The European regulatory framework governing Data Center energy efficiency is entering a new phase. The article we published on DataCenter Forum in March 2025 told a relatively straightforward story: the European Union was introducing mandatory reporting requirements for Data Centers. Operators with more than 500 kW of installed IT power were required to submit data on energy and water consumption.

The developments introduced in 2026 take this framework one step further: the same data will now underpin a common rating system, based on standardized indicators and automatically generated electronic labels.

The new scheme proposed by the European Commission on 21 September is technically significant because it introduces a common set of metrics for assessing Data Center performance across Europe. Until now, operators have reported data on energy consumption, water use and infrastructure performance. The next step is to turn this data into comparable indicators that will also be publicly visible.

According to the timeline presented by the European Commission, the first electronic labels are scheduled to be issued on 15 September 2027 and will be regenerated annually.

From Measurement to Performance Assessment

Data Center efficiency cannot be assessed solely on the basis of total energy consumption. Two facilities with the same IT capacity may have very different performance profiles depending on the efficiency of their power and cooling infrastructure, the source of the energy they use, their water consumption and their ability to reuse waste heat.

This is precisely why the European framework relies on several indicators.

• PUE – Power Usage Effectiveness measures the ratio between the total energy consumed by a Data Center and the energy actually used by its IT equipment. It is one of the most widely used metrics for assessing the efficiency of physical infrastructure, particularly power and cooling systems.
• WUE – Water Usage Effectiveness measures water-use efficiency and is becoming increasingly relevant as Data Center cooling is more closely scrutinized in relation to the availability of local water resources.
• ERF – Energy Reuse Factor introduces a different perspective: not only how much energy a Data Center consumes, but also how much of the resulting waste energy can be reused.
• REF – Renewable Energy Factor reflects the proportion of electricity supplied from renewable energy sources.

Together, these indicators provide a more comprehensive picture of a facility’s performance than total energy consumption alone.

What Needs to Be Measured Across the Infrastructure

For operators, one of the most important consequences of the new framework is the need for sufficiently accurate and consistent data to calculate these indicators.

This means that monitoring infrastructure becomes an integral part of the Data Center’s operational infrastructure.

Submetering, for example, makes it possible to separate individual energy loads and identify sources of losses or inefficiency. Instead of relying on a single facility-wide metric, operators can monitor the consumption associated with different systems and areas of the infrastructure.

In practice, preparing for the new system involves:

• separately monitoring IT energy consumption and auxiliary infrastructure consumption;
• collecting the data required to calculate PUE and WUE;
• measuring and documenting reused energy;
• identifying the energy sources used and the share supplied from renewables;
• maintaining the historical data required for annual reporting;
• integrating monitoring systems with reporting processes.

Commission Delegated Regulation (EU) 2024/1364 has already established categories of informational, energy, operational and sustainability indicators, and the new framework builds on this foundation.

Cooling Becomes a Strategic Element

One area where the impact of the new requirements may be particularly visible is the cooling system.

As power densities increase, particularly in infrastructure designed for AI workloads, cooling becomes an important factor both in overall energy consumption and in maintaining IT equipment at its intended performance level.

In this context, operators need to view PUE not simply as a reporting metric, but as an outcome of the Data Center’s technical architecture.

Free cooling, optimized operating temperatures, airflow containment, improved chiller and pump efficiency, and the use of liquid-cooling technologies can all significantly alter a facility’s energy profile.

In high-power-density Data Centers, liquid cooling becomes particularly relevant. However, the efficiency of such technology needs to be assessed at system level, rather than solely at the IT equipment level. Liquid distribution infrastructure, heat exchangers, pumps and heat-rejection systems all contribute to total energy consumption.

From PUE to a Multidimensional Assessment

One important consequence of the new system is that a single metric is no longer sufficient to describe the performance of a Data Center.

A low PUE, for example, does not indicate how much water is being used for cooling or whether waste heat is being recovered and reused. Likewise, a high share of renewable energy does not automatically compensate for inefficient energy infrastructure.

By considering PUE, WUE, ERF and REF simultaneously, the assessment becomes multidimensional.

This approach is particularly relevant when designing new facilities. Decisions regarding the cooling system, electrical architecture, electricity sources, waste heat recovery potential and even the location of a Data Center can influence its reported performance over the long term.

The 500 kW Threshold and the Types of Data Centers Covered

The European framework applies to operators of Data Centers with at least 500 kW of installed IT power. It covers several types of facilities, ranging from Data Centers operated in-house by an organization to colocation facilities and multi-tenant environments where several customers share the same infrastructure.

In terms of size, the framework distinguishes between several categories, ranging from 500 kW–1 MW to facilities exceeding 10 MW. Facility size will also form part of the information associated with the rating, allowing the indicators to be interpreted more meaningfully according to the scale of the infrastructure.

