Data Centers as Thermal Power Plants: Reusing AI's Waste Heat for District Heating
- 1 day ago
- 9 min read
Discover how Europe is turning digital infrastructure into district heating assets, and how Urbio helps utilities and data center operators seize those opportunities.
European data centers are growing fast and their impact on the power and heat sectors is getting harder to ignore. In 2025, data centers consumed 108 TWh, representing roughly 3.3% of total electricity demand. By 2035, that figure could double to 236 TWh, reaching almost 6% of Europe's power demand.
As hyperscalers and colocators race towards more compute and grid capacity to power generative AI models, the opportunity this represents for the heat sector is still largely overlooked.
When you prompt ChatGPT, Claude or Gemini, nearly 100% of the electricity required to run the CPUs in a data center is converted to heat. Today, this heat is generally dumped into the atmosphere through cooling towers, at enormous cost.
Meanwhile, massive efforts are deployed across Europe to renovate buildings and reduce their heat demand. A study from Aalborg University estimates that buildings will need "only" 2,338 TWh by 2035 following current renovation rates.
If data center operators and district heating operators worked hand in hand, data centers could cover up to 10% of European buildings' heat demand by 2035, creating a clear business win for both sides.
The Business Case for Data Center Heat Reuse
The problem is straightforward. Data center currently operators pay twice for energy: once to power servers, then again for cooling. Meanwhile, district heating utilities struggle to find affordable low-carbon heat sources to meet decarbonization targets and increase margins.
Yet for decades, both industries have mostly operated in separate silos, leaving billions in value on the table. The CEO of Finland's largest energy company, Helen, told Bloomberg what data center operators and colocators like Equinix, Telia or Microsoft can expect when approaching utilities:
"Data centers have a problem with heat, and need to cool down their premises. Our job is to sell heat. So the starting point for discussion is pretty good: we can actually monetize their problem." — Olli Sirkka, CEO of Helen
The benefits of reusing data center waste heat for district heating are becoming evident.
For data center operators, cooling represents on average 40% of power consumed, and as much saving in electricity bills when taken over by a utility. In some cases, data center operators even generate additional revenue from stable heat commitments, as Microsoft in Finland. More importantly, demonstrating heat reuse avoids permitting delays and compliance fees.
For utilities, data centers offer cheap, predictable 24/7 baseload heat supply that complements other renewable heat sources (biomass, geothermal, industrial waste heat...). A UK study found that waste heat from data centers could be two times cheaper than regulated price caps. It also represents a straightforward way to qualify as "efficient district heating system" under European regulation and avoid fines and penalities.
However, to harvest the benefits, three barriers must be addressed to ensure heat reuse is profitable.
Barrier 1: High-grade waste heat with liquid cooling
Traditional data centers have relied on air cooling, which vents waste heat to the atmosphere at 25–35°C, too low for most district heating networks without additional heat pumps.
With the rise of AI, rack power densities have jumped from ~5–15 kW in traditional servers to hundreds of kilowatts per rack in future AI configurations.
To survive, the industry is pivoting to liquid cooling, which creates a massive opportunity for the heat sector: instead of lukewarm air, data centers produce high-grade hot water over 45°C requiring marginal effort to pump through a district heating network.
Today, air cooling still accounts for more than half of the global market, but this share is declining. In 2026 already, liquid cooling should already dominate new AI and hyperscale builds reaching 72%.
Barrier 2: Proximity of heat demand
Even with high-grade heat, the distance between rack and radiator is a critical factor in project viability. Identifying the right location is a complex "spatial matchmaking" exercise that requires high-quality geospatial data and specialized thermo-hydraulic modeling software. Without these tools, spotting profitable opportunities early on is prone to data risk and lengthy manual analysis.
