Data centres in Germany: between billion-euro investments and capital outflow

The data centre market in Germany is experiencing an investment boom that is difficult to explain at first glance. Leading cloud providers and institutional financial investors have recently pledged billions for sites in this country. At the same time, 69 per cent of German CFOs surveyed by us for the Business Destination Germany 2026 study rate the country’s digital infrastructure as one of the five weakest in Europe.

Anyone who understands why billions are flowing into the market despite these conditions will also recognise which reforms can make the country a permanently attractive destination for institutional capital.

Investors under pressure to act: a physical presence in the country is required

The reason for the continued investment lies not in particularly attractive risk-return profiles, but in regulatory logic: The GDPR, the requirements of public administrations and highly regulated sectors such as finance and healthcare, as well as the importance of the DE-CIX internet exchange in Frankfurt, together create a demand that no global provider can ignore. After all, anyone wishing to process German corporate and public authority data must have a physical presence in Germany. The multi-billion-euro projects announced by global providers for the development of sovereign infrastructure in Germany are a direct result of this requirement for data localisation.

However, this locational advantage masks a weakness: as soon as workloads are involved that do not require physical proximity to economic centres – such as AI training, high-performance computing or other latency-insensitive applications – Germany systematically loses its appeal in an international comparison. Capital flows show where the energy-intensive AI infrastructure of the next generation is heading.

Energy arbitrage: Germany in competition with other countries

A leading institutional infrastructure fund recently invested around US$10 billion in an AI campus in Sweden rather than in one of the traditional German locations. The decision is based on a simple calculation: AI training does not have to take place at a specific location, but requires hundreds of megawatts of electricity at competitive prices. In Finland, the commercial industrial electricity price in the second half of 2025 was around 7 cents per kilowatt-hour, in Sweden just under 10 cents, whereas in Germany it was around 23 cents.

Furthermore, hyperscalers are not only looking at price, but increasingly also at the carbon footprint of their electricity. Here, the German electricity mix has a double disadvantage: It is not only more expensive but, calculated over the year, also more carbon-intensive than the electricity in Scandinavia, which is predominantly generated from hydro and nuclear power. Added to this are annual average temperatures below ten degrees Celsius, which enable natural cooling in Northern Europe and further reduce operating costs.

Alongside energy costs, speed is a second locational disadvantage. It is not the server halls themselves that require approval, but their peripheral equipment: emergency generators and recooling systems trigger an environmental impact assessment procedure under the Federal Immission Control Act. For large projects in Germany subject to an Environmental Impact Assessment (EIA), it regularly takes four to six years from initial project planning to approval, as expert reports, completeness checks and public consultation follow one another sequentially. In Sweden and Norway, comparable projects are approved up to 24 months earlier. In a market where the technological landscape shifts on an annual basis, this time difference is a significant factor in the decision on where to locate a facility.

Ready to talk? Contact us

auto_stories

How private individuals and companies assess the infrastructure in Germany and what this means for future decisions.

Download now

Three areas for reform to boost regional competitiveness

An analysis of the location’s shortcomings highlights three areas for reform, on which Germany’s ability to remain an attractive investment destination for digital infrastructure in the long term depends.

Parallelised approval procedures

The coalition agreement has enshrined the expansion of digital infrastructure as a matter of overriding public interest. However, this commitment must now be urgently translated into administrative practice: in future, environmental impact assessments and planning permissions should be processed in parallel rather than sequentially. Standardised planning documents and digital administrative processes can reduce the duration of procedures to the level seen in neighbouring countries.
A reliable and affordable energy supply

The electricity demand of the German data centre industry has been growing rapidly for years and, according to Bitkom, is set to rise to over 30 billion kilowatt-hours by 2030. Even today, projects are failing less because of a lack of investor capital and more because of insufficient grid capacity. In hubs such as Frankfurt, the high-voltage grids are reaching their physical limits. At the same time, the expansion of transmission grids is lagging years behind demand. That is why, in addition to accelerated grid expansion, we need competitive grid tariffs and a legally secure framework for long-term power purchase agreements with renewable energy producers.
A practical Energy Efficiency Act

The Energy Efficiency Act requires new data centres, from 2028 onwards, to feed at least 20 per cent of their waste heat into district heating networks. On the one hand, this target is a sensible initiative in theory. On the other hand, however, there are practical issues regarding its implementation. Where no fourth-generation heating networks exist that can accommodate such thermal loads, the regulation falls flat. The exemption for sites located more than five kilometres from a heating network exacerbates this problem by incentivising investors to choose peripheral locations where neither networks nor consumers for the waste heat are available. It would make more sense to base efficiency assessments on case-by-case evaluations using local cost-benefit analyses and actual operational data rather than theoretical planning figures.

Two markets, one location issue: what businesses should be doing now

For investors and operators in the TMT sector, this complex environment means that projects must be underpinned by an interdisciplinary approach right from the start. The entire value chain should be taken into account: from site analysis and the identification of future-proof regions with available network capacity, through the tax structuring of complex investment models, to regulatory compliance with the Energy Efficiency Act and the NIS2 Directive. 

For the foreseeable future, there is no alternative to Germany as a location for data-sovereign cloud infrastructure. DE-CIX, the GDPR and industrial demand ensure a base level of investment that cannot be shifted by short-term location decisions. However, the next wave of digital infrastructure – from AI training clusters and HPC facilities to gigawatt-scale projects – is driven by different criteria. Here, Germany is competing not only with Dublin or Amsterdam, but increasingly with Stockholm and Oslo. Capital allocation decisions follow a return-on-investment logic that only rewards national location preferences if the framework conditions are right.

Where the US and China are pulling ahead – and where Germany is in a promising position

When it comes to deciding where to train large language models, the competition is largely decided: this race is being contested between the US and China, and Europe is likely to find it difficult to catch up. A second area, in which Germany can bring its own strengths to bear, offers better prospects. The application of AI in production and operations – that is, inference based on industrial process data – does not require gigawatt-scale training clusters, but rather proximity to machinery, specialist knowledge and reliable data sets. This is precisely where Germany’s industrial strength lies. The three areas for reform described above create the conditions to ensure that this strength does not fail due to energy costs and processing times, but instead becomes a key selling point for the next generation of infrastructure.

Interested in our services? Contact us

More KPMG Insights

Podcast

The backbone of digitalisation: How data centres are becoming a key factor in AI

High-performance data centres are a key component of the digital infrastructure and a prerequisite for the further development of AI. Katja Modder, Head of TMT at KPMG in Germany, talks to Günter Eggers, a board member of the German Data Centre Association, about power grids at breaking point, locations as a competitive advantage and regulatory hurdles. The two also assess what European ‘gigafactories’ can realistically achieve and what role Germany plays in an international comparison.

Listen now