Nordic blueprint for sustainable, scalable AI data center infrastructure

atNorth (article author) presents a Nordic operational model as a blueprint for scaling AI infrastructure while minimising environmental impact.

• Main announcement/action: atNorth outlines a Nordic blueprint combining free-air cooling, closed-loop liquid cooling, municipal heat reuse, and near-zero carbon electricity as a replicable model for sustainable AI data centers. Key concrete details: atNorth data centers operate at PUE ~1.2 (compared with a European average of 1.36), liquid cooling captures 90-95% of server heat, and their Espoo (Finland) data center supplies excess heat to Kesko, which aims for a 50% decrease in scope 1 and 2 emissions. The article also cites a Financial Times estimate of $635 billion in AI infrastructure spend by Google, Meta, Microsoft and Amazon this year.
• Background and context: The piece situates the blueprint within existing and forthcoming regulation (GDPR, Data Act 2025, EU Energy Efficiency Directive, and the expected Cloud and AI Development Act (CADA) — not yet published). It describes technical and market enablers: trans-arctic cable landings, virtually carbon-free hydro/geothermal power in the Nordics, and the operational distinction between training (energy intensive) and inference (latency sensitive) workloads. The article is an opinion/blueprint piece by atNorth leadership rather than a formal policy or regulatory announcement.