Leading industrial players are now reframing this challenge, treating waste heat not as a byproduct to be discarded, but as a value-adding product that should be deliberately routed to existing demand. This shift fundamentally changes the economics of the facility, as Drew Turner from Danfoss explains:
"Any time you reject heat to somebody reusing it, that is free cooling for the data center. You're no longer using power to conduct heat rejection other than getting it from point A to point B." — Drew Turner, Danfoss
Barrier 3: Stable demand throughout the year
Data centers operate steadily all year round. To fully benefit from free cooling via district heating networks, they must ensure there is sufficient heat demand throughout the year.
While space heating demand peaks in winter, buildings have a baseload heat demand for domestic hot water (washing, cooking...), which typically represents around 15–25% of total demand. This baseload can absorb a significant share of data center waste heat, but only if seasonal load curves are properly anticipated. Early-stage planning must therefore ensure an adequate mix of building uses and assess economic viability in scenarios where supplementary cooling is required in summer.
Colder weather also has a benefit on cooling needs. As grid congestion already limits data center developments in major hubs like Frankfurt, London, Amsterdam, Paris, and Dublin (the "FLAP-D" markets), Ember analysts predict a diversification of locations, with a slight advantage for colder countries.
In non-urban areas, it is important to consider also alternative heat sinks like greenhouses, swimming pools or aquafarming, which may offer more seasonal stability than residential buildings.
Regulatory Tailwinds Boost Heat Reuse
European regulation
The EU's Energy Efficiency Directive is the primary regulatory framework governing data center energy use. At its core, the directive aims to improve the efficiency of data centers through reporting and performance standards.
Concretely, as of 2025:
all data centers over 500 kW must annually report a dozen key performance indicators detailed in the Delegated Regulation (EU) 2024/1364.
all data centers over 1 MW must reuse their waste heat, unless they can demonstrate it is technically or economically unfeasible through a cost-benefit analysis
Two main KPIs drive the data center heat reuse by district heating networks.
Power Usage Effectiveness (PUE) is the industry standard metric for data center efficiency:
PUE = Total Facility Power (IT, cooling, lighting...) ÷ IT Equipment Power
A PUE of 2.0 means half the electricity goes to cooling and infrastructure; a PUE of 1.2 means only 20% overhead. The global average has improved from 2.5 in 2007 to approximately 1.6 today. The next efficiency gain should come from the adoption of direct liquid cooling technologies, and ICIS forecasts the European average PUE will reach 1.35 by 2035, with values up to 15% lower in northern Europe thanks to lower temperatures.
The Energy Reuse Factor (ERF) measures what proportion of waste heat is captured for productive use rather than rejected to the atmosphere.
ERF = Energy Reused ÷ Total Facility Power
For example, a data center consuming 1,000 MWh of electricity that supplies 200 MWh of waste heat to a district heating operator has an ERF of 20%. This means 20% of its total energy consumption is reused rather than wasted.
Country-specific regulation
Member states are entitled to stricter regulations, and Germany has set the gold standard. The Energy Efficiency Act (EnEfG) covers all data centers above 300 kW—roughly 1,000 facilities nationwide.
The German law provides clear targets:
Category | Requirements for facilities over 300 kW |
Power Usage Effectiveness (PUE)
| New facilities:
Existing facilities:
|
Energy Reuse Factor (ERF)
| New facilities:
Existing facilities (over 1MW):
|
Renewable Energy
| New and existing facilities
|
Beyond performance targets, Section 17 of the EnEfG also obliges operators to provide detailed waste-heat data to district heating companies upon request, ensuring utilities have the technical visibility needed to design data center-powered networks.
Failure to comply is costly: violations can result in fines up to €100,000, while reporting failures can reach €50,000.
How Urbio Accelerates Data Center Heat Reuse and Compliance With District Heating Planning
Both data center operators and district heating utilities face the same challenge: identifying viable heat reuse opportunities before committing resources. Urbio's early-stage planning software bridges this gap by turning weeks of manual analysis into minutes of automated diagnosis, providing immediate clarity for the most frequent stratic questions.
Data Center Operators frequently ask:
"Is there enough heat demand near my data center to make reuse profitable?"
Urbio's plug-and-play geospatial data instantly maps surrounding buildings, heat density, and seasonal demand profiles within your target radius. In one session, you can assess whether nearby heat demand justifies investment—or whether your site faces summer cooling gaps that require alternative solutions.
