By Eva Sóley Guðbjörnsdóttir
A Nordic blueprint for AI infrastructure demonstrates scalability and sustainably, combining clean energy, efficient cooling and heat reuse, to balance technical excellence and growth with environmental responsibility.
Are we destined for a world where “the entire surface of the earth will be covered with solar panels and data centers,” as predicted by Ilya Sutskever, chief scientist at OpenAI? We hope not and are committed to balancing AI growth with environmental sustainability. But how can businesses, nations and economic regions capitalize on the AI boom without risking the planet? We believe that there is a specific operational model that is already striking this balance in the Nordics and that could be used as a blueprint for other regions.
The Financial Times recently estimated that the big four tech companies (Google, Meta, Microsoft and Amazon) will spend $635 billion on AI infrastructure this year. This month, the European Commission is expected to publish its Cloud and AI Development Act (CADA) aimed squarely at securing Europe’s future in this rapidly evolving market. We are in the midst of a ‘gold-rush’ for data centers to support AI growth.
As the different types of AI workloads become more sophisticated, there are further opportunities to combine opportunities for growth with enhanced environmental performance.
But, amidst all the investment hype, there is growing concern about the environmental impact of data centers, especially with the increasing development of AI facilities specifically. Exponential growth in power demand, water usage, and impact on land, local ecosystems and communities must all be weighed against the advantages AI delivers. In the Nordics, we believe that measurable and managed impact on our environment, combined with integrated planning and technological innovation are a model that should define data center development. In this way, we can capitalize on the AI revolution and still protect our planet.
Staying Cool
For AI operators, the attraction of the Nordics is a matter of fundamental physics and economics. The computers in data centers get hot as their chips process work from customers and they need to be constantly cooled. The less energy used to cool hot servers, the more efficient and environmentally sustainable the data center. Generative AI clusters generate 3.5 times more heat than legacy applications, making thermal management a key operational challenge. As AI chips create more heat, they will be increasingly cooled by new technologies that use liquids in direct contact with the chips.
The cool average temperatures in the Nordics allow free-air cooling for most of the year. This reduces the need for electromechanical cooling systems required to chill liquids used for direct to chip cooling. This lowers overall power consumption and improving a key metric of environmental sustainability Power Usage Effectiveness (PUE). Nordic data centers are among the most efficient in the world on this metric: our own data centers operate at a PUE of around 1.2, well below the European average of 1.36.
Circular Intelligence: Using Green Energy Twice
Not only are these closed-loop liquid cooling systems much more efficient in energy and water use, but they provide a more effective heat transfer. This fact, combined with cooler average climate makes the most distinctive feature of the Nordic model for AI data centers possible: the systemic integration of data centers into municipal heating infrastructures. Commonly referred to as “waste heat,” the heat created by servers can be a valuable resource. Through partnerships with municipal district heating networks, data centers effectively ‘reuse’ energy by offering residual heat to warm houses, workplaces and specialist facilities. For example, in Finland, excess heat from our data center in Espoo keeps the neighbouring Kesko superstore warm year-round helping the retailer achieve their goal of 50% decrease in scope 1 and 2 carbon emissions. Liquid cooling captures 90-95% of server heat and when combined with heat pumps can easily reach the temperatures required by municipal heating systems. This model transforms the data center from a passive consumer of resources into an active utility.
Strategic Division of AI Workloads
As the different types of AI workloads become more sophisticated, there are further opportunities to combine opportunities for growth with enhanced environmental performance. One specific area of consideration is the differentiation between model training, which is highly energy-intensive but latency-tolerant, and inference, the point at which AI is used to make decisions, which requires less raw compute and storage capacity, but faster responsiveness.
The Nordic region remains one of the best-connected areas for data centers with high-speed fiber connecting it not only to the large populations and economic hubs of Western Europe, but also to the US, and with the imminent landing of trans-arctic cables, to high growth markets in Asia. However, the cold climate and relatively low-cost geothermal and hydro power also make it an ideal location for AI training hubs. If AI has a power consumption issue, then the Nordics, with virtually carbon-free electricity generation, have a solution.
But, of course, local and regional businesses are also customers for AI and want their data used for AI inference to remain close by and defended by Europe’s leading privacy and data security laws. Plus, the high digital penetration among citizens in Nordic countries, and the highly skilled workforce make them excellent to nurture new AI-based businesses. So, there are also huge opportunities to leverage sustainable data centers that host AI inference. Climate, economic and political advantages all align to create significant pull for AI data centers in Europe’s North.
The Regulatory Advantage: Balancing Growth and Trust
Europe continues to lay the foundations that have helped create trust in the digital economy, which puts European operators at an advantage as scrutiny over environmental impact increases. From the General Data Protection Regulation (GDPR) in 2018 to the Data Act of 2025 strong foundations allow businesses to plan. The forthcoming Cloud and AI Development Act (CADA) may build on this approach by bringing together initiatives designed to drive investment and growth in cloud and AI with environmental safeguards.
Whilst the text has yet to be published, we expect to see funds and locations for AI investments tied explicitly to environmental performance.
Whilst the text has yet to be published, we expect to see funds and locations for AI investments tied explicitly to environmental performance. The EU Energy Efficiency Directive (EEED) has already established benchmarks for reporting to allow for data-driven assessment of environmental performance of data centers. Our data center of the future blueprint embeds these regulations as competitive advantages for cloud and AI customers, and ultimately citizens and consumers who want the services they consume to be housed in ‘green’ data centers.
AI as an Ally to Sustainability
The Nordic markets prove that AI and sustainability are not in opposition, provided there is careful planning and a model that places infrastructure at the heart of economic and environmental development. While the aggregate energy consumption of the AI sector is growing, the aggregate savings from optimized, high-density Nordic data centers represent a way to balance this with the advantages and opportunities of Europe’s AI ambitions.
For investors, technology providers and those seeking to leverage the potential of AI, the takeaway is clear: the most competitive edge in the AI era is not just the number of chips in your racks, but how sustainably you power them and how efficiently you reuse their waste. By aligning energy policy, municipal infrastructure, and technological innovation, the Nordics have provided a global case study for responsible, innovative, and future-oriented data center development.
Eva Sóley Guðbjörnsdóttir
