The UK startup captures up to 95% of server waste heat, using it to warm swimming pools and city buildings while keeping data center efficiency near theoretical limits.
Deep Green, a UK-based company, builds modular, high-density data centers designed specifically for AI and high-performance computing workloads. Its closed-loop, waterless cooling systems capture and reuse up to 95% of waste heat, routing it to nearby buildings, pools, and district heating networks. The result is a Power Usage Effectiveness rating between 1.03 and 1.07.
Using the electrons twice
Deep Green’s proof of concept sits in Urmston, Greater Manchester. A 400 kW facility there pipes excess heat directly to the Trafford Leisure Centre’s swimming pool, saving the facility roughly £80,000 per year in heating costs and cutting CO2 emissions by an estimated 100 to 150 tonnes annually.
CEO Mark Lee has described the philosophy as a commitment to “use the electrons twice.”
Deep Green’s 5.6 MW facility in Bradford secured planning consent on May 14, 2026, and is expected to connect to the Bradford Energy Network. Once commissioned by late 2028, the site will deliver low-carbon heat to multiple city-center buildings and is projected to reduce CO2 emissions by more than 4,500 tonnes per year.
The company’s Energy Reuse Factor, a metric measuring how much waste heat actually gets put to productive use, ranges from 82% to 93% across its facilities.
Financial backing and deployment speed
Deep Green has secured £200 million in investment from Octopus Energy Generation. The funding supports the company’s ambition to deploy 300 MW of heat-reusing data center infrastructure across the UK, Ireland, Europe, and North America.
Deep Green claims its modular colocation facilities can become operational in as little as four weeks, compared to traditional data center construction timelines that stretch into years.
The broader power problem
The EU’s revised Energy Efficiency Directive, which took effect in 2024, requires new data centers above 1 MW to report energy performance metrics and explore waste heat reuse opportunities.