"Where should I locate my next facility to ensure there is a buyer for my waste heat?"
While local grid capacity is a key criterion for new data centers, factoring in heat reuse potential early on can pay off down the road. With Urbio's location intelligence and digital twin, screen entire regions to find the sweet spot where grid capacity meets high heat demand and existing pipe infrastructure.
"Can I reach my 20% Energy Reuse Factor (ERF) target with buildings within 500m of my site?"
Get a quick techno-economic assessment of how much heat could be supplied within a cost-effective range. Easily conduct sensitivity analyses by simulating different scenarios with varying perimeters and future heat demand trends.
"How do I demonstrate technical feasibility to avoid permitting delays and regulatory fines?"
Generate preliminary feasibility reports with a few clicks. Urbio's automated reporting features help you document heat reuse potential for regulators and satisfy EU mandates for facilities over 1 MW.
Utilities and District Heating Developers are asking:
"Where are the data centers in my territory, and which ones have real heat reuse potential?"
Urbio's digital twin overlays data center locations with heat demand hotspots, identifying profitable matches in your service area. No GIS expertise required—your commercial team can run the analysis themselves and prioritize outreach to operators with the strongest business case.
"How can I quickly evaluate if a data center partnership will help us meet efficiency targets?"
Urbio's AI-powered design tools let you model integration scenarios—baseload contribution, seasonal variation, network extension costs—and compare techno-economic outcomes side-by-side. This enables you to approach data center operators with concrete pricing proposals, rather than exploratory conversations.
"How many new customers can I connect if I integrate this 10 MW waste heat source?"
Generate and compare multiple network expansion layouts to find the most cost-effective routing, and which customers to prioritize in Urbio's spatial CRM so your commercial team engages in menaingful conversations.
"How much CO2 will we save by replacing our gas peak boilers with data center waste heat?"
Urbio's automated feasibility studies compute the share of renewable energy and carbon intensity of operating your network. Proceed confidently with profitable projects that drive your internal impact targets and help you win more customers.
Case Study: Connecting a 1.7 MW Data Center to a Local District
Let’s walk through a typical data center heat reuse project in the Frankfurt outskirts.
The goal of the utility is to assess whether there is sufficient heat demand nearby, which buildings to target, and is the business case profitable. Urbio supports each step of the project.
Step 1. Spatial matching
First, locate the data center on the map. Note its power and waste heat temperature output, in this case, a 1.7MW data center with liquid cooling rejecting 45°C water.
Then explore nearby heat demand. Look out for: high heat densities, diversity of building types and proximity to the data center.
Step 2. Network design
Once a site is identified, Urbio optimizes the pipe layout from data center to district, and automatically sizes the pipes, heat exchangers and heat pumps.
Step 3. Scenario analysis
Assess the techno-economic feasibility of your project with Urbio's computed metrics, charts and tables, including CAPEX, OPEX, cost price of heat, required peak demand, load duration curves and total supplied heat, which allows to compute the data center's Energy Reuse Factor (ERF).
Finally, sensitivity analyses—such as simulating a 50% building connection rate—allow the team to de-risk the project with multiple scenarios before committing to execution.
Conclusion
As the race for AI compute intensifies, Urbio allocates a small fraction of that very same power to design more sustainable data centers.
The goal is to ensure that every kilowatt-hour works twice: once to power the digital economy, and again to keep our cities warm. By treating data centers as "thermal power plants" and planning them carefully, we can heat up to 10% of Europe's buildings by 2035.
The conditions are finally in place:
Liquid cooling is making waste heat higher-grade and easier to integrate into heating networks.
Planning software now leverages AI to to match heat supply and demand in space and time.
Proactive regulation promotes profitable heat reuse through transparency and new standards.
The result is a paradigm shift that extends beyond data centers: a new mindset where waste heat—regardless of its source—is no longer a liability to be rejected, but a strategic asset to be mapped, modeled and monetized.