How AI Is Driving EMEA Data Centre Growth (Second Only to Public Cloud)
AI has moved from hype to hard demand. Savills' latest EMEA Data Centre Spotlight shows AI is now the second-biggest growth driver across the region, just behind public cloud.
International Data Corporation projects AI spending will reach US$144.6bn by 2028, a 30.3% CAGR between 2024 and 2028. The compute needed to train and deploy large language models is pushing regional infrastructure harder than planned, and it's showing up in capacity, pricing and timelines.
Where demand is actually landing
Cameron Bell, Director, EMEA Data Centre Advisory at Savills, puts it plainly: "Geographically, we have seen demand remain highly concentrated. Despite discussion around location agnostic strategies when it comes to AI, very few such projects have translated into transactions."
Operators are doubling down on availability zones with proven scale: hyperscale footprints, dependable energy and space to grow. The bulk of new requirements still clusters in FLAP-D (Frankfurt, London, Amsterdam, Paris and Dublin), supported by dense populations and large enterprise tenants. Expansion is happening beyond the big five-but it's still a smaller slice.
Capacity growth: FLAP-D leads, new markets surge
Live capacity across EMEA is up 12% in the past 12 months. France grew 15%, Germany 10%, and the UK and Ireland 9%. The Netherlands rose just 6%, reflecting the ongoing government moratorium on new builds.
Non-traditional markets are accelerating from a smaller base: Portugal (+60%), Saudi Arabia (+49%), Spain (+25%), UAE (+20%) and Sweden (+17%). Even with these gains, delivery is lagging demand-creating a tighter market.
Supply delivered vs. take-up
Since the start of 2025, 850MW of power capacity has been delivered across EMEA, down 11% year over year. New take-up in 2025 sits at 845MW-around half of 2024's total leased capacity.
That slowdown isn't about weaker demand. Savills attributes it to limited new facilities coming online, not a lack of buyers.
Pre-lets, occupancy and pricing pressure
Total contracted power capacity is now close to 14,500MW, up 12% year over year. About a quarter of take-up is pre-let, compared with less than 20% three years ago. Occupancy reached 91% in Q3 2025, up from 87% in Q3 2022.
Bell adds: "The persistent imbalance between surging demand and restricted supply continues to underpin rental values. Following three years of sharp increases, average rents have stabilised across the region. Nonetheless, with accelerating AI related requirements, rising energy costs and sustained construction inflation, further upward pressure on pricing is widely anticipated for the rest of 2025 and beyond."
Capital flows: private equity stays active
Investor interest hasn't cooled. Since 2021, 80-90% of the value of closed deals has been backed by private equity, real estate funds or infrastructure investors, up from 50% in 2020.
Fundraising for data centre vehicles by the end of Q2 2024 is almost 40% higher than the total raised in 2023. The sector's link to the digital economy and ongoing capital needs keeps money moving, even with power constraints.
Build costs keep climbing
Across EMEA, build costs now range from US$7.3m to US$13.3m per MW of commissioned IT load. That covers land, shell, electrical and mechanical systems, cooling, fire safety and fit-out.
The sharpest annual increases: Vienna (+27.5%), Warsaw (+25.4%), Stockholm (+18%) and Copenhagen (+17%). A Turner & Townsend survey shows most respondents saw 5-15% cost increases in the past year, with 22% reporting even higher jumps. See the Data Centre Cost Index
What builders and operators are doing about it
Marc Edmondson, Director and data centre expert in Savills' building & project consultancy team, notes: "With construction timelines lengthening, capital expenditure is also rising. Greenfield construction costs have risen sharply, reflecting labour shortages, land scarcity and persistent supply chain issues."
Globally, average costs are up more than 6% year on year; in EMEA, it's 9.7%. Developers are responding with forward purchasing, tighter supplier relationships and a pivot to markets with more accessible land and available energy.
Practical takeaways for IT, Dev and Ops
- Prioritise locality. Keep latency-sensitive AI workloads close to FLAP-D, but evaluate emerging markets where energy and land are available.
- Lock in earlier. With more pre-lets and longer lead times, reserve capacity sooner-especially for GPU-heavy deployments and high-density cooling.
- Budget for inflation. Model 5-15% build cost increases and stress test for higher energy prices. Expect timelines to slip and price that risk.
- Engineer for constraints. Right-size models, use efficient inference endpoints, and schedule training during off-peak windows to smooth load and cost.
- Strengthen procurement. Forward-order critical equipment and build redundancy into your supplier base.
- Upskill your team. Align infra, data and engineering skills with AI deployment realities. Explore focused options by role at Complete AI Training.
The bottom line
AI demand is surging. Supply is tight. Costs are rising. The teams that plan early, commit to the right regions and design for constraints will ship on time while competitors wait for capacity that never shows up.
Your membership also unlocks:
