Data Centres and Heat Networks: Expectations and Reality

There is strong momentum in the UK heat network sector around the idea that data centres could become major contributors to low-carbon heat.  After all, data centres operate continuously and release huge amounts of heat - why wouldn’t we use it? 

But as a heat network developer who also understands the technical pressures on the data centre industry, I see a significant disconnect between: 

·       The expectations of heat network developers; & 

·       The operational and commercial realities faced by data centre developers. 

There is clearly opportunity, but not alignment. 

Eye-level view of a modern heat network facility

This brief article covers six themes: 

1.     Expectations heat network developers have for data centres 

2.     The reality data centres face before heat offtake is even considered 

3.     Challenges data centre developers face when heat offtake is added 

4.     The evolving data market and the pace of cooling change 

5.     The relevance of PUE - and how heat networks can improve it 

6.     Where heat network developers can have the most impact with the least disruption. 

1. Expectations Heat Network Developers Have for Data Centres 

From a heat network perspective, data centres appear to be ideal heat sources: 

·       They operate 24/7 

·       They produce vast quantities of heat 

·       They can be close to areas of high heat demand 

·       Their waste heat is “always on” and predictable 

·       Their heat aligns with net zero. 

This leads to an assumption that: 

·       Data centres should provide waste heat 

·       They should design cooling for heat reuse 

·       They should co‑locate with heat networks 

·       They should prioritise external heat customers 

·       They should be mandated through planning or zoning frameworks. 

These assumptions come from a genuine desire to accelerate decarbonisation, but they often overlook one central reality: 

·       A data centre’s purpose is not to be a heat generator; it is to provide resilient digital services. 

2. The Reality Data Centre Developers Face - Before Heat Offtake Is Even Mentioned 

Data centres face major technical and regulatory hurdles before heat reuse is on the table.  The industry is under immense pressure. 

A. Grid capacity and TM04+ queue reforms 

Grid access is currently the single biggest constraint for data centres.  Developers must navigate: 

·       Transmission queue reforms (TM04+) 

·       Harder “readiness tests” to secure grid slots 

·       Substation capacity shortages 

·       Long reinforcement lead times 

·       Batch‑based connection approvals 

·       Constraints on fault‑level headroom. 

Even a modest increase in electrical load - such as adding a heat pump - can jeopardise connection dates. 

B. Rising power density 

AI and high‑performance computing have increased rack power density dramatically.  Servers that once consumed 5–10 kW per rack now exceed 40–80 kW, with some systems heading beyond 100 kW.  This places extreme stress on: 

·       Cooling systems 

·       Electrical distribution 

·       UPS and backup infrastructure 

·       Floor loading and layout. 

Cooling becomes an engineering bottleneck long before heat reuse is considered. 

C. Cooling complexity 

Cooling infrastructure is shifting rapidly from traditional air-cooled systems to water‑based solutions.  Operators need flexibility to upgrade cooling as workloads evolve.  Anything that locks them into a specific architecture introduces risk. 

D. Uptime at all costs 

Tier III and Tier IV standards demand highly resilient systems with: 

·       Redundant plant 

·       Compartmentalised cooling 

·       Diesel backup 

·       Independent power feeds. 

Heat export systems cannot undermine this.  This single factor dominates every conversation. 

E. Market growth pressures 

Data centres are building at unprecedented speed.  Operators focus on delivering compute capacity quickly; extra obligations slow this down.  For heat reuse to succeed, we must understand these pressures.  Data centres are not resisting heat reuse - they are prioritising fundamental operation and service delivery. 

3. The Additional Challenges Data Centres Face When Heat Offtake Is Added 

When we overlay heat recovery, further complexities emerge. 

A. Cooling system constraints 

Heat offtake interacts directly with cooling mode and configuration.  Air‑cooled data centres provide low‑grade heat and will not dominate future design.  Modern water‑based systems (rear‑door, direct‑to‑chip, immersion) provide higher and more stable temperatures. But: 

·       Designs are evolving quickly 

·       Operators want freedom to upgrade. 

Heat offtake infrastructure must not constrain future cooling choices. 

B. Reliability and resilience risks 

Heat networks need consistency, data centres experience variability due to: 

·       IT load changes 

·       Climate conditions 

·       Cooling mode changes 

·       Server upgrades 

·       Phased expansions. 

Data centres fear being held to heat supply obligations that create operational risk. 

C. Commercial and energy risks 

Heat export generally requires: 

·       Heat pumps 

·       Upgraded pipework 

·       Additional plant rooms 

·       More monitoring and maintenance. 

These add cost and electrical load - directly competing with compute. 

