As Europe swelters through another summer of record-breaking temperatures, incredibly hot tube rides and swimming in canals, concerns about heat-related disruption are spreading far beyond agriculture, transport and public health.
Increasingly, attention is turning towards the digital infrastructure powering everything from cloud computing to artificial intelligence.
This has always been a topic for discussion, mostly from a global warming and environmental standpoint. But now, even more so with Europe investing heavily in AI and with governments and businesses racing to build new data centres and computing capacity. At the same time, however, what’s different now is that the continent is experiencing more frequent and more intense heatwaves, placing growing pressure on energy grids, water supplies and cooling systems.
The question at this point, however, is whether these two trends are on a collision course.
AI’s Growing Appetite For Power
The rise of generative AI has triggered a huge increase in demand for computing power. As a result, according to European Business Magazine, AI is creating a new “data heat” challenge as increasingly powerful models require vast amounts of electricity and cooling infrastructure to operate.
Unlike traditional enterprise workloads, AI training and inference often run at extremely high densities, generating substantial amounts of heat. As more organisations adopt AI tools, demand for data centre capacity is expected to continue rising throughout the decade.
But, AI’s infrastructure challenge isn’t just about building more servers. As Emil Romanus, Founder of Farang, explains, “heatwaves exposed an existing European problem. Its digital infrastructure is under pressure because the AI industry has built an enormous and largely unquestioned dependency on brute-force compute.”
“The real question is not how we keep the data centres cool, but whether the compute demands driving their growth are actually necessary.”
And that, in itself, brings a broader question to light: are heatwaves exposing a climate problem, or simply revealing deeper inefficiencies in how the AI industry currently operates? It’s a chicken-egg situation.
A Stress Test for Europe’s Infrastructure?
Several experts argue that the real concern isn’t just a few individual, isolated heatwaves, but the way extreme weather compounds existing infrastructure weaknesses.
According to CNN, data centres are becoming increasingly vulnerable to extreme temperatures as climate change places greater strain on cooling systems and electricity networks. At the same time, operators face growing pressure to reduce energy and water consumption.
James Barnes, CEO of StatusCake, believes firmly that the issue extends beyond data centres themselves. “The much bigger issue is that the UK, US and Europe have underinvested in their grids, water infrastructure and energy storage for decades.”
“My view”, Barnes adds, “is the real risk is poor resilience planning and investment. Without stronger grids, renewable storage etc, more websites, apps and online services could be affected during extreme weather.”
Peter Juhasz, Co-Founder and CEO of Syrvi AI, echoes this view: “Heatwaves are not going to melt Europe’s data centres, but they are a stress test most of the estate was never designed to pass.”
Perhaps more importantly, he notes that, “the hottest days are exactly when the grid is tightest, when cooling demand peaks and when water is scarcest, all at the same moment.”
In other words, the danger comes when multiple pressures converge simultaneously. And unfortunately, as we’re starting to see now, this is exactly what’s beginning to happen.
Cooling Is Becoming A Strategic Issue
Historically, cooling was largely viewed as an engineering problem, but now, it’s a lot more of a strategic one. According to AI Insight Note, many operators are already rethinking where future AI infrastructure should be located, with cooler climates becoming increasingly attractive.
Alex MacColl, PM EMEA at DataMove, says this shift has already begun: “Northbound moves are taking place – especially in the AI / hyperscale sector – to the Nordics where natural cooling and abundant power is available.”
He adds that, “heat and the associated cost of cooling has become a major strategic factor, not just a seasonal inconvenience.”
Now, others believe new cooling technologies may help reduce the problem. Paul Quigley, Chief Strategic Relations Officer at Airsys Cooling Technologies, argues that older facilities are consuming excessive amounts of power simply to stay cool. He says that “older data centers across Europe are pulling more power than they need to.”
Indeed, according to Quigley, the solution for this is to ” deploy precision liquid cooling.” Kate Steele, Director, EMEA HPC/AI at Lenovo Infrastructure Solutions Group, says that “climate volatility, such as Europe’s recent heatwaves, is already reshaping how data centers are designed and operated.”
She points to closed-loop liquid cooling systems as one way operators can improve resilience while reducing dependence on fresh water supplies.
Is This The Beginning Of A Bigger Problem?
Perhaps the most important takeaway is that heatwaves themselves are not necessarily the problem. Instead, they may be highlighting a larger challenge facing Europe as AI adoption accelerates, and in turn, perhaps that gives us an opportunity to come up with a solution.
Dan Herbatschek, CEO and Founder of Ramsey Theory Group, argues that energy has become the limiting factor for AI growth.
“Forget worrying about compute capacity, the biggest limiting factor to AI performance is the power driving it, meaning energy and cooling capacity.”
Meanwhile, François Le Scornet, President of Carbonexit Consulting, warns that rising temperatures can create cascading pressures on both infrastructure and energy systems.
“The most likely scenario is not ‘France’s internet collapses, of course. Instead, we can expect potential localised outages and reduced cloud performance.”
For now, Europe’s digital infrastructure remains highly resilient. But as AI workloads continue to grow and climate pressures intensify, operators may find themselves planning for a future where extreme heat is no longer an occasional disruption but a regular operating condition.
The challenge facing Europe isn’t just whether data centres can survive the next heatwave. Rather, the difficulty is whether the continent can build an AI ecosystem that remains resilient in a world where extreme weather becomes the norm rather than the exception.
More from Artificial Intelligence
Our Experts:
- Emil Romanus: Founder of Farang
- James Barnes: CEO of StatusCake
- Paul Quigley: Chief Strategic Relations Officer at Airsys Cooling Technologies
- Peter Juhasz: Co-Founder and CEO at Syrvi AI
- Dan Herbatschek: CEO and Founder of Ramsey Theory Group
- Alex MacColl: PM EMEA at DataMove
- François Le Scornet: President, Cleantech and Climate Tech Senior Consultant at Carbonexit Consulting
- Patrick Keaney: Global Engineering Executive at SLR
- Kate Steele: Director, EMEA HPC/AI at Lenovo Infrastructure Solutions Group
- Colin Rees: Co-Lead of Consultancy at IES
Emil Romanus, Founder of Farang
“Heatwaves exposed an existing European problem. Its digital infrastructure is under pressure because the AI industry has built an enormous and largely unquestioned dependency on brute-force compute. More capability has become synonymous with more power, more cooling, more infrastructure.
“The real question is not how we keep the data centres cool, but whether the compute demands driving their growth are actually necessary. Approaches that find simpler internal representations of hard problems, rather than searching exhaustively for answers, can achieve strong results at a fraction of the compute cost and power.”
James Barnes, CEO of StatusCake
“Yes, I do think heatwaves should be a wake-up call for Europe’s digital infrastructure, but the risk needs to be put in context. The reductions in power output from French nuclear power stations is due to environmental safeguards; limiting the discharge of heated water into rivers. The much bigger issue is that the UK, US and Europe have underinvested in their grids, water infrastructure and energy storage for decades.
“That doesn’t mean data centres should be built blindly; their impact on local power and water supplies should be properly assessed. But we should handle through local planning, investment and smarter designs (e.g. using data centre discharged heat to warm houses, greenhouses etc), rather than treating data centres as the problem in isolation. Compared with many industries, data centres remain a relatively small water user.
“My view is the real risk is poor resilience planning and investment. Without stronger grids, renewable storage etc, more websites, apps and online services could be affected during extreme weather.”
Paul Quigley, Chief Strategic Relations Officer at Airsys Cooling Technologies
“Europe is hot and understandably frustrated. Here’s what actually works right now, without waiting for grids or permits to catch up. Older data centers across Europe are pulling more power than they need to—these sites consume 35–45 percent of total power just to cool equipment, often via water-intensive systems that fail during heatwaves.
“Here’s the fix: Deploy precision liquid cooling. Direct to server spray cooling that is rack-ready. It permits hybrid AI data center upgrades without facility-wide overhauls, can eliminate water consumption, and deploys in months, not years. The result: recover wasted megawatts and dramatically increase compute output per megawatt of power consumed. Return that power to heat-stressed grids—as Texas already does and Europe is planning—or grow compute at existing sites without new builds. This infrastructure is deployable today. It solves three problems at once: power waste, water waste, and stranded assets.”
Peter Juhasz, Co-Founder and CEO at Syrvi AI
“Heatwaves are not going to melt Europe’s data centres, but they are a stress test most of the estate was never designed to pass. The honest answer is that the risk is real and rising, just not in the dramatic way the headlines imply.
“The danger is the pile-up. The hottest days are exactly when the grid is tightest, when cooling demand peaks and when water is scarcest, all at the same moment. A data centre does not fail because of one hot afternoon, it fails when those pressures arrive together and the design assumed they never would.
“Heat is not the root problem. We are scaling AI compute faster than we are scaling resilient, water-aware infrastructure, and a 40-degree June simply makes that visible. Plenty of facilities were built for a cooler climate and sited with cheap power in mind, not water stress.
“This is becoming a competitive issue, not just an environmental one. Operators who treat 40 degrees as the new normal, who plan for water as carefully as power, and who build resilience into siting and cooling will keep running. Those treating this summer as a freak event are quietly storing up trouble. With AI demand set to double by 2030, the heat question is really a planning question: build for the climate we are getting, not the one we used to have.”
Dan Herbatschek, CEO and Founder of Ramsey Theory Group
“Forget worrying about compute capacity, the biggest limiting factor to AI performance is the power driving it, meaning energy and cooling capacity. I’ve seen in Europe everything from enterprise AI platforms to critical public services failing as they experience operational overload in the heat. I have seen reduced generation capacity, transmission network overload and loss of critical cooling capabilities.
“The end result is that millions of people are affected, so there has to be much better planning for climate resistance by all organizations, both public and private. The solution includes making sure workloads are distributed geographically and not all in one region, create scheduling that is energy aware, and most importantly there needs to be infrastructure monitoring in real-time with better human oversight. Heatwaves are a big threat to both Europe’s digital infrastructure and also quality of life as a result.”
Alex MacColl, PM EMEA at DataMove
“Heatwaves aren’t just a risk – they’re a growing operational pressure, and we are already seeing the consequences in where organisations are choosing to locate their infrastructure.
“Extreme heat can hit data centres two-fold. Cooling demand will spike precisely when the grid is already under heavy stress and less able to supply it. Major AI workloads with dense compute face a threat to uptime, not to mention cost.
“On the ground, what we are seeing is relocation and forward planning as a direct response. Northbound moves are taking place – especially in the AI / hyperscale sector – to the Nordics where natural cooling and abundant power is available. If an organisation is choosing to relocate or plan new build out in Sweden or Norway rather than risk existing infrastructure, that tells you that heat – and the associated cost of cooling has become a major strategic factor, not just a seasonal inconvenience.”
François Le Scornet, President, Cleantech and Climate Tech Senior Consultant at Carbonexit Consulting
“In case of a heatwave, higher power consumption occurs as cooling systems work harder. French data centres already use about 8 TWh per year. This figure rises to nearly 10 TWh if you include distributed server rooms, making up almost 2% of national electricity demand. Loss of redundancy happens at temperatures between 38 and 40°C, putting stress on chillers, pumps, compressors, and batteries. If one part fails, the system can shift from “N+1 redundancy” to “no safety margin.” This means that service degradation may push operators to throttle servers, shut down racks as a precaution, migrate cloud workloads, or tolerate higher latency.
“For users, this can lead to slower services, partial outages, failed transactions, or reduced cloud availability. Grid stress also becomes an issue. During the current heatwave, French nuclear output reportedly dropped by 4.1 GW, which is about 7% of demand, becaus river temperatures limited cooling (2 nuclear reactors were stopped). If such a heatwave were to happen again, the most likely scenario is not “France’s internet collapses” of course. Instead, we can expect potential localized outages and reduced cloud performance. For the operators, this also means increased operating costs and more competition for grid resources during simultaneous heat and AI demand peaks.”
Patrick Keaney, Global Engineering Executive at SLR
“To be net positive on water, the tech sector has to move past just being more efficient and start taking real accountability for its impact. That starts with measuring water the right way – at the basin level, not just total consumption – and shifting toward non-potable and recycled sources wherever possible. It also means putting water back into the system through replenishment, not just reducing use.
“But the real unlock is how AI gets applied. If it’s focused on solving meaningful water problems – reducing losses in utilities, improving industrial reuse, optimising agriculture, the impact can far outweigh the sector’s own footprint. A few companies are starting to head in this direction with water-positive commitments, but most are still early. The ones that will get there are the ones that treat water as a real operating constraint and aim their technology at problems that actually move the needle.”
Kate Steele, Director, EMEA HPC/AI at Lenovo Infrastructure Solutions Group
“Climate volatility, such as Europe’s recent heatwaves, is already reshaping how data centers are designed and operated. For Europe’s AI infrastructure that makes resilience a present-day priority rather than a future concern.
“The key point is that sustainability and resilience can no longer sit in separate conversations. Reducing reliance on energy and fresh water does not just support emissions goals; it also makes data centers less exposed when resources come under pressure during extreme weather. That means moving beyond static design assumptions and building around local climate risk, from site selection through to cooling and hardware choices.
“Closed-loop liquid cooling is a strong example of this mindset, retaining and reusing water while helping high-performance systems operate more efficiently under stress.”
Colin Rees, Co-Lead of Consultancy at IES
“Heatwaves are becoming a serious design and operational test for Europe’s data centres. Higher ambient temperatures reduce cooling efficiency just as AI workloads push up rack density and electricity demand.
“Conditions around the site also need closer scrutiny. Heat discharged by cooling equipment can recirculate into air intakes, increasing energy use and the risk of equipment overheating.
“We must utilise building performance modelling alongside CFD analysis to test facilities under a range of weather conditions. This can reveal hotspots, airflow short circuits and whether cooling systems have enough capacity during extreme heat or equipment failure. Performance digital twins then help operators compare predicted behaviour with live data and understand the effect on annual PUE and operating costs.
“Europe can expand AI capacity more safely by designing for future climate conditions from the outset. Early testing gives project teams stronger evidence for cooling decisions and helps operators protect uptime without wasting energy.”
