Record temperatures are putting Europe’s digital infrastructure under growing strain, just as AI and high-density computing push internal heat loads higher. For data centre operators, cooling resilience is becoming a critical issue for business continuity, energy efficiency and long-term facility planning.
Cooling demand rises with computing power
Data centres underpin cloud services, payment systems and communications, but their reliability depends heavily on effective heat management. During extreme weather, cooling systems must work harder while increasingly powerful servers generate more heat inside the facility.
Mathias Franke, Head of Data Center Consulting at Drees & Sommer SE, warns that heat can no longer be treated as a secondary concern. Rising outdoor temperatures are combining with higher internal loads from AI applications and powerful chips, increasing the risk of disruption where cooling capacity has not kept pace.
The challenge is amplified by rapid sector growth. According to facility-management.de, Germany’s installed data centre capacity has more than doubled since 2010, reaching more than 2,730 MW in 2024. Further substantial expansion is expected by 2030.
Climate exposure is also becoming harder to ignore. A 2026 analysis by physical climate risk specialist First Street found that 54% of global data centre capacity is located in markets exposed to chronic heat or drought stress.
Designing for tomorrow’s temperatures
There is a growing need to consider future climate conditions when selecting sites, specifying cooling technology and developing heatwave operating procedures.
Research by Google and the University of Pennsylvania reinforces the point. Their Prometheus study estimates that data centres may need around 11% more cooling capacity on average by 2044 simply to maintain today’s level of outage risk. At particularly vulnerable locations, additional cooling requirements could reach 48%.
Operational measures can also reduce pressure during heatwaves. Energy-intensive activities such as backups, data analysis and major migrations can be moved to cooler night-time periods. Cooling redundancy is equally important: if one system fails, another must be capable of taking over immediately, while emergency power must support cooling as well as IT equipment.
Liquid cooling gains ground
As rack densities approach 50 kW and above, conventional air cooling can struggle to remove enough heat. This is encouraging operators to adopt water-based and liquid cooling technologies, particularly for AI and high-performance computing applications.
The shift also has an efficiency dimension. Under Germany’s Energy Efficiency Act, newly commissioned data centres face stricter Power Usage Effectiveness (PUE) requirements, although proposed amendments to the legislation could change some thresholds.
Water availability, however, must form part of the resilience calculation. In drought-prone regions, cooling demand can create another operational vulnerability. Closed-loop systems can mitigate this by substantially reducing ongoing water consumption.
The greatest exposure may ultimately lie outside large purpose-built data centres. Smaller decentralised server rooms in offices, basements and converted technical spaces often lack sophisticated cooling, redundancy and emergency power – making them particularly vulnerable as extreme summer temperatures become more frequent.
Planning for a Hotter Future
For facility managers and data centre operators, resilience starts well before the next heatwave. Future temperature extremes should be factored into site selection, cooling design and capacity planning, alongside local water availability, grid resilience and opportunities for on-site energy generation and storage.
Existing facilities should regularly review cooling capacity, redundancy and emergency power systems, while operational plans should identify which computing loads can be shifted away from peak-temperature periods.
With higher-density AI infrastructure adding to internal heat loads, climate resilience can no longer be treated as an emergency response: it needs to become a core consideration in data centre design, investment and day-to-day facility management.