Marco Kisic

27-01-2025 14:45

Powering the AI revolution: Opportunities for Nordic companies

Head of ESG Research in Nordea Equities, Marco Kisic, shares the findings from the latest analysis, “Powering the AI revolution,” including the outlook for data centres’ growth, their exponential increase in energy needs, key bottlenecks and the implications for Nordic companies.
Woman operating digital interface technology

In the near term, the AI revolution will likely drive increased greenhouse gas emissions, with its heavy need for computer power from servers housed in data centres. Longer out, however, it has the potential to give fresh impetus to the energy transformation and the green transition. 

In our recent report, we built a scenario around data centre expansion in 2024-30, and find global data centre capex growing to around USD 650bn by 2030. We see three key levers to make our scenario happen: energy supply, grid infrastructure and specialised cooling. We also see the Nordic region standing out as a potential data centre hub of the future.

The data centre revolution

There are currently around 8,000 data centres in the world, consuming some 1-1.5% of global electricity and generating around 1% of the world’s total greenhouse gas emissions, according to the International Energy Agency (IEA). The rise of AI will transform the size of data centres and their infrastructure, as well as exponentially increase the amount of energy they consume.

In fact, AI will multiply data centres' energy requirements, due to the massive processing power required for specific tasks, including training large language models such as ChatGPT. With more energy-thirsty graphic processing unit (GPU) architecture replacing central processing units (CPU), these data centres will have vast power and cooling needs. 

Today's data centre infrastructure is often not equipped to handle AI workloads, as traditional cloud computing was built for more general functions, such as web servers or e-commerce websites. As a result, data centre infrastructure and the energy supply will need to transform.

The AI revolution will not be possible unless data centres become smarter, more sustainable and more strategically-located assets. In our view, if this were successful, it would offer tremendous opportunities to companies exposed to data centre infrastructure, ranging from HVAC to power, grid and energy supply firms. 

Source: Nordea estimates, Schneider Electric and Dell'Oro Group

The Nordics as an emerging data centre hotspot

The US is by far the number-one region right now when it comes to data centres, accounting for some 40-45% of global capacity, according to the IEA. Europe trails in second place, with the bulk of European data centres located in Frankfurt, London, Amsterdam, Paris and Dublin. 

As we see it, the Nordic region in the medium and long term has the potential to emerge as a winner in the data centre AI transition, making this space highly relevant to Nordic companies. The region has several key ingredients for the AI-driven development of such industry: cool temperatures, low energy prices, plentiful CO2-free electricity, abundant water and political stability. 

Source: Data Centre Dynamics, Klimstatus, Data Centre Mapping, RISE, The Norwegian Datacenter Database, Nordea estimates

Our scenario: A USD ~600bn market by 2030

For our latest report, we designed a proprietary model to estimate the energy consumption and size of the data centre market by 2030, including the related cooling market. Our estimates are based on market consensus expectations for sales of GPUs and CPUs from NVIDIA and other key players in the field. 

We find that global data centre capex could expand to USD 650bn by 2030 (representing a 15% compound annual growth rate in 2024-30), of which around 10% is for cooling and around 15-20% is for electrical systems. Data centres are set to become one of the fastest-growing consumers of electricity and account for 8.4% and 6.3% of power demand in the US and Europe by 2030, respectively – double the current levels.

The key bottlenecks in our scenario include the energy supply, infrastructure, regulation and equipment availability – especially of transformers, which could slow down growth rates.

We expect data centre operators to strive as much as they can for renewable energy sources, accelerating the wave of renewable commissioning and installations. However, given the quick turnaround time needed for data centre construction, in the near and medium term, we believe data centres will continue to rely on fossil fuel-based energy. In our scenario, data centre operators ultimately transition to renewables, but this will take time.

Glossary of terms

Co-location: Also known as "carrier hotels", these are data centres where companies can rent equipment, space and bandwidth from data centre owners. 

Enterprise: These data centres are fully owned by individual companies and are used to process internal data and host mission-critical applications. 

Hyperscale: Engineered for large-scale workloads and minimised latency, hyperscale data centres are similar to enterprise data centres, but offer extreme scalability capabilities and better economics. The largest hyperscalers are Amazon Web Services, Microsoft Azur and Google Cloud. 

Edge: A small version of a data centre, as close to the end user as possible, just handling data processed in a specific region to minimise latency and lag.

Author

Name:
Marco Kisic
Title:
Head of ESG Research, Equities
After reading this article, is your perception of Nordea?