Occupier market: three engines, one choke-point

The UK data centre market is no longer best explained by a single technology cycle. Demand is being propelled by three overlapping engines: cloud-led scale, AI-driven power intensity, and a persistent enterprise and colocation undercurrent. All are converging on a limited pool of capacity that can be delivered in the right places and within occupiers’ timeframes. That tension is most acute in London, where cloud ecosystems, network density and established availability zones continue to anchor site selection and, in doing so, make infrastructure constraints the practical limit on choice.

Recent metrics illustrate the consequences. Annual take-up has moved onto a higher runway since 2020, reaching 217MW in 2025, a record high. The subdued 19MW recorded in Q1 2026 is better read as a symptom of limited near-term options than a reversal in requirement depth. Vacancy has compressed sharply, from 29% in 2006 to 8% in Q1 2026, with London tighter at 7%. Commitment timing has therefore shifted forward: in 2024, pre-sold capacity from future projects totalled 257MW, alongside 212MW of take-up and against deliveries of 190MW, reinforcing that a significant share of demand is being secured well before assets become operational.

Cloud-led demand remains the principal volume driver, but procurement has become more fluid. Hyperscalers still set the marginal price for well-located, power-secure campuses, yet capacity is increasingly accessed through wholesale platforms and self-build. Wholesale colocation accounts for 1,163MW of live UK capacity, and rises to 2,729MW when live, under-construction and committed schemes are combined.

AI-related demand is changing the character of requirements as much as their scale. Compared with traditional cloud deployments, AI places disproportionate weight on power density, cooling specification and energy economics, and it tends to arrive in larger, more time-sensitive blocks. Market feedback suggests early European momentum was interrupted while design standards and investment priorities were recalibrated, but confidence returned later in 2025 as major occupiers re-entered the market. Reflecting that shift, AI now appears as a discrete category in the UK development landscape, with 50MW recorded within self-build AI capacity in the combined live-plus-pipeline view.

Enterprise and broader colocation demand provide the market’s constant, and it is increasingly expressed through hybrid strategies that blend public cloud with third-party facilities for control, resilience and compliance. Retail colocation remains material, with 347MW of live capacity, but tighter supply conditions are encouraging larger suites and longer commitments that are beginning to resemble wholesale procurement. For investors, this convergence blurs traditional segmentation and increases the premium placed on platforms that can flex across tenant types without relying on a single anchor to underwrite the business case.

London’s gravity, supply reality

London’s dominance is not a legacy artefact; it has strengthened as the market has scaled. UK live capacity has reached 1,803MW in 2026, and London accounts for 1,637MW, around 91% of the national stock, up from 79% in 2006. This concentration reflects the continued pull of the capital’s connectivity and cloud ecosystems, but it also means that UK-wide capacity expansion does not automatically translate into choice for occupiers. Outside the London orbit, the cloud market remains comparatively shallow and less able to absorb requirements at pace.

As a result, London operates as both the market’s demand engine and its constraint amplifier. Where power and land are scarce, timelines elongate, and occupiers face a tighter trade-off between location fidelity and delivery certainty. It is this dynamic, rather than any sudden weakening of London’s role, that is prompting a broader search for alternative geographies.

Market discussions reinforce that decentralisation is, for now, incremental. Activity remains anchored to tier-one locations around London, with only selective regional depth emerging. Manchester is the most cited secondary market, but follow-through beyond London and Manchester has been limited, reflecting how difficult it remains to replicate the capital’s combination of network density, occupier ecosystems and timely access to power.

Where regional opportunities do look credible, they are typically those that can offer what London cannot: scale, land and a clearer pathway to power at scale. This is not a redirection of cloud demand, which continues to cluster around established availability zones, but a structural addition driven by AI's fundamentally different site-selection criteria: geographically agnostic, less latency-sensitive, and able to follow power and land rather than population centres. The North East is increasingly positioned in that bracket, supported by renewable infrastructure and grid connectivity, and by a planning context that is more receptive to large-scale investment. Even so, this is best understood as a medium-term diversification story rather than an immediate release valve for the broader market.

Bigger facilities, fewer marginal locations

Alongside geographic concentration, the market’s development profile is shifting, with growth increasingly delivered through fewer, larger facilities. Average UK data centre size has risen from 2.1MW in 2006 to 6.3MW in 2026, and while sub-5MW sites still dominate (171 of 258 existing facilities), incremental capacity is being delivered through fewer, much larger projects. The direction of travel is unambiguous: there is only one operational facility above 100MW today, Vantage Data Centres’ Cardiff (CWL 1) in Newport, offering 130MW of IT power, while 28 projects of that scale sit in the wider pipeline.

This re-scaling concentrates both opportunity and risk. Larger campuses can relieve scarcity through step-changes in capacity, but they also increase exposure to programme and energisation risk, and they deepen reliance on a smaller pool of anchor occupiers. Operator concentration remains high today, with the top ten operators controlling 62% of UK live capacity. These structural dynamics explain why the pipeline appears extensive, and why deliverability, rather than headline megawatts, is the critical lens.

A large pipeline on paper, a narrow window in practice

Last year, 231MW of IT capacity was delivered in the UK, following 191MW in 2024, both record highs. That step-change has not yet repeated in 2026, with 48MW delivered in Q1. Of the 242MW currently under construction across the UK, around 66MW is scheduled to complete within the remainder of 2026 (Q2–Q4), underscoring how limited the immediate delivery window is relative to the headline pipeline.

Headline pipeline numbers can be misleading without a clear view of what is actually deliverable. The UK pipeline totals around 13,500MW, approximately 7.5 times current live stock yet the executable component is concentrated in a relatively small number of schemes. In addition to the 242MW under construction, a further 2,084MW is categorised as committed, while the balance (9,371MW) remains early stage, where grid access, planning outcomes and financing are still uncertain. London’s dominance persists in the most advanced stages, accounting for 240MW of capacity under construction and 1,652MW of committed capacity.

In a FLAPD context, this concentration underpins London’s continued leadership over the medium term: by 2030, London’s stock is expected to reach around 2,679MW, still well ahead of other FLAPD markets. That implies London grows by roughly 1.1 GW between 2025 and 2030, materially outpacing all other FLAPD cities over the same period.

Even within the under-construction cohort, availability is thin. Of the c.66MW scheduled for completion in 2026, 96% is already pre-sold, leaving less than 3MW available to transact. The forward picture is only marginally looser, across the 37MW currently programmed for 2027 delivery, just over half is already taken up. The market, therefore, does not move from tight to balanced simply because cranes are visible. Instead, pre-letting has become part of the supply mechanism, pulling future capacity forward into today’s decision set and reducing the stock of walk-up options for both hyperscale and enterprise occupiers.

Constraints and consequences: power, land and the economics of time

The UK’s supply–demand imbalance is ultimately a constraint story, and power sits at the apex. London’s growing share of national capacity reflects rational demand clustering around established cloud zones, but it also heightens exposure to grid scarcity and long connection lead times. Market conversations suggest that, in some of the most constrained West London locations, developers are now planning for multi-year waits for firm capacity, long enough to undermine traditional development sequencing and force a rethink of how projects are energised. The direction of travel is clear: delivery certainty is increasingly shaped by energy strategy as much as by real estate fundamentals.

Land constraints reinforce the power challenge. Where power and connectivity coincide, land values and competitive tension rise sharply, and the market can generate speculative behaviour, applications submitted in the hope of capturing an outsized uplift if power is secured. However, the volume of proposals should not be mistaken for deliverable supply. Practitioner feedback indicates that a significant proportion of speculative applications are unlikely to progress to development, whether due to grid allocation outcomes, planning friction, or the capital intensity required to meet modern design requirements. For developers, this places a premium on origination capability and stakeholder management; for occupiers, it increases the value of partnering early with platforms that can demonstrate credible pathways through planning, procurement and energisation rather than relying on announcements.

Construction timelines and delivery risk further shape market outcomes. The UK’s average facility size has increased, reflecting the shift towards larger, more complex facilities where mechanical and electrical fit-out drives both cost and programme. While retrofitting existing industrial shells can be technically feasible in some cases, the economics are often less compelling than occupiers expect because the critical cost sits in M&E rather than the envelope. For investors, these dynamics sharpen the distinction between ‘pipeline narrative’ and investable reality: consenting and delivering powered capacity becomes the scarce capability, while underwriting must account for the possibility that timing slippage turns today’s contracted demand into tomorrow’s renegotiation.

Taken together, these forces sustain a landlord-leaning market even after periods of heavy building. Tight vacancy, high pre-letting and volatile delivery mean that pricing and lease structures are increasingly anchored in scarcity and certainty rather than marginal cost alone. At the same time, constraints are beginning to create a selective rationale for activity beyond the traditional London core, not because the market has become location-agnostic, but because the economics of time and power increasingly reward places where those constraints are less binding.

>>  For more information, please contact our EMEA Data Centre Advisory team

_{Read the articles within Spotlight: UK Data Centre Building Capacity below.}