European Startups Target Data-Centre Power Crunch Amid AI Boom

European startups are emerging as key players in tackling the surging electricity demands of data centres driven by the AI boom, as projections show European data-centre consumption rising from 96 terawatt-hours (TWh) in 2024 to 236 TWh by 2035, according to UK-based think tank Ember.

Startups such as EkkoSense, etalytics, Submer, GridBeyond, and Deep Green are rolling out technologies that target the energy-intensive components of AI infrastructure, notably cooling systems, battery management, and waste heat utilisation. Their innovations are attracting interest from major telecommunication and technology companies seeking to contain operational costs and curb carbon emissions.

EkkoSense’s optimisation system, deployed across 20 sites for Virgin Media O2 in 2024, delivered an average 15% cut in cooling energy, translating to savings of more than £1 million (about €1.17 million) annually and a reduction of 760 tonnes of CO₂. Meanwhile, German startup etalytics provided simulation software for cooling at NTT data centres, with trial results showing a 19.1% drop in chiller electricity use and expected annual savings of up to 25%.

Submer, based in Spain, has developed liquid-cooling technology that can boost data-centre energy efficiency by up to 50%. Its solution, adopted by Telefónica, also enhances the total cost of ownership, offering competitive advantages as operators scale up AI capabilities.

On the grid-management front, GridBeyond’s software enables two Dublin data centres, owned by Keppel DC REIT, to provide a combined 8 megawatts (MW) of flexible capacity—helping balance demand and support grid stability. In the UK, Deep Green repurposes waste heat from data centres for local heating, with one swimming pool cutting gas use by 62% and saving over £20,000 (about €23,300) annually.

These advances are seen as crucial to helping Europe manage rising AI-driven power demand while postponing the need for costly new grid infrastructure, which can take up to 15 years to develop. For investors and policymakers, the trend highlights both the business and regulatory opportunities in supporting energy-smart AI infrastructure.

However, analysts caution that the pace of AI growth may still outstrip the mitigating effects of these innovations, keeping pressure on power grids and reinforcing the need for continued investment in generation and transmission capacity. Still, the startups are positioning themselves to lead in data-centre efficiency—a market expected to expand as AI reshapes the global technology landscape.