Water/wastewater
The European Commission's draft regulation published in March 2026, proposing an EU-wide rating scheme for data centres, signals a shift from internal engineering metrics to public comparative assessment.
For environmental monitoring professionals, the important question is not primarily about electricity bills or carbon targets. It is about whether the measurement and reporting systems that underpin those ratings are technically credible.
The current state of EU regulationThe regulatory context matters. In March 2024, the Commission introduced legislation creating a European database on data centres and a systematic reporting scheme for key performance indicators.
The 2026 rating scheme is the next layer – using that reported data to generate ratings that will be electronically labelled and publicly accessible.
The scheme is explicitly designed to inform both procurement decisions and policy-making. A hyperscale cloud provider's data centre would carry an EU performance rating visible to corporate customers buying cloud services.
That turns sustainability data from an internal metric into a commercial differentiator.
Power Usage Effectiveness (PUE) is the most established metric in the sector and is already reasonably well understood. A PUE of 1.0 would mean every watt drawn from the grid goes directly to computing; real-world figures for efficient modern facilities run from about 1.1 to 1.3, with older facilities sometimes significantly higher.
But PUE has known limitations: it captures the efficiency of cooling and power distribution relative to IT load, but says nothing about the carbon intensity of the electricity consumed, the renewable fraction or how the facility interacts with the grid during peak demand periods.
A facility with a PUE of 1.2 running entirely on coal-backed grid power has a very different environmental profile from one with a PUE of 1.4 running on dedicated renewable generation with grid interactive demand management.
Water consumption is the metric where data centres face the most scrutiny growth. Evaporative cooling systems, which are highly energy-efficient, use large volumes of water.
Hyperscale data centres can consume millions of litres of water per day. In water-stressed regions, this creates genuine resource competition with agricultural, domestic and ecological uses.
Water Usage Effectiveness (WUE) – litres of water consumed per kWh of IT load – is the emerging standard metric but the metrology is not yet consistent across operators.
What counts as water consumed versus water used and returned? How are water sources accounted for – mains supply, recycled water, rainwater harvesting, cooling tower blowdown? How are indirect water uses from electricity generation included or excluded? These are metrology questions as much as engineering ones.
Heat reuse is the sustainability story that data centres are most eager to tell but struggle most to measure convincingly. Waste heat from server rooms can theoretically be recovered and exported to district heating networks, industrial processes or nearby facilities.
Several large European installations are doing this at scale. But quantifying the heat exported, demonstrating its quality, accounting for seasonal variation, and connecting it to a verified reduction in total system energy use requires metering infrastructure and data management that many facilities do not currently have.
For a rating scheme that rewards heat reuse, the measurement protocol needs to be robust enough that operators cannot claim heat reuse credit without demonstrable delivery.
Renewable energy accounting is another area where the gap between narrative and verifiable evidence is often wide. Purchasing renewable energy certificates guarantees the renewable provenance of electricity statistically – it does not mean the electrons powering the servers were generated renewably at the moment of consumption.
For a rating scheme to meaningfully distinguish between facilities with genuine renewable supply and those using certificate purchases to offset baseload fossil consumption, it needs to assess temporal matching, geographic proximity, additionality and grid interaction.
These are technically demanding assessments that the monitoring sector has the skills to support, but they require standardised methodologies that do not yet fully exist for the data centre context.
For instrumentation and monitoring suppliers, the commercial opportunity is in helping data centres build the metering and data management infrastructure that ratings will require.
That means sub-metering to IT load level, cooling system flow and temperature measurement, water flow monitoring, renewable energy delivery metering, heat export instrumentation and integration of these data streams into sustainability reporting platforms that can generate auditable outputs.
Facilities that have been monitoring only at the incoming utility meter level will need significant investment in monitoring granularity.
The broader significance is that the digital sector is now being treated as a general-purpose infrastructure service.
As AI model training, cloud computing, data storage and edge computing demand expand, the energy and water intensity of digital infrastructure will become harder to minimise or explain away.
Operators who get ahead of the rating scheme by building robust measurement systems now will be better placed when the ratings are public and procurement teams start asking questions. Those who do not will find that sustainability marketing without credible underlying data is an increasingly untenable position.
IET 36.3 May
Product directory