This is important because comparisons between two Data Centers need to take into account their architecture, size and the type of workloads they host.

For those wondering how many such facilities there are in Romania, estimates indicate approximately 25–30, enough for the new requirements to have a tangible impact on the local market.

Reused Energy: The Next Step in Efficiency

ERF brings into focus an aspect that has so far received less attention in the operational assessment of Data Centers: what happens to the thermal energy generated by the infrastructure.

In a conventional Data Center, a significant proportion of the energy consumed ultimately becomes heat that needs to be rejected. From an overall efficiency perspective, however, this energy can also represent a valuable resource.

Connecting a Data Center to local district heating networks or using the recovered heat for other applications can improve the overall energy utilization of the infrastructure. Waste heat recovery and reuse should therefore be considered as early as the design stage, rather than treated solely as a subsequent optimization project.

Renewable Energy Does Not Replace Infrastructure Efficiency

REF introduces another dimension into the assessment: the source of the electricity used.

For operators, access to renewable energy and the use of mechanisms such as long-term Power Purchase Agreements (PPAs) can contribute to improving the facility’s energy profile. However, this does not remove the need to optimize the underlying infrastructure.

A Data Center that consumes less energy for the same workload has an inherent structural advantage, regardless of the source of that energy. Infrastructure efficiency and the decarbonization of the power supply should therefore be treated as complementary objectives.

Timeline: 2026 Is the Year for Technical Preparation

For operators, the period between the development of the framework in 2026 and the first generation of electronic labels in September 2027 is relatively short.

The timeline is important:

• 2023: the Energy Efficiency Directive establishes the general framework;
• 2024: Commission Delegated Regulation (EU) 2024/1364 introduces reporting requirements for Data Centers with at least 500 kW of installed IT power;
• March–April 2026: public consultation on the common rating scheme;
• May 2026: publication of the revised technical version;
• 2026: implementation of the technical framework and reporting systems;
• 15 September 2027: first automatic generation of electronic labels;
• 31 March 2029: first review of the indicators and performance targets.

For operators, the technical message is straightforward: preparation does not begin when the label becomes public. The data underpinning the label must already have been measured, validated and collected beforehand.

What Operators Should Review

Preparation for the new framework can begin with a technical assessment of the existing infrastructure.

First, operators need to determine whether all the required metering points are in place and whether the collected data can be correlated to calculate the relevant indicators.

Second, data quality needs to be assessed, including measurement frequency, consistency, historical records and auditability.

Third, the results need to be linked back to the physical infrastructure. PUE is an indicator; improving it requires changes to the systems that drive energy consumption, including cooling, electrical distribution, UPS systems, pumps, fans and other auxiliary systems.

Against this background, a technical preparation program should include at least the following:

1. Auditing the metering system and identifying areas where submetering is missing.
2. Validating the calculation methodology for PUE, WUE, ERF and REF.
3. Establishing a baseline for current performance.
4. Identifying the main sources of auxiliary energy consumption.
5. Assessing the cooling system and opportunities for improving its efficiency.
6. Analyzing opportunities for waste heat recovery and reuse.
7. Evaluating renewable energy sources and energy procurement agreements.
8. Automating data collection and reporting wherever possible.

What Changes for Data Centers in Romania?

For the Romanian market, the challenge goes beyond complying with a European requirement. It lies in operators’ ability to turn operational data into verifiable and comparable technical information.

Data Centers that already have mature DCIM, submetering and energy-monitoring systems in place start from a stronger data infrastructure position. For facilities that still rely on aggregated measurements, the period leading up to 2027 may become an important phase of modernization.

At the same time, the expansion of cloud and AI infrastructure is placing additional pressure on efficiency. Increasing power densities, greater cooling requirements and increasingly demanding power-supply needs mean that design decisions made today can influence operational performance indicators for many years.

From this perspective, the new European system can be viewed less as a simple reporting obligation and more as a framework for standardizing Data Center performance.

A New Definition of Performance

For the European Data Center industry, the transition from reporting to rating represents an important paradigm shift. Until now, the main question was whether an operator could provide the required data. The next question will be what that data reveals about the performance of the infrastructure.

PUE, WUE, ERF and REF will provide a more granular view of how a Data Center uses energy and resources. For operators, this means that efficiency needs to be addressed simultaneously across electrical infrastructure, cooling, water use, waste heat recovery and energy sourcing.

For new projects, these criteria need to be integrated from the design stage. In an industry where computing capacity is expanding rapidly, efficiency is no longer simply an operational characteristic. It is becoming an integral part of infrastructure design, investment decisions and the competitiveness of digital infrastructure.

The first electronic labels are scheduled for September 2027. For operators, however, the most relevant point in time is not 2027 itself, but the preparation period leading up to it.