Insights/Article

On-site power generation for data centres in the UK: why the grid queue is accelerating the shift

On site power data centres

For most of the past decade, on-site power generation for UK data centres was treated as a contingency — a backup for grid outages, a sustainability credential, or an engineering complexity to be managed rather than embraced. The grid was reliable and accessible. The queue was manageable. There was no commercial imperative to invest in alternative power infrastructure when a grid connection could be secured in months.

That calculus has changed. The UK grid connection queue now runs to 125GW of requested demand, with average connection timelines stretching to seven years or more for new 50MW facilities. On-site generation and private wire have moved from contingency to strategy — and for some of the largest projects now being proposed, they have become the primary power model.

This article sets out the on-site power options available to UK data centre developers, the regulatory and legal framework that governs them, when the economics make sense, and what the shift means for site selection and development strategy.

From contingency to strategy: why the shift is happening now

The trigger is straightforward. The UK’s grid connection queue tripled from 41GW in November 2024 to 125GW by June 2025, with data centres accounting for more than 50GW of new requests. Ofgem’s TMO4+ reforms are beginning to clear speculative applications from the queue, but the underlying infrastructure constraint — the physical gap between available grid capacity and genuine demand — is not resolved by regulatory reform alone.

The commercial pressure is now sufficient to make on-site generation genuinely competitive. Recent modelling shows microgrids can be deployed in roughly half the time of waiting for grid connections, at 43% lower cost, while meeting 80% of annual data centre demand from renewables. When the alternative is a decade-long queue, those economics become compelling.

The clearest public signal of this shift came in June 2026, when Xlinks proposed a 1.5GW AI data centre campus in Devon connected to a 1.8GW battery energy storage facility via private wire, bypassing the grid connection queue entirely. The site was selected in part because of spare highly reliable capacity at the Alverdiscott substation — but the design is built around private wire and on-site storage, not around a grid connection as the primary power source.

On-site generation has moved from contingency to core strategy for UK data centre developers who cannot wait on the grid queue. The question is no longer whether to consider it — it is which model fits the project and how to structure it.

The main on-site generation options for UK data centres

Battery energy storage systems (BESS)

BESS is the most rapidly deployed on-site energy asset available to UK data centre developers. Lithium-ion battery systems can be installed and commissioned significantly faster than generation assets, and they serve multiple functions: smoothing intermittent renewable generation, providing grid frequency response services, and acting as a buffer against short-term supply interruption.

BESS is also increasingly a route to grid income. From April 2026, NESO’s Demand Flexibility Service expanded to include bi-directional flexibility, lowering the eligibility threshold to 0.1MW, which materially improves accessibility for data centre participation in balancing services markets. A BESS-equipped data centre can generate revenue from grid services while providing operational resilience — a combination that improves the economics of the asset significantly.

The practical constraint on BESS participation in frequency response markets is that UPS batteries are specified and warranted for resilience, not regular cycling. Repurposing them for grid services risks warranty invalidation and degradation. Purpose-built BESS assets, sized and specified for dual-use from the outset, are the right approach.

Solar PV and onshore wind

Solar and onshore wind are the most cost-effective generation technologies for UK data centres with available land. Both can be co-located with a data centre or connected via private wire from a nearby generation asset, with the private wire arrangement allowing the data centre to access renewable power without the transmission and distribution charges that apply to grid-sourced electricity.

The challenge for solar and wind is intermittency. Neither generates continuously, and data centres require firm, continuous power. BESS is required to firm renewable generation for data centre use — the combination of renewable generation and storage is the model that makes on-site power genuinely viable as a primary supply strategy.

The Xlinks Devon proposal is the most prominent UK example of this model at scale: 1.5GW of data centre demand served by 1.8GW of BESS, drawing on nearby solar and wind generation via private wire. The design is intended to be power-reliable without relying on the grid as primary supply.

Gas generation

Gas-fired generation — typically gas engines or combined-cycle gas turbines — has become the short-term power solution of choice for larger UK data centre projects that cannot wait for grid connections. Gas networks received around three times as many enquiries from data centre developers in 2025 compared to 2024, with 113 enquiries over the two-year period.

Gas offers firm, dispatchable power that can match data centre load profiles in a way that intermittent renewables cannot without substantial storage. It can be deployed faster than a grid connection in many locations, and the infrastructure — gas network connection, engine installation, switchgear — is well understood by contractors.

The sustainability trade-off is significant. The UK’s first nationally significant data centre development in Bedfordshire is planned to run on gas-fired generation until 2034, drawing criticism from environmental campaigners. Gas is most defensible as a bridge — providing firm power while renewable and storage capacity is built out — rather than as a long-term power strategy.

Hydrogen

Hydrogen as a fuel for data centre power generation is at an earlier stage of commercial deployment than BESS, solar, or gas. Fuel cell technology for data centre applications exists and is operational in some facilities, but the economics and the hydrogen supply chain in the UK are not yet at a point where hydrogen is a primary power strategy for most projects.

