queue is attributable to data centre projects, making them the single largest driver of the connection backlog. Some developers are being offered connection dates in 2037 and beyond. In that context, a study published in Energy and Climate Change by researchers from WU Vienna University of Economics and Business proposes reframing how data centres are assessed and connected, arguing that on-site batteries could reduce peak electricity imports from the grid by between 10% and 15%, and that regulatory rules should be redesigned to treat data centres as potential grid assets rather than purely sources of additional demand.
The study analysed energy consumption at 96 UK data centres using data from UK Power Networks. Its core finding is that on-site battery storage could supply a meaningful share of power during periods of highest demand, reducing the size of grid connection required and easing pressure on distribution networks that are already operating close to capacity in key locations. Parts of London’s grid reached full capacity in 2022, temporarily halting new housing connections in Hillingdon, Hounslow, and Ealing. The same network constraints that are delaying housing are delaying data centre connections, and the competition between these categories of demand for the same scarce grid headroom has become a structural conflict requiring a policy response.
The Connection Constraint as the Binding Limit
The scale of the problem dwarfs what battery storage alone can resolve, but the WU Vienna research addresses a genuine bottleneck. Global data centre electricity consumption reached approximately 415 TWh in 2024, roughly 1.5% of world electricity consumption, and the IEA projects this will more than double to 945 TWh by 2030. In the US, data centres could consume as much as 17% of electricity by the end of the decade, 60% higher than projections made in 2024. In the UK, NESO identified approximately 140 data centres representing 50 GW of demand in the connection queue, of which only 71, representing around 20 GW, had reached Final Investment Decision. The remaining 30 GW consists largely of speculative or early-stage applications occupying queue positions without the capital commitments or planning permissions that characterise viable projects.
Ofgem’s demand connections reform consultation, launched in February 2026, identified three interrelated challenges: a queue that is large and growing and contains many non-viable projects; delays for well-progressed schemes caused by both network build times and speculative applications blocking queue capacity; and the absence of mechanisms to prioritise strategically important developments. Phase one of the proposed reforms targets the data centre market specifically, with stricter financial and readiness tests designed to filter out speculative positions. New grid assets take five to ten years to build, whereas a data centre can be constructed in 18 to 24 months, a mismatch that means the grid is structurally incapable of keeping pace with AI-driven demand growth under the current connection architecture.
What On-Site Storage Changes in the Calculation
A 10 to 15% reduction in peak grid imports, the range the WU Vienna study identifies from batteries at 96 UK data centres, is meaningful in this context because peak demand determines connection capacity requirements. A data centre that imports 100 MW at peak but can shave that to 85 to 90 MW through battery discharge during high-demand intervals requires a smaller grid connection, moves up the viability queue relative to a connection sized for its full load, and reduces the infrastructure cost that the network operator must build to serve it. The cumulative effect across many facilities could reduce the total capacity that the queue is asking the grid to provide, addressing the structural demand-supply mismatch at the source rather than requiring proportionate grid expansion.
The study also identifies additional battery value beyond peak shaving. On-site generation from solar or wind co-located at data centre sites, or nearby through power purchase agreements, reduces grid reliance further. Battery storage at data centres can provide frequency response, voltage support, and other ancillary services to the wider electricity system, generating revenue that improves project economics and reduces the net connection capacity the facility needs to draw from the shared network. The research argues that regulatory barriers preventing data centre batteries and on-site generators from supplying grid services should be removed, and that computing demand itself could become more flexible, with operators rewarded for shifting workloads to periods of abundant clean power or reducing consumption when networks are under stress.
The Policy Framing and Its Implications
The WU Vienna paper’s most consequential recommendation is that grid connection rules should be redesigned to prioritise projects that deliver measurable benefits to the wider energy system, rather than assessing data centres purely on the volume of demand they place on the grid. Under current rules, a data centre that commits to significant battery storage, flexible demand response, and co-located clean generation gains no priority advantage over one that simply requests a large firm connection. The research argues this is the wrong incentive structure at the wrong moment: the electricity system needs flexibility as much as it needs generation, and data centres that can supply flexibility should be valued accordingly.
Ofgem is exploring non-firm, ramped, or self-supply connections as alternatives to conventional firm demand connections for data centres. A ramped connection allows a developer to connect at a lower initial capacity and step up as the network is reinforced, aligning the build-out of grid infrastructure with actual data centre commissioning phases rather than requiring full capacity upfront. A non-firm connection provides power at lower priority with the possibility of curtailment during network stress events, which for data centres with on-site batteries and flexible workloads is a more viable commercial proposition than it would be for a facility entirely dependent on grid supply.
The Planning and Infrastructure Act 2025, which received Royal Assent in December 2025, enables large-scale data centres to be classified as Nationally Significant Infrastructure Projects, allowing developers to bypass local planning and seek development consent directly from the Secretary of State. This streamlines the planning dimension of the connection problem. Whether the Ofgem connection reforms produce a queue that reflects genuine investment readiness and system benefit, rather than speculative positioning, will determine whether the WU Vienna battery integration model can translate from research findings to standard practice across the 50 GW of data centre demand currently waiting for grid access.
