Daniele Viappiani, Economics and Venture Capital Advisor, explains why power constraints, slow permitting and limited land availability could prove more decisive than ownership models in Europe’s digital future.

The digital sovereignty debate sweeping Europe has largely focused on ownership models, certification schemes and regulatory frameworks. While these are important considerations, the continent faces a more fundamental challenge: the data centre and AI infrastructure needed to support rising demand is not being built fast enough. For data centre operators and cloud infrastructure strategists, this gap highlights the practical constraints limiting European capacity growth.

Europe’s share of global AI computing capacity remains disproportionately small compared to North America, despite hosting some of the world’s most sophisticated digital economies. Currently, Europe accounts for just 4-5% of global AI computing capacity, or approximately 0.25 GW of the world’s total, while the United States holds roughly 70% of global AI computing power. This limitation is not only a question of who owns the infrastructure or which regulatory framework governs it, but of far more practical factors such as access to power, permitting timelines, available land and investment certainty.

The power wall

Modern hyperscale facilities require anywhere from 50 to several hundred megawatts of reliable power, and grid congestion affects most major European markets. In markets like Frankfurt, London, Amsterdam, Paris and Dublin – often referred to as FLAP-D – power availability has become the primary gatekeeper for new deployments, with grid connection wait times averaging seven to ten years. In 2023, data centres consumed between 33% and 42% of electricity demand in Amsterdam, London and Frankfurt, and nearly 80% in Dublin.

Unlike North America, where data centre operators can often secure dedicated substations or work with utilities to upgrade local infrastructure within reasonable timeframes, European grid operators are dealing with ageing infrastructure, complex ownership structures and regulatory frameworks that can significantly slow capital deployment.

AI workloads in particular demand consistent, high-density power delivery. A single training cluster for a large language model can require substantial dedicated power, and Europe’s fragmented grid infrastructure makes such deployments more challenging compared to regions with abundant, reliable power.

Permitting as a competitive disadvantage

The European permitting environment adds to the power challenge. While certain hyperscale facilities in the United States progress from site selection to operational status in 18-24 months, European timelines frequently extend to 36-48 months or longer. In most EU Member States, permitting processes alone can take 24 months or more, creating a structural disadvantage when infrastructure deployment is racing against technological evolution.

Adding to this, brownfield redevelopment, which could theoretically accelerate deployment by repurposing existing industrial sites, often faces slow approval processes. Even when suitable former manufacturing or logistics sites exist with established power infrastructure, local planning authorities frequently lack clear frameworks for data centre applications, leading to case-by-case evaluations that introduce significant uncertainty. For infrastructure investors evaluating multi-market deployment strategies, this permitting uncertainty creates material risk.

Land availability in a crowded continent

The continent’s high population density creates another fundamental constraint, especially considering hyperscale facilities typically require 10 to 30 hectares. Land suitable for such development within reasonable proximity to fibre infrastructure and power substations is rare, and therefore expensive, in most European markets. This is driving operators towards second-tier markets where land costs are lower, but connectivity and power infrastructure may be less developed.

The trade-off between land economics and infrastructure quality creates a tension that North American operators are less likely to face in markets such as Northern Virginia or central Texas, where land, fibre and power are more readily available. The Nordics and parts of Southern Europe, with uncongested grids and short wait times for connection, are expected to see data centre demand grow 150% by 2030, nearly double the rate of traditional FLAP-D markets at 55%.

For AI infrastructure specifically, which benefits from high-density campus configurations to minimise east-west latency between training clusters, Europe’s land constraints can force architectural compromises that affect performance and economics.

The capital and skills gap

Beyond physical infrastructure, Europe’s venture and growth capital environment significantly affects the pace of innovation in data centre and cloud technologies. Compared to North America, European markets generate less patient capital for infrastructure-intensive technology deployments.

This matters because next-generation data centre architectures require substantial upfront investment, and ecosystems with deeper capital pools can iterate faster, deploying and refining new approaches. European competitors, meanwhile, may remain constrained by more conservative financing structures.

The regulatory complexity layered across European markets further increases the premium investors require. The AI Act, NIS2 cybersecurity directive and various sector- and region-specific frameworks add compliance overhead that established hyperscalers can absorb, but which may hinder newer, smaller entrants attempting to scale regionally.

At the same time, data centre operations require specialised expertise in power systems, cooling architectures, network design and, increasingly, AI infrastructure optimisation. Europe produces strong technical talent, but cross-border mobility within the EU remains more complex than intra-US movement, and competition for skilled professionals from North American employers creates retention challenges.

Site selection metrics that actually matter

The European Commission’s AI Continent Action Plan set a goal of tripling data centre capacity within the next five to seven years. However, current forecasts from the IEA, ICIS, IMF and McKinsey suggest the EU may only double, rather than triple, capacity by 2030, due to grid access constraints and power procurement challenges.

The practical path to enhanced European digital resilience runs through infrastructure fundamentals: streamlined permitting for data centre development, accelerated grid infrastructure investment and clear frameworks for securing dedicated power; pre-approved sites with established power and fibre connectivity; and reduced regulatory fragmentation across EU markets to improve capital efficiency.

European expansion should be evaluated not only against regulatory compliance scorecards, but also against concrete metrics such as grid connection timelines, permitting predictability, power costs and reliability, fibre density and talent availability. Markets that improve these fundamentals are likely to attract capacity deployment regardless of ownership structures or certification frameworks.

Competition over control

The digital sovereignty conversation ultimately brings together two distinct objectives: ensuring resilient, secure infrastructure and promoting domestic industry champions. The former requires competitive markets where multiple providers deploy capacity, creating genuine choice, while the latter can tend towards consolidation and regulatory protection that may reduce resilience by limiting provider diversity.

When it comes to the data centre and cloud infrastructure industry, Europe’s technological position can strengthen through better infrastructure and competitive markets, not through regulatory frameworks alone. A durable form of digital resilience emerges when operators can deploy capacity efficiently, customers can choose among competitive providers, and regulatory and infrastructure constraints do not limit supply.

The sovereignty debate will continue. However, until Europe addresses power availability, permitting speed, land economics and investment certainty with the same focus it devotes to regulatory architecture, it will struggle to achieve a proportionate share of global data centre and AI infrastructure deployment.