In brief
Waste heat has commercial value where available heat, temperature and operating hours match a buyer’s network. A heat pump may be needed. The connection must include clear metering, operating responsibility and fallback heat rejection so IT cooling remains available.
In practice all the electricity fed into a data center leaves it as heat. A one megawatt site produces a megawatt of thermal output that has to go somewhere. The question is not whether heat is produced, but whether it is rejected to ambient air or sold.
In Finland the second option has become more interesting than it was. Data center electricity moved from the reduced tax class II to the general class I on 1 July 2026, which raised operating costs appreciably. Heat sold into a district heating network is a revenue stream that offsets part of that. At the same time the EU Energy Efficiency Directive requires sites above 1 MW to make use of their waste heat where this is technically and economically feasible. The wider picture is in our complete guide to data center cooling.
What a buyer should provide
For a feasibility discussion, share hourly heat and temperature profiles, district heating operator requirements, the connection point, metering boundary and fallback cooling arrangement. Finkova can discuss the pump, valve, piping and instrumentation scope.
Temperature level decides everything
Whether recovery is possible depends on one thing above all others: how hot the heat leaves the site.
In an air-cooled hall the return temperature is low and the heat is spread through the whole volume of hall air. Collecting it for reuse is laborious and inefficient. Liquid cooling changes this, because the heat is concentrated into one loop at a higher temperature. The server loop return is appreciably warmer than the exhaust air of an air-cooled hall.
A district heating network, however, requires a higher temperature still. The supply temperature of a Finnish network varies with the season and is considerably higher in winter than in summer, while the return temperature is closer to what a data center produces. An industrial heat pump is therefore needed in between, lifting the temperature to the level the network requires.
One of the most important early decisions on a project follows from this. The higher the ASHRAE temperature class chosen for the site, the smaller the lift required of the heat pump and the better the efficiency of the recovery. The temperature class is therefore not only a cooling decision but also a decision about the value of the waste heat. The classes are covered in the article on TCS and FWS: data center cooling loop architecture and scope splits.
Two ways to connect
Heating the return line. The heat is fed into the return side of the district heating network. The required lift is smaller, so the heat pump is smaller and its efficiency better. The price paid for the heat is lower, however, and the capacity to absorb it depends on the return flow at that point in the network.
Heating the supply line. The heat is lifted to the network’s supply temperature. The lift is larger and the heat pump more expensive, but the heat commands a higher price and the absorption capacity is not limited in the same way.
Which suits depends on the structure of the network and where in it the data center sits. This is a question the network owner answers, not the designer.
What the economics rest on
The viability of recovery rarely comes down to heat pump efficiency alone. The decisive factors are:
- distance to the district heating network and the cost of building the connection
- the network’s temperature level and its seasonal variation
- the price paid for the heat and the length of the contract
- summer demand, which in a district heating network is a fraction of winter demand
- the heat pump’s electricity consumption and its price
- who owns the heat pump and carries the investment
Summer is the most underestimated of these. A data center produces heat steadily all year, while a district heating network buys it mainly during the heating season. The site therefore has to be able to reject its entire thermal load without recovery in any case, meaning the dry coolers are sized for full load regardless of whether heat is sold. Recovery is an addition, not a substitute.
Availability is an absolute constraint
This is the most important point in the article, and it belongs in the contract.
Data center cooling must never depend on the sale of heat. If the district heating network stops accepting heat, if the heat pump fails, or if the connection is under maintenance, the site has to carry on as normal. Recovery must therefore be bypassable automatically and without interruption, and the changeover between recovery and heat rejection has to happen fast enough.
In practice this means the recovery loop gets its own isolation and control valves, the bypass is designed in from the start, and the changeover is tested for real at commissioning rather than on paper. The commissioning stages are covered in the article on data center cooling: commissioning and handover.
The same applies to the contract. A heat supply agreement should not carry a delivery obligation that cannot be met without putting cooling availability at risk.
