The Boom That Looks Unstoppable
The UK data center market is booming. More than GBP 25 billion has been committed since 2020, with hyperscalers, private equity and institutional investors all backing large scale projects. In 2024, data centers were designated as Critical National Infrastructure, cementing their role in Britain’s sovereign AI ambitions. The headline story is that the UK has become Europe’s leading hub for digital infrastructure. But the sharper point is that growth alone is not the full story. Beneath the surge in capital lies a potential fragility that could determine which projects succeed and which stall.
Hyperscalers Betting Big
The evidence of hyperscale commitment is striking. Blackstone is advancing a GBP 10 billion AI campus in Northumberland, with a capacity of 720MW. AWS has pledged GBP 8 billion over five years, supporting 14,000 jobs annually. Equinix is investing GBP 3.9 billion in Hertfordshire, adding 250MW of compute capacity. CloudHQ is developing a GBP 1.9 billion campus in Oxfordshire, adjacent to a 400kV substation. BlackRock has committed GBP 500 million to enterprise modernization through its Gravity Edge platform.
Technavio forecasts UK market will grow from GBP 16.6 billion in 2023 to GBP 44.5 billion by 2028, a 168% increase. London alone accounts for 80% to 90% of installed capacity, making it the dominant hub. Hyperscalers are operating at scale, and the investment narrative is compelling. Yet the concentration of capital in power secured hyperscale projects tells us something important: Investors are already filtering for resilience.
The signals we are seeing show that investors are now far more cautious. Due diligence requirements are higher, and purpose and location matter as much as scale, raising the question of whether investors would make the same decisions today compared to when they first invested. AI data centers built purely for training can be located anywhere with cheap energy, such as Scandinavia or Canada. In contrast, inference workloads for applications that drive financial services and metropolitan demand need to be located close to major UK hubs. Many U.S. banks, for example, favor sites west of London because they provide faster connections to U.S. markets than facilities in the east or north. This demonstrates how commercial strategy and location have become decisive factors in investment.
Fragility That Cannot Be Ignored
Power availability is the clearest red flag. Grid constraints in London and Slough are forcing development beyond the M25, and electricity demand is projected to rise from 5 TWh in 2025 to 22 TWh by 2030. Water and cooling resilience add another layer of risk. Large campuses in water stressed regions could face permitting scrutiny and community opposition, even though many UK facilities use waterless cooling.
Retrofitting older halls is also problematic. AI workloads already reach 80 kW per rack, far beyond traditional enterprise densities. While retrofit programs can reduce capex and embodied carbon, structural limits often make rebuilding more economical. Valuation compression is emerging as transaction volumes slow and bid ask spreads widen. Prime, power secured UK campuses remain resilient, but speculative projects face pressure. Leverage exposure is evident, with debt available mainly for de risked, fully leased assets, leaving development stage projects vulnerable to refinancing shocks. Finally, technological obsolescence looms large, as AI hardware evolves faster than building lifecycles, rendering air cooled assets obsolete without liquid cooling and structural redesigns.
Even when projects overcome energy and grid hurdles, wider commercial and strategic issues are emerging. Facilities built for the wrong purpose or in the wrong location risk becoming structurally unprofitable or redundant. This marks a shift from purely technical fragility to broader market fragility.
Policy That Enables but Does Not Deliver
UK government policy has enabled growth through designation, planning reform and regulatory support rather than direct infrastructure investment. Capacity is expected to rise from 1.6 GW in 2024 to 6.3 GW by 2030, but demand is outpacing supply. Electricity consumption could quadruple by 2030, creating a widening supply–demand gap.
The lesson from Ireland’s moratorium is clear. If grid reform lags, power bottlenecks can halt development entirely. In the UK, pre-leasing capacity before construction is now common, underscoring the urgency of securing power. Policy is supportive, but delivery remains the bottleneck. That is the real constraint investors must weigh.
Stress Signals Already Visible
Evidence of stress is already visible in the UK and Europe. OpenAI’s “Stargate UK” project has been paused due to energy costs and grid constraints. Amazon Europe is facing grid delays of up to seven years. Ireland’s moratorium stalled investment until policy clarity returned. These examples show that even hyperscale players are vulnerable to infrastructure bottlenecks, resulting in liquidity pressures, stalled developments and refinancing risks.
The Sharp Point
The UK data center market is a paradox: structurally attractive, yet increasingly fragile. Hyperscale investment is surging, but only a narrowing subset of assets will remain future proof enough to capture AI driven demand. The sharp point is that growth alone is no longer the story. The real differentiator is resilience to power access, cooling design, financing stability, technological readiness and commercial viability of project’s purpose and location.
For Kroll, by focusing on early risk signals such as power availability, financing and technological obsolescence we can support investors, lenders and operators to identify fragility before it becomes distress.
