In Northern Europe, datacenter operators are grappling with rising costs and project delays due to U.S. tariffs and global geopolitical tensions. The demand for AI infrastructure to support larger model training has surged significantly, amplifying pressure on both existing and planned datacenter capacities. A study by Onnec highlighted a 42% increase in AI workloads over the past year, with two-thirds of operators reporting unexpected demands that stretch their resources.

The report, based on feedback from 250 senior decision-makers in the UK, Ireland, and Nordic countries, cites skills shortages, geopolitical turbulence, and complex designs as key hurdles. Britain’s Prime Minister Keir Starmer may find these findings concerning, especially as the government prioritizes rapidly building datacenters to support AI growth while navigating concurrent political challenges.

Tariffs imposed by the U.S. administration have been highlighted by 69% of participants as cost-inflating and delay-inducing factors. Past warnings already suggested that such tariffs could increase server prices and disrupt supply chains reliant on parts from multiple sources.

Onnec Nordics’ General Manager cautioned that protectionist measures are compelling operators to make cost-driven decisions, which inadvertently limits the quality of infrastructure they can provide. The report further suggests that the current AI frenzy could significantly reduce the lifespan of datacenters, as they struggle to meet growing power, cooling, and bandwidth demands.

Adaptation plans have arisen, with 74% of operators reassessing strategies for power, cooling, and site selection to build AI-ready facilities. This industry shift has already impacted major players like Microsoft, which reconsidered its datacenter requirements to better align with power and cooling needs.

AI’s rapid evolution demands resilient datacenters built with future capacities in mind. Meanwhile, construction costs in Europe continue to climb, with Savills reporting a 6.5% rise in 2024 costs to $9.1 million per MW, making Zurich, London, and Frankfurt particularly costly. Despite land, mechanical, and electrical systems being the primary cost drivers, skill shortages pose a notable risk of project delays. Ignoring cabling, often an overlooked aspect of datacenter setups, can jeopardize performance and resilience, with 70% of operators acknowledging its impact. Onnec’s findings underscore the importance of strategic planning to overcome these challenges and sustain AI infrastructure development long-term.