UK grid capacity to 2029: why credible demand will matter more than a queue position

The thesis: grid access is becoming a test of deliverability

For major electricity users, the commercial question is changing. Securing a place in a connection process is no longer enough. A project must show that its demand is real, its site and capital are ready, its network works can be delivered, and its operating model can accommodate the constraints of the system.

This matters for the UK now because the rapid expansion of large-load applications, particularly from data centres, has revealed the gap between reserved capacity and electricity that will actually be consumed. Ofgem reports that contracted transmission-demand offers rose from 41 GW in November 2024 to 125 GW between November 2024 and June 2025. At least 80 GW of that increase was linked to data-centre projects (Ofgem, 2026) ↗. These figures demonstrate exceptional pressure on the connection system. They do not show that 125 GW of new demand will be financed, built, energised or operate at scale.

Ofgem’s proposed data-centre reforms respond directly to that problem. They include a commitment fee and milestones designed to test whether applicants have a credible end user, plans to procure long-lead electrical equipment, and adequate financial and technical capability (Ofgem, 2026) ↗. The detailed arrangements remain subject to consultation, but the policy direction is clear: holding scarce capacity without evidence of delivery is becoming harder to justify.

**Sanctuary’s interpretation:** the competitive asset to 2029 is not a large nominal connection request. It is credible access: an integrated case connecting customer demand, land rights, finance, network engineering, a construction programme and a phased operational plan. That proposition matters for data centres, but also for electrified manufacturing, logistics, industrial estates and other major loads.

A real demand trend should not be confused with the queue

Treating all data-centre applications as speculative would be as misleading as treating the queue as a build-out forecast. NESO estimates that existing Great Britain data centres use 7.6 TWh of electricity from 2.4 GW of connected capacity. Its 2025 Future Energy Scenarios put data-centre electricity demand at 20–41 TWh in 2035, depending on the pathway, and connected demand at 6.1–14.6 GW in 2050 (NESO, 2025) ↗. These are scenarios, not commitments, but they establish that computing could become a system-scale source of demand alongside electrified transport, heat and industry.

The distinction has practical consequences. A queue position can reflect a developed investment with an identifiable customer and equipment programme. It can also represent commercial optionality: securing infrastructure ahead of a funding decision, a customer contract or a competing developer. Connection volumes therefore cannot be read as near-term consumption forecasts. Yet the scale of applications still gives planners a legitimate reason to improve the quality of demand information and to prevent weak claims from blocking viable projects.

Nor does a national demand forecast answer the question facing an individual investor. Capacity is constrained at particular transmission boundaries, grid supply points, substations, distribution feeders and peak periods. A viable investment case must separate three tests: whether the system can meet aggregate demand; whether capacity can be allocated at a specified location; and whether the necessary works can be designed, consented, procured and commissioned on the required timetable.

This is why a smaller, staged or partly flexible load can sometimes have a stronger investment case than a larger inflexible request. It may reduce exposure to reinforcement delays and allow productive activity to begin before ultimate capacity is available. That is not a universal solution: some industrial processes and digital services require continuous, high-quality supply. But it is a material option that conventional headline-megawatt comparisons can obscure.

Location and flexibility are economic choices, not technical afterthoughts

NESO’s scenarios make the locational dimension explicit. It suggests that sufficiently strong locational signals could support up to 20% of future data-centre demand locating in Scotland, helping to reduce network constraints. It also identifies cold thermal storage and potentially flexible, non-time-critical computing as ways to shift cooling demand away from peak periods (NESO, 2025) ↗.

The implication is not that every project should move north. Latency, fibre routes, customer proximity, resilience standards, skills, planning and water availability can be decisive. The more useful conclusion is that demand is not as geographically fixed as many site strategies assume. The value of a location should be assessed across power, communications, workforce and market access, rather than treating grid access as a late-stage utility issue.

Flexibility has a similar trade-off. Cold stores, charging depots, pumping and treatment loads, batteries, some industrial processes and some computing workloads have different abilities to reshape consumption. But a flexible connection is investable only if its terms are operationally intelligible: the circumstances in which demand may be reduced, the notice period, the expected duration, compensation, and accountability for disruption to customers. Otherwise, flexibility simply transfers uncertainty from the network to the end user.

Government’s strategic-demand consultation considers wider flexible-connection arrangements alongside stronger entry and progression requirements for demand projects (DESNZ, 2026) ↗. **Sanctuary’s judgement** is that the value of those reforms will depend on whether they create products with usable commercial terms, rather than merely more conditional offers. Where they do, flexibility could become both a route to earlier capacity and an operating capability that reduces energy and resilience risk.

Better allocation is not the same as faster physical delivery

The emerging policy architecture is intended to make connection allocation more disciplined. Ofgem’s 2025 decision on connection reform introduced “ready” and “needed” criteria for transmission-generation connections (Ofgem, 2025) ↗. Government has subsequently consulted on applying stronger requirements to strategic demand, including capacity reservation or reallocation, potential self-build of high-voltage infrastructure and flexible connections (DESNZ, 2026) ↗. NESO’s Strategic Spatial Energy Planning work is intended to provide a longer-term view of locations, technologies and infrastructure pathways to inform later network planning (NESO, 2026) ↗.

Together, these measures address a coordination failure: networks need confidence that demand will materialise before committing capital, while credible projects need confidence in network access before taking irreversible investment decisions. Queue discipline can improve the information available to planners and release capacity held by projects that do not progress.

It cannot, however, eliminate the physical work. Design, land rights, consents, equipment procurement, construction and commissioning all retain their own schedules and risks. A revised connection date is valuable only if the associated works, costs and responsibilities are sufficiently defined for a board or lender to rely on it.

There is also a legitimate counterargument to strict commitment tests. Requirements for customer evidence, equipment orders and financial capability may advantage established or well-capitalised applicants, including those best able to absorb deposits. That may be defensible where public infrastructure is scarce, but it can exclude earlier-stage innovators whose projects are credible but phased. Effective reform should therefore be transparent and proportionate, recognising genuine staged development without rewarding speculative scale.

The likely 2026–29 outcome is improved sorting, not the disappearance of constraint. Well-evidenced projects should gain clearer routes through the process, while local delivery will remain uneven.

Industrial power-cost relief improves economics, not connection certainty

Electricity-intensive industry faces a related but distinct issue. The UK Steel Strategy states that the Network Charging Compensation Scheme began in April 2026 and can provide eligible firms with up to 90% relief on electricity network charges. Government analysis expects this to reduce large industrial users’ costs by a further £7–£10/MWh, alongside existing measures worth around £24–£31/MWh on average for eligible businesses. For supported steel producers, the strategy cites industrial power prices of £86/MWh after the relief package, compared with £168/MWh before it (Department for Business and Trade, 2026) ↗.

This can materially improve the case for eligible industrial production and electrification. It is not a general forecast of business electricity prices, nor a substitute for connection capacity. Eligibility is sector- and scheme-specific, while project economics still depend on wholesale costs, utilisation, financing, power quality and the timing of network works.

Businesses should keep two investment questions separate: can the operation afford electricity at the required utilisation rate, and can the site secure and operate the required capacity on its investment timetable? A positive answer to the first cannot compensate for an uncertain answer to the second.

For local economic strategy, the same discipline applies. Major loads can support construction activity, supply chains, skilled work and investment in local services, but only once they become operating businesses. Combined authorities and councils should test headline job and investment claims against load phasing, site readiness, network works, skills demand and procurement opportunities for local firms. SMEs may not compete directly for strategic-demand capacity, but they can be affected when local constraints delay depot charging, workshop expansion or the electrification of commercial premises. Cluster-level planning makes those dependencies visible earlier.

What a credible project and place proposition now requires

For developers and other large loads, the immediate priority is an evidence pack that can withstand connection scrutiny: a credible end user or demand case, site control, an electrical design, capital readiness, plans for long-lead equipment, phased energisation and a candid account of operational flexibility. Under Ofgem’s proposed approach, these are increasingly relevant not only to lending diligence but to retaining access to scarce capacity (Ofgem, 2026) ↗.

Manufacturers, industrial landlords and logistics operators should begin estate strategy with the load profile rather than a maximum theoretical connection size. An initial connection sized to early operations, backed by expansion options, storage or demand management, may be more bankable than an oversized request with little near-term utilisation. Firms should verify eligibility before incorporating industrial electricity-cost relief into forecasts.

Places should offer more than land and planning support. A credible local proposition includes an honest account of network constraints and connection routes, as well as fibre, transport, construction access, water, resilience, skills and local economic value. NESO’s spatial-planning work creates an opportunity to align those elements earlier with national infrastructure choices, though it does not remove the need for detailed local delivery planning (NESO, 2026) ↗.

**Sanctuary’s recommendation:** make grid access a board-level sequencing issue from the first site-search decision. Assess demand shape, point of connection, flexibility, resilience, construction programme, supplier readiness and local value as one investment problem. Reform can allocate capacity more intelligently. Only delivery of the physical connection converts that allocation into productive investment.