Artificial intelligence is transforming data centres from passive infrastructure into strategic assets that determine economic competitiveness and digital sovereignty. As power constraints tighten across Europe, site selection has become one of the most consequential decisions hyperscale and AI operators will make for the next generation of compute.
For much of the cloud era, data centre location followed a relatively predictable logic. Connectivity, land availability and proximity to established digital hubs shaped expansion decisions, while power was assumed to be accessible with sufficient planning and investment. Artificial intelligence has disrupted that equilibrium. The explosive growth of high-density compute is forcing operators to reconsider not only where infrastructure is built, but how geography itself shapes the economics and performance of intelligence.
Across Europe, demand for capacity is accelerating faster than the infrastructure designed to support it. Vacancy rates continue to tighten even as record levels of new supply enter the market, reflecting a structural shift rather than a temporary surge. AI workloads, particularly training and inference clusters operating at unprecedented power densities, are reshaping how operators evaluate risk, resilience and long-term competitiveness. Decisions that once balanced cost and connectivity now determine whether a facility can operate at all.
Rob Cooper, Senior Director at CBRE, argues that this transition marks a fundamental change in how site selection should be understood. “Data centre site selection is no longer simply about identifying land with available power,” he says. “The decision you make today directly influences operational resilience, regulatory exposure and economic performance for decades. AI has extended the planning horizon dramatically because infrastructure must now anticipate future density and demand, not just current requirements.”
The consequence is that site selection has moved from a real estate exercise to a strategic infrastructure decision, one increasingly shaped by energy systems, regulatory policy and long-term market dynamics.
Power becomes the primary constraint
Power availability has emerged as the defining factor in European data centre expansion. AI training environments and high-performance computing deployments routinely exceed 50 kilowatts per rack, pushing campus requirements into the hundreds of megawatts and placing unprecedented strain on transmission networks already under pressure from electrification and renewable integration.
“Power is now the gating factor for growth,” Cooper explains. “Across Europe we are seeing grid congestion and extended connection timelines becoming the main limiting factor on new capacity. Operators are discovering that securing land is often easier than securing electricity at the scale AI requires.”
The challenge is compounded by timing. Infrastructure development cycles extend over years, while AI demand evolves in months. Operators must therefore commit capital before market conditions fully reveal themselves, introducing a level of strategic risk unfamiliar to earlier phases of cloud expansion.
CBRE’s research shows that even with significant new supply entering the European market, vacancy rates are forecast to fall sharply, reflecting how quickly AI workloads absorb available capacity. This imbalance is reshaping competitive dynamics, rewarding regions capable of delivering scalable power infrastructure and penalising locations constrained by ageing grids or slow planning processes.
“The difference today is that operators are no longer planning for incremental growth,” Cooper says. “They are planning for exponential demand. That requires regions with credible long-term grid reinforcement strategies, access to multiple substations and confidence that expansion will remain viable over time.”
The rise of new European data centre geographies
Traditional hubs such as London, Frankfurt and Dublin remain central to Europe’s digital ecosystem, yet power constraints and planning delays are pushing operators toward secondary and emerging markets. This shift does not represent abandonment of established clusters, but rather an evolution toward diversified infrastructure portfolios.
“Hyperscale operators are increasingly balancing two objectives,” Cooper says. “They want the ecosystem advantages of established hubs, but they also need locations where power is available both today and in the future. Secondary markets are becoming attractive because they offer scalability rather than immediate saturation.”
The decision increasingly revolves around speed versus longevity. Established markets provide immediate interconnection density and customer proximity, while emerging regions offer long-term expansion potential and improved economics. Successful strategies increasingly combine both approaches, creating distributed portfolios that mitigate risk while maintaining performance.
Network connectivity remains critical within this calculus. AI inference workloads, real-time analytics and latency-sensitive enterprise applications depend on predictable performance across distributed environments. Fibre density, carrier diversity and proximity to internet exchange points now influence not only user experience but also infrastructure valuation.
“Latency has become a business metric rather than a technical specification,” Cooper notes. “Operators are looking for locations where network resilience and connectivity diversity reduce operational risk. In many cases, short-term available capacity with strong connectivity is commanding a premium because it enables faster deployment.”
