London nonetheless hosts the largest focus of UK knowledge centre capability, however the centre of gravity is beginning to transfer. AI workloads are altering the infrastructure maths, pushing energy, house, and planning issues up the choice listing. That’s precisely the place regional areas begin to seem like the smart choice.
Authorities knowledge exhibits how concentrated the market stays: as of Autumn 2024, London is estimated at 1,048MW of colocation it load. Evaluate that with 44MW within the east of England, 17MW within the northeast and 30MW in Scotland. The hole is big, but it isn’t a everlasting benefit.
Many AI use circumstances don’t want to sit down inside London. They want predictable efficiency, safe internet hosting, robust connectivity, and a viable path to scaling. That shifts the placement query away from postcode gravity and in direction of power entry, latency necessities and construct timelines.
Reuters, citing Barbour ABI, reported UK knowledge centre spending projected to rise to £10 billion a 12 months by 2029, up from £1.75 billion in 2024, pushed largely by AI functionality wants, with near 100 new initiatives in progress. Even when the precise mixture of initiatives shifts, the course is evident: extra websites, extra energy, extra urgency.
Right here, Mark Lewis, Chief Advertising and marketing Officer at Pulsant, presents his insights.
Why London’s dominance is beginning to fray
London’s conventional edge has been proximity to finance, dense connectivity, and established ecosystems in locations just like the M4 hall. The constraint is not demand. The constraint is supply.
Nationwide planning and coverage our bodies have began to handle knowledge centres as an infrastructure class with actual trade-offs. A UK parliament briefing notes TechUK’s view that west London is “beginning to reach saturation point”, citing land and grid capability constraints. London’s personal establishments have been blunt about electrical energy demand pressures too, linking knowledge centre progress to wider constraints within the capital.
AI workloads intensify the difficulty as a result of they alter the form of demand. Coaching, inference and mannequin fine-tuning pull giant, sustained hundreds. That makes grid entry, connection queues, and substation availability the gating elements. The websites that win are usually those that may safe energy with fewer unknowns.
Lewis feedback: “A lot of organisations still default to London in early planning, then run into delivery friction. AI has made the power question impossible to defer. The smart move is to start with the workload, the latency tolerance, and the power profile, then choose the geography that can deliver on those constraints.”
AI progress zones and why they matter exterior London
The UK authorities’s AI progress zones programme is successfully a sign that regional build-out is now a part of nationwide industrial coverage. The acknowledged intention is to unlock funding in AI-enabled knowledge centres and assist infrastructure by bettering entry to energy and offering planning assist. That framing issues as a result of it strikes the dialog from “regional capacity is nice to have” to “regional capacity is a planned route to delivery”.
The coverage element is explicitly tied to energy system effectivity. The federal government’s “Delivering AI growth zones” paper units out potential electrical energy price reductions linked to location, giving a north east instance of as much as £14/MWh for a 500MW knowledge centre. That’s not a advertising line, it’s a system-level argument: place giant new hundreds nearer to the place era and community capability make the entire system cheaper to function.
Lewis notes: “Growth zones make one point unmissable. Power and planning are now first-order design constraints for ai infrastructure. Regional sites can move faster on both, and that changes the investment case. New sites may be years away, but there’s plenty of regional capacity available right now. Existing regional operations are a vital part of the mix.”
The “regional edge” angle is changing into the smarter play
For inference, retrieval, and real-time decisioning, distance begins to matter once more. Latency budgets might be tight, particularly the place AI is embedded into customer-facing companies, industrial management, fraud detection, or near-real-time analytics, which will increase the worth of regional knowledge centres and edge-adjacent capability.
This isn’t a name to desert London, however it’s a recognition {that a} single-centre technique seems brittle as energy turns into contested and AI load grows. Hybrid architectures are extra frequent now: core platforms in established hubs, paired with regional capability that retains knowledge nearer to customers, websites, or operational techniques.
With UK regional knowledge centres and high-performance connectivity choices, Pulsant can assist architectures the place workloads sit nearer to the place they’re generated or consumed, plus present resilient interconnect paths into public Cloud and companion ecosystems.
Lewis provides: “A lot of AI projects stall because the infrastructure plan arrives too late. If you start with regional capacity in mind, you can place data, compute and connectivity in a way that keeps latency predictable, reduces exposure to one grid area and gives you options as your workload grows.”
Grid reform provides urgency to location selections
Vitality regulation is shifting in parallel. Ofgem launched a session in February 2026 on overhauling grid connection guidelines, with knowledge centre progress as a focus. That issues as a result of connection queues and allocation standards can reset the aggressive image. Tasks that look viable on paper can slip if grid entry can’t be secured on the required timeline.
Future-proofing AI infrastructure means making selections that survive actual constraints: energy entry, construct time, resilience, latency, operational management, and regulatory strain. Regional capability will increase optionality. It provides organisations a approach to scale with out betting every little thing on one congested zone. The following wave of funding seems extra distributed, pushed by ai load and a coverage setting that’s explicitly attempting to tug capability into areas.
Lewis concludes: “The organisations that do best with AI infrastructure planning are the ones that treat geography as part of the architecture. Regional data centres are not a fallback. For many workloads, they are the route to predictable delivery and a cleaner scaling path.”