The longer-term picture is more interesting. Great British Energy Nuclear signed a contract with Rolls-Royce SMR in April 2026 for the UK’s first small modular reactors at Wylfa, with first power anticipated in the mid-2030s. For operators planning long-horizon campuses, the structuring work for nuclear-adjacent power offtake starts now, even if the actual power is a decade away.

The most resilient power strategy for a UK data centre in 2026 combines renewable generation, battery storage, and a gas bridge — with a private wire structure that bypasses the grid queue while retaining a grid connection for backup and export.

The private wire regulatory framework

Private wire is the mechanism that makes on-site and near-site generation viable for data centres at scale. Rather than supplying power via the public distribution or transmission network, a private wire connects a generation asset directly to the data centre load through a dedicated cable. The data centre operates largely or entirely behind the meter, avoiding the grid queue, the transmission and distribution charges, and much of the grid connection uncertainty.

The UK regulatory framework for private wire is workable but requires careful structuring. Any entity generating or supplying electricity beyond single-premises self-supply must hold a generation licence from Ofgem or qualify for an exemption. For large-scale data centre microgrids, the generation licence position depends on the scale of the generation asset and whether it supplies power to multiple customers or tenants.

For multi-tenant colocation data centres, a supply licence may also be required if electricity is supplied to tenants within the facility. Ofgem has granted private wire exemptions for specific developments on a case-by-case basis, but each arrangement requires individual assessment. Legal and regulatory advice from specialists in electricity licensing is essential before committing to a private wire structure.

The physical PPA structure that typically underpins a private wire arrangement involves a long-term power purchase agreement between the generation asset owner and the data centre. Physical PPAs deliver electrons through a licensed supplier; virtual PPAs settle financially against a wholesale reference price. For private wire, the physical PPA is the appropriate structure — it delivers actual electrons from the generation asset to the data centre through the dedicated cable.

When the economics stack up

On-site generation is not right for every UK data centre project. The economic case depends on several variables that differ by project, location, and technology choice.

Scale matters. The economics of on-site generation and private wire are most compelling at larger scales — 50MW and above — where the capital cost of the generation and storage infrastructure is spread across a sufficiently large load to produce competitive per-unit costs. Smaller facilities are generally better served by grid connection or standard colocation arrangements.

Location matters. The economics of on-site generation depend heavily on the proximity and cost of the generation asset. A data centre that can co-locate with an existing renewable generation asset, or that can secure a private wire connection from a nearby wind or solar farm, has materially better economics than one that needs to build generation from scratch.

Grid connection timing matters. The core economic argument for on-site generation is the alternative cost of waiting for a grid connection. A project that can access the grid in 18 months does not have the same incentive to invest in private wire as one that faces a decade-long queue. The economics of on-site generation improve as the grid queue lengthens.

Power cost matters. The UK is a relatively expensive power market in European terms. Bypassing transmission and distribution network charges through private wire can produce meaningful cost savings relative to grid supply — particularly for energy-intensive AI training workloads where power cost is a significant proportion of operational expenditure.

What this means for site selection

The shift toward on-site generation and private wire is changing what makes a good data centre site. Historically, site selection was driven primarily by location — proximity to fibre, population centres, cloud availability zones, and labour markets. Power was assumed.

That has changed. Developers are pursuing private wire agreements, on-site generation, and co-location arrangements with energy providers to guarantee supply. Site selection now begins with power availability and the feasibility of on-site or private wire supply, with location as a secondary filter.

The sites that benefit most from the on-site generation shift are those with proximity to existing renewable generation assets, available land for BESS and generation co-location, access to the gas network for bridging supply, and existing high-voltage infrastructure that can support a grid tie-in for backup and export.

These sites are not always in the locations that data centre development has traditionally concentrated. Devon. The Scottish Central Belt. The Humber. Former industrial land with legacy power infrastructure. The geographic spread of viable data centre sites is widening, and it is widening along the axis of power availability rather than proximity to London.

The sites that will define the next phase of UK data centre development are not necessarily the best located. They are the best powered. On-site generation is making viable locations out of sites that the grid queue would have made impossible.

Armason: power advisory and site access across the UK

Armason tracks power availability, grid connection status, and on-site generation opportunities across the UK data centre market. We advise developers and investors on power strategy — including private wire structuring, BESS integration, and PPA frameworks — and we hold active positions in sites where land and power have already been secured.

We have sites available now with land secured and power in place, including positions that are not subject to the grid connection queue. If you are evaluating on-site generation as part of your power strategy, or looking for sites where the power question has already been answered, we would like to talk.

Sources referenced in this article

How microgrids can solve the UK’s data centre grid capacity crisis — Tim Harper

Cleaner in, useful out: the maturing sustainability stack for UK data centres — Bird & Bird

Navigating UK and Ireland data centre development in 2026 — Browne Jacobson

1.5GW data center campus proposed in Devon, UK — Data Centre Dynamics

Ofgem launches sweeping review of grid connections — Computing

Insights on meeting UK data centres’ growing energy and water infrastructure needs — New Civil Engineer

Spotlight: UK Data Centre Building Capacity H1 2026 — Savills