Europe is redefining how datacentre efficiency and sustainability are measured. Through the revised Energy Efficiency Directive (EED), and informed by a technical assessment conducted by the Borderstep Institute — an independent sustainability research organisation — in collaboration with EY and the Austrianian Institute of Technology (AIT), the EU is moving away from single-metric evaluation toward a multidimensional performance framework.
The Borderstep assessment was commissioned to evaluate how the EED’s new datacentre reporting obligations work in practice. It analysed early reporting data from across the EU, reviewed how operators interpret and apply the required metrics, and assessed whether these indicators are suitable for transparency, comparability, and future policy use. In doing so, it provides the first evidence-based picture of how European datacentres perform under the new framework, as well as the practical challenges of implementation.
For more than a decade, Power Usage Effectiveness (PUE) served as the dominant benchmark for datacentre efficiency. While useful, the Borderstep assessment confirms what many operators and regulators have long understood: PUE alone does not capture the full environmental footprint of modern facilities. It does not reflect water consumption, carbon intensity, renewable energy sourcing, or the reuse of waste heat.
The revised EED formalises this broader perspective. It requires qualifying datacentres to report a standardised set of metrics covering energy efficiency, water usage, renewable energy integration, and the reuse of waste heat. Together, these indicators are intended to improve transparency, support comparability, and form the basis for future sustainability classification schemes.
Why Europe Is Moving Beyond PUE
The EED was introduced because digital infrastructure has become too important — and too resource-intensive — to operate without oversight. AI workloads are accelerating energy demand, traditional air-cooling systems are placing increasing pressure on water resources, and public authorities are seeking clearer evidence that operators are managing these impacts responsibly.
Under the Directive, datacentres above a defined capacity threshold must submit detailed performance data to a central EU database. The objective is not to mandate specific technologies, but to establish transparency and enable comparability. Over time, this data is expected to support benchmarking and potentially underpin a European sustainability rating scheme comparable to energy labels used for buildings or appliances.
The Borderstep assessment shows that this approach is both feasible and necessary. Early reporting cycles reveal three consistent patterns: participation is incomplete, data quality varies, and performance differences between facilities — even within the same region — are larger than anticipated. These findings reinforce the need for standardised metrics that reveal underlying trade-offs rather than headline efficiency figures.
From Metrics to Design: What the EED Means for Datacentre Decisions
The purpose of sharing and analysing early EED reporting results is not only to critique data quality, but to ensure that operators, investors, and technology buyers understand the full set of criteria against which datacentres will increasingly be assessed.
The EED makes clear that future compliance and competitiveness will be shaped at the design and technology-selection stage, not retroactively through reporting alone. Decisions that optimise one metric at the expense of others — a common outcome under PUE-centric design — are increasingly misaligned with regulatory expectations.
For datacentre operators and those responsible for procurement or design decisions, this implies a shift from single-objective optimisation to multi-criteria system design.
Designing for Multiple Metrics, Not a Single Score
Under the EED framework, no single metric determines performance. Instead, evaluation increasingly focuses on how trade-offs are managed across energy efficiency, water use, renewable integration, and heat reuse.
In practice, this means:
• Low PUE achieved through high water consumption will be subject to scrutiny.
• Facilities without a credible heat reuse pathway may face challenges in urban or district-heated regions.
• Renewable sourcing strategies must align with operational load profiles, not only contractual procurement.
• Technologies that simplify measurement and reduce variability will become more attractive as reporting matures.
Datacentres designed narrowly around legacy assumptions — particularly around air cooling and climate dependency — may face higher retrofit costs, regulatory friction, or reduced long-term flexibility.
Technology Selection as a Regulatory Risk Decision
Technology choices are no longer purely operational or financial. Under the EED, they increasingly carry regulatory, disclosure, and reputational risk.
When evaluating cooling, power, or infrastructure technologies, operators should assess:
• How the technology performs across all EED metrics, not just PUE
• Whether it introduces structural dependencies, such as water stress or climate sensitivity
• How easily its impacts can be measured, verified, and reported
• Whether it enables or constrains future heat reuse and renewable integration
Technologies that lock facilities into narrow optimisation paths may become liabilities as the EED framework evolves toward benchmarking, labelling, and differentiated permitting or financing conditions.
Practical Guidance for Datacentre Operators and Buyers
To align design and procurement decisions with the EED, operators should consider the following principles:
1. Design for balance, not single-metric excellence
Select architectures that perform well across PUE, WUE, ERF, and REF simultaneously.
2. Eliminate structural trade-offs early
Cooling and power choices largely determine long-term water use, heat reuse feasibility, and reporting complexity.
3. Prefer low-variance technologies
Systems that are less dependent on climate, seasonality, or operating mode reduce regulatory and reporting risk.
4. Treat heat as a design output
Where district heating or industrial reuse is feasible, heat quality and stability should be evaluated alongside IT performance.
5. Simplify compliance through technical clarity
Architectures with fewer interacting subsystems reduce measurement ambiguity and audit complexity.
The Four Core Metrics of the EED Framework
The EED reporting framework is structured around four primary indicators that together describe datacentre performance across multiple dimensions.
Power Usage Effectiveness (PUE) remains central to energy reporting, but is positioned as a technical efficiency indicator rather than a sustainability proxy.
Water Usage Effectiveness (WUE) exposes the water cost of cooling, making visible trade-offs that