Summary
- Turner & Townsend’s Global construction market intelligence 2026 covers 112 markets across 44 countries.
- The report identifies data centres and industrial logistics as the highest-demand global construction sectors.
- Labour shortages are now the primary driver of cost increases, with Europe among the regions facing acute skills pressure.
Turner & Townsend has warned that data centre construction is being pulled into a two-speed global building market, with AI-led demand accelerating while labour, contractor capacity, MEP equipment, and supply chains constrain delivery.
The firm’s Global construction market intelligence 2026 report covers 112 construction markets across 44 countries and more than 1,000 data points. It identifies data centres and industrial logistics as the sectors with the highest global construction demand, while more traditional sectors such as residential and commercial development are moving more slowly under economic and structural pressure.
The report says labour availability is now the primary driver of construction cost increases, with more than 70% of markets reporting skills gaps and limited surplus capacity. It also flags pressure across mechanical, electrical, and plumbing components, where demand continues to outpace supply.
In the UK, Construction Management’s summary of the report says data centres have become the second most in-demand construction sector, behind defence and ahead of industrial logistics. London ranks as the fifth most expensive construction market in the world at $6,032 per sq m, while Birmingham and Glasgow are experiencing significant cost increases.
The same trades are being chased
Data centre construction has always required a narrower labour base than general commercial development. Electrical specialists, commissioning engineers, controls technicians, mechanical installers, fire-suppression specialists, and clean-build contractors all sit on the critical path. AI infrastructure raises the pressure because projects are larger, denser, and more technically demanding.
That makes labour availability a data centre capacity issue. A campus cannot be accelerated simply by adding more general labour to site. The constrained roles often require training, certification, site experience, and familiarity with mission-critical standards. When those people are already committed to grid upgrades, industrial decarbonisation, defence work, battery projects, and energy infrastructure, data centre programmes have to compete harder for them.
Europe faces a particular squeeze because several infrastructure cycles are arriving at once. Data centre operators want more capacity for AI and cloud. Network operators are reinforcing grids. Governments want renewables, transmission, batteries, and industrial electrification delivered faster. Each programme pulls on overlapping electrical and mechanical skills.
MEP pressure adds another layer. Data centres consume large volumes of switchgear, transformers, UPS systems, generators, chillers, pumps, pipework, containment, controls, and fire systems. As AI shifts facilities towards higher densities, the MEP package becomes harder to design, install, and commission. Cooling architecture, power distribution, and integrated systems testing all require experienced teams.
Cost inflation changes shape
Turner & Townsend says global construction cost inflation is expected to rise modestly from 4.2% in 2025 to 4.5% in 2026 before remaining broadly flat in 2027. The figures suggest the most intense post-pandemic materials shock has eased, but delivery has not become simple. Cost pressure is shifting from general inflation towards specific constraints: labour availability, local contractor capacity, energy price volatility, MEP components, and risk pricing.
That changes procurement strategy for data centre developers. Early contractor engagement, realistic risk allocation, equipment reservation, modularisation, repeatable design, and disciplined scope control become more valuable when the market is tight. A project that enters procurement late, with uncertain connection milestones or unresolved design risk, is likely to pay a premium or lose capacity to better-prepared schemes.
The labour question also affects geography. Secondary markets may offer land and power advantages, but they do not automatically provide a deep pool of specialist data centre labour. Moving away from established hubs can reduce one bottleneck while creating another around workforce availability, accommodation, logistics, and contractor mobilisation.
That is especially relevant for AI campuses proposed near power-rich rural or industrial sites. The electrical connection may be stronger, but the construction labour market may be thinner. If multiple large projects arrive at once — data centres, substations, battery storage, grid upgrades, renewables, and industrial plants — local delivery capacity can become the limiting factor.
The report leaves a harder version of the AI build-out story. Demand is strong, but demand does not install switchgear, commission cooling loops, pour concrete, or validate failover sequences. Capacity growth will depend on whether the sector can turn capital into coordinated delivery while the same people and equipment are being chased by the rest of the infrastructure economy.