Metering and billing
Heat sold is measured with a thermal energy meter consisting of a flow meter, a matched pair of temperature sensors and a calculator. The European product standard is EN 1434, and for billing use the meter is typically required to hold type approval under the Measuring Instruments Directive 2014/32/EU.
In Finland the measurement of district heating and district cooling is covered by the current Finnish Energy metering guidance. It is a natural starting point, but the final requirements come from the buyer, meaning the network owner, and they are worth establishing before the metering is designed. The implementation is covered in the article on flow and energy metering in data center cooling loops.
A small but expensive detail: billing measurement and the site’s own energy monitoring are different things with different requirements. The same meter may not serve both, and that is worth noticing before the metering points are fixed.
Common misjudgements
Recovery is sized to replace heat rejection. It cannot, because in summer the network does not buy.
The temperature class is set low as a precaution. This sinks the economics of recovery before a heat pump has even been specified.
The network owner is consulted too late. The connection point, temperature level and absorption capacity are properties of the network, not choices the data center makes.
The availability risk is not assessed. The contract obliges heat delivery in exactly the situation where the site should be concentrating on cooling itself.
Metering is designed for monitoring rather than billing. A meter replaced afterwards is expensive and requires opening the loop.
Finkova’s scope and next step
Finkova installs and commissions data center cooling in the Nordics, with instrumentation and lifecycle support. The work scope, OEM requirements and acceptance criteria are agreed for each project.
Contact Finkova with the project inputs above to discuss the relevant installation, testing and handover interfaces.
Frequently asked questions
Can data center waste heat be sold to district heating?
Yes. In practice all the electricity fed into a data center becomes heat, and liquid cooling recovers it at a temperature high enough to be useful. A district heating network still needs a higher temperature, so an industrial heat pump lifts the heat to the network’s level.
Why is waste heat topical in Finland now?
Data center electricity moved from the reduced tax class II to the general class I on 1 July 2026, which raised operating costs appreciably. Selling heat into district heating is a revenue stream that offsets this. The EU also requires sites above 1 MW to reuse waste heat where technically and economically feasible.
Is a heat pump needed for waste heat recovery?
In practice, yes. Server loop return water is considerably warmer than the exhaust air of an air-cooled hall, but district heating supply requires a higher temperature still. An industrial heat pump lifts the heat to the network’s level. The higher the data center’s temperature level, the smaller the lift needed.
Is heat fed into the supply or return side of district heating?
Either is possible. Feeding the return side needs a smaller lift and a smaller heat pump, but the heat commands a lower price and absorption capacity is limited. Feeding the supply side costs more but earns more. The network owner decides which suits.
What does the viability of heat sales depend on?
Distance to the network, the network’s temperature level and seasonal variation, the heat price and contract length, summer demand, the heat pump’s electricity consumption, and who owns the heat pump. Heat pump efficiency alone rarely decides the question.
Can waste heat recovery replace data center cooling?
No. A district heating network buys heat mainly in the heating season, and summer demand is a fraction of winter demand. A data center produces heat steadily all year, so dry coolers are sized for full load regardless. Recovery is an addition, not a substitute.
What happens if the district heating network stops accepting heat?
The data center has to carry on as normal. Recovery must be bypassable automatically and without interruption, and the changeover between recovery and heat rejection must be tested at commissioning. Cooling availability must never depend on the sale of heat.
How is sold heat measured?
With a thermal energy meter consisting of a flow meter, a temperature sensor pair and a calculator. The European standard is EN 1434, and billing use typically requires MID type approval. In Finland the current Finnish Energy metering guidance covers district heating metering, but final requirements come from the network owner.
How does the temperature class affect the value of waste heat?
The higher the ASHRAE temperature class chosen for the site, the warmer the heat leaves and the smaller the lift required of the heat pump. This improves recovery efficiency. A temperature class set low as a precaution can sink the economics of recovery before a heat pump has even been specified.
What should a heat supply agreement cover?
The delivery obligation has to be worded so that meeting it never puts cooling availability at risk. The agreement should also cover the connection point, temperature level, metering and billing basis, and who owns the heat pump and carries the investment. The network owner is best engaged at design stage.