D. Air cooling becoming obsolete 

Air‑cooled systems will gradually disappear in high‑density environments.  This is good news for heat networks - but only if we align with water‑cooled designs rather than trying to retrofit air‑based systems. 

4. The Evolving Data Market and Cooling Landscape 

The data centre industry is undergoing radical change: 

·       AI workloads are surging 

·       Chip power densities are increasing rapidly 

·       Liquid cooling is becoming mainstream 

·       Immersion cooling is emerging for GPU clusters 

·       New build timescales are shortening 

·       Energy efficiency regulation is tightening. 

Cooling systems that were “cutting edge” five years ago are already being replaced.  The challenge for heat reuse is that: 

·       Heat networks operate on 30–50 year horizons. 

·       Data centres redesign themselves every 3–5 years. 

This mismatch makes long-term heat obligations difficult unless heat extraction is designed to be adaptable to future cooling systems. 

5. The Relevance of PUE - and How Heat Networks Can Help Improve It 

PUE (Power Usage Effectiveness) remains the dominant energy metric in the data centre industry. It compares: 

Total Facility Energy versus IT Energy 

Depending on allocation, heat pumps and heat recovery equipment increase facility energy use (by adding electrical overhead), which can make PUE appear worse - even if the overall environmental outcome is better.  This is crucial for two reasons: 

·       Operators optimise around PUE because it affects cost, efficiency benchmarks, and investor confidence. 

·       Improving PUE matters more to them than selling heat. 

This is where the heat network industry can help. 

·       Take on technical risk: Provide, maintain, and operate heat pumps and heat extraction plant.  This removes electrical overhead from the data centre’s scope. 

·       Take on commercial risk: Guarantee certain performance levels or provide availability‑based payments to de‑risk the investment. 

·       Package solutions as PUE‑neutral or PUE‑improving: If heat networks supply all the extraction plant’s power or offset it through commercial arrangements, the data centre avoids a PUE penalty.  

This is a major opportunity; heat networks can offer value by improving a KPI the data centre already cares about. 

6. Where Heat Networks Can Have the Most Impact With the Least Disruption 

Here is the opportunity space where heat networks can genuinely collaborate with data centres. 

A. Focus on low‑impact, low‑risk integration with existing water‑cooling 

Most large data centres already have: 

·       Chilled‑water pre‑cooling circuits 

·       Glycol loops 

·       Heat rejection via cooling towers or dry coolers 

·       Sufficient pumping and redundancy. 

These circuits offer the cleanest interface for heat extraction with minimal operational impact. 

Rather than tapping into downstream IT cooling loops (high‑risk), heat networks can integrate with: 

·       the condenser water loop 

·       the recirculating water loop 

·       the pre‑cooling circuit before the cooling towers. 

This is similar to the integration concepts highlighted in several engineering studies; take heat from the water circuit before it is rejected. 

B. Capture heat while remaining cooling‑agnostic 

If heat networks draw heat from upstream water circuits, the solution becomes: 

·       modular 

·       repeatable 

·       compatible with both air‑assisted and fully liquid‑cooled systems 

·       agnostic to future cooling upgrades. 

As liquid cooling and immersion cooling grow, operators will continue to rely on some form of water‑based heat rejection - giving heat networks a stable, long‑term integration point. 

C. Lock in long-term certainty (without constraining the data centre) 

By securing rights to extract heat from the pre‑cooling circuit - not from IT equipment directly - heat network developers can: 

·       avoid interfering with mission‑critical cooling 

·       ensure long-term heat availability 

·       adapt to new cooling technologies 

·       provide a stable heat supply to the network. 

This is a low‑risk pathway for both sides. 

D. Aligning with the data centre’s long‑term PUE goals 

Because upstream heat extraction: 

·       reduces cooling tower use 

·       reduces fan speeds 

·       lessens mechanical rejection energy 

·       can offset other cooling loads 

…it can contribute to lower long‑term PUE in a way that is meaningful to operators.  This is the strongest value proposition heat networks can offer. 

Conclusion: A Path Forward Based on Alignment, Not Obligation 

Heat reuse from data centres offers a major decarbonisation opportunity.  But the solution must fit the realities of data centre operation - not impose additional burdens or risks. 

By focusing on: 

·       water‑based cooling integration 

·       low‑risk pre‑cooling extraction points 

·       solutions that evolve with cooling technology 

·       and commercial models that support PUE improvements 

·       heat network developers can create partnerships that work for both industries. 

This is where real progress will be made - not through expectation or obligation, but through alignment of engineering, incentives, and long‑term operational reality.