The report, which analyses construction benchmarks across 52 markets and surveys more than 280 industry experts, identifies 2025 as a tipping point in the shift from traditional air-cooled facilities to high-density, liquid-cooled data centres engineered for AI workloads.
However, the industry’s rapid pivot towards more advanced facilities is outpacing infrastructure readiness, sparking concern among operators and developers. Turner & Townsend warns that without urgent adaptation, these pressures risk delaying delivery and inflating costs across one of the world’s fastest-growing digital infrastructure segments.
Cooling tech bottleneck and power squeeze
A major finding is that 83% of surveyed experts do not believe supply chains are equipped to deliver the advanced cooling systems required for AI-ready centres. These newer facilities typically rely on liquid-cooling to support power-dense AI clusters, placing new demands on both equipment manufacturers and skilled labour.
Power constraints also continue to dominate boardroom anxiety. Nearly half (48%) of respondents cite power availability as the primary obstacle to on-time project delivery, amid ongoing shortages in grid capacity and long lead times for new connections.
Turner & Townsend notes that power-hungry AI facilities are intensifying competition for energy access, across both urban clusters and emerging hub markets, as utilities struggle to respond to soaring digital demand alongside wider electrification efforts.
Rising costs and widening premium for AI builds
While cost inflation remains a factor, Turner & Townsend argues that capacity rather than price is now the strongest headwind to delivery. Traditional data centre construction costs are expected to rise 5.5% globally in 2025, driven by robust demand and labour constraints.
For the first time, the consultancy has quantified a measurable cost premium for AI data centre construction in the United States, identifying a 7–10% uplift for like-for-like capacity due to more complex cooling, power, and engineering requirements.
The report urges investors and operators to rethink procurement strategies, decouple risk exposure, and work more proactively with suppliers to accelerate industrial readiness for AI technologies. It also calls for innovation in energy-efficient design and stronger engagement with regulators and utilities to mitigate power-connection delays.
Tokyo, Singapore and Zurich top global cost league
Turner & Townsend’s index also charts notable disparities across global markets. Tokyo remains the world’s most expensive data centre construction market at US$15.2 per watt, followed by Singapore ($14.5) and Zurich ($14.2).
Established hubs continue to dominate the upper tier, with Silicon Valley at $13.3, London at $12.0 and Frankfurt at $11.6 per watt. Persistent capacity constraints, high construction costs and surging AI demand underpin the pricing.
Industry must adapt at speed
Paul Barry, data centres sector lead, North America at Turner & Townsend said: “Data centres are increasingly at the forefront of many governments’ long-term policy ambitions and their significance is better understood and recognised, providing greater opportunities for clients in the sector.
“Yet our report highlights key challenges that must be addressed to avoid putting a brake on investment and the benefits of AI transformation.”
He notes that power availability remains a critical barrier, with long-lead times for grid connection the main constraint. There is also “stronger competition than ever before” for power due to both increased business and consumer demand placing added pressure on grids.
“Developers and operators must adapt quickly to the evolving market landscape. AI data centres are more advanced, larger, and by extension, costlier. They come with greater power demands and modern cooling solutions.
“Clients need to navigate the power conundrum with greater openness to off-grid design solutions, while also securing reliable supply chains capable of providing the technology and talent needed for this new wave of data centres.”
RELATED STORIES
AWS, OpenAI sign $38bn cloud computing deal
China tightens custom checks on Nvidia AI chips amid growing tech tensions
Datacloud Energy 2026
After a standout 2025 edition, we’re back with an even sharper focus on the intersection of data centres, energy, and ESG. As power demand rises and regulations evolve, there’s a growing urgency to rethink how infrastructure is powered, financed, and built for long-term impact.
