London still dominates Britain's datacenter map, but the regions are powering up

Britain has 555 datacenter facilities providing ~1.6 GW of IT capacity, with London housing 219 facilities (about 40%) and ~1,048 MW (~two-thirds of capacity). Expansion is shifting to regions outside London due to power/planning constraints, with 15 London expansion projects and larger queues in places like South East (10 planned) and North West (5 planned). The article flags major new builds (e.g., Equinix near South Mimms, Blackstone’s planned ~720 MW near Blyth, and a potential ~1.5 GW AI-optimized campus in Devon), while noting OpenAI’s Stargate plans at Cobalt Park are on hold due to UK energy costs.

Analysis

This is less a pure AI demand story than a scarcity-of-power-and-permits story. The economic moat accrues to operators that can secure grid access, fiber, and planning approvals near demand centers; that supports premium rents and lower vacancy for established colocation platforms, especially EQIX, while commoditized greenfield projects face a widening execution gap. The second-order effect is a pipeline bottleneck: capital will increasingly chase sites with pre-existing substations and network connectivity rather than the cheapest land, which should lift the value of powered land banks and create a spread between “approved watts” and speculative capacity.

For BX, the key upside is optionality across infrastructure, not immediate fee revenue. If UK hyperscale development shifts regional and the capex burden moves to large sponsors, Blackstone can monetize both development and financing, but only if power costs normalize enough to keep projects alive; otherwise, the headline scale of planned campuses is mostly paper. The near-term loser set is not public cloud demand, but developers that underwrite aggressive utilization before securing cheap power—those projects can stall, get repriced, or be redesigned downward.

The contrarian read is that the market may be overestimating how much of this translates into near-term build activity. Energy price sensitivity in the UK is a hard constraint, so the likely path is a slower, more regional buildout over 6-18 months rather than a London-led surge, and that favors existing operators over new entrants. Watch for planning approvals, grid connection queues, and forward power costs; if UK industrial power stays elevated or approvals slip, the AI datacenter capex narrative gets pushed out rather than accelerated.

Key Decisions for Investors

• Long EQIX on pullbacks over the next 1-3 months: treat London scarcity as a margin-supporting backdrop for colocated inventory. Risk/reward is best if leasing spreads and backlog hold; thesis is falsified if UK power costs or churn rise enough to delay rent resets.
• Buy BX as a 6-18 month infrastructure optionality trade, but only on weakness after any AI buildout enthusiasm fades. This is a slower monetization story; trim if UK project announcements keep getting deferred or if financing spreads widen.
• Do not chase speculative UK data-center developers absent secured power; prefer a wait-and-see posture until planning/grid milestones are visible. The fastest way to lose money here is underwriting MW that cannot be connected on time.
• Watch list: any public update on UK power-price policy or major grid connection approvals. A meaningful decline in forward power prices or faster permitting would be the catalyst that turns this from a scarcity trade into a broader UK datacenter re-rating.