This convergence of power and network considerations is redefining how digital infrastructure ecosystems emerge, turning geography into a competitive differentiator once again.
Economics in the age of density
Electricity has always represented a significant operational cost, but AI-driven density magnifies its importance dramatically. Even marginal differences in energy pricing now translate into substantial long-term financial consequences, particularly for facilities operating continuously at high load.
“Energy cost is no longer just an operating expense,” Cooper says. “It is a strategic determinant of competitiveness. When density increases, small variations in pricing or volatility can materially affect the economics of a deployment over its lifetime.”
Operators are responding through long-term power purchase agreements and, increasingly, direct investment in renewable energy generation to stabilise pricing and meet sustainability commitments. The objective is not simply cost reduction but predictability, allowing infrastructure planning to align with long-term investment horizons.
Total cost of ownership therefore extends beyond construction and operational expenditure to include regulatory exposure, taxation frameworks, incentives and scalability economics. In constrained markets, operators may accept higher upfront costs if locations provide secure access to future power capacity.
“The supply-constrained environment means availability sometimes outweighs price,” Cooper explains. “Access to reliable power can justify higher costs because the alternative may be delayed deployment or lost market opportunity.”
The result is a more sophisticated evaluation framework, one that integrates infrastructure economics with long-term operational strategy.
Sustainability as operational necessity
Sustainability considerations have shifted from reputational concern to operational requirement as investors, regulators and customers scrutinise the environmental impact of digital infrastructure. AI workloads, with their intensive energy demands, have accelerated this transition.
“ESG expectations are now embedded within site selection,” Cooper says. “Operators must demonstrate energy efficiency, responsible resource use and alignment with net-zero objectives from the outset. Sustainability is no longer an optional overlay; it directly influences operating cost and regulatory approval.”
Access to renewable energy sources has become a decisive factor, particularly as operators seek to reduce exposure to carbon pricing and energy volatility. Climate conditions also play a growing role, with cooler regions enabling more efficient cooling strategies and lower operational expenditure.
Advances in liquid and hybrid cooling technologies are reshaping facility design, allowing higher density deployments while reducing environmental impact. At the same time, water usage has emerged as a critical consideration, particularly in regions facing resource constraints.
“Cooling efficiency, renewable availability and water management are increasingly interconnected,” Cooper explains. “Operators are evaluating how environmental conditions influence both sustainability performance and long-term cost structures.”
Transparency through efficiency metrics and ESG reporting has become standard among leading hyperscale operators, reinforcing sustainability as a core component of infrastructure strategy rather than a compliance exercise.
Planning for a different future
Regulatory scrutiny is intensifying as governments recognise data centres as critical national infrastructure. Planning approvals, environmental assessments and community engagement now play a central role in determining project timelines and feasibility.
‘“Successful site selection requires early engagement with utilities and regulators,” Cooper says. “Planning certainty is becoming just as important as physical suitability because delays can undermine business cases in fast-moving markets.”
Brownfield redevelopment is gaining traction as operators repurpose industrial sites with existing power and fibre infrastructure, accelerating deployment while supporting sustainability goals. These projects illustrate how infrastructure evolution increasingly intersects with urban regeneration and industrial policy.
The broader implication is that data centres are no longer isolated technical facilities. They are becoming foundational components of national digital economies, shaping employment, investment and energy strategy simultaneously.
Infrastructure decisions that define decades
The acceleration of artificial intelligence has transformed data centre development into a long-term strategic discipline. Site selection now determines not only operational efficiency but also competitiveness, resilience and adaptability across decades of technological change.
“The most successful developments are those informed by rigorous evaluation and long-term thinking,” Cooper says. “Operators must anticipate future density, regulatory evolution and environmental expectations rather than reacting to immediate demand.”
This perspective reframes infrastructure planning as an exercise in foresight rather than optimisation. Decisions made today will govern how effectively organisations can scale intelligence tomorrow.
As AI continues to reshape compute demand, the geography of digital infrastructure is being rewritten. Power availability, network performance, sustainability and regulatory certainty are converging into a single strategic equation, one that determines where innovation can happen and how quickly it can scale.
The organisations that recognise this shift early will treat site selection not as a logistical step, but as a defining strategic choice. In the age of artificial intelligence, where infrastructure determines capability, location is no longer merely about place. It is about possibility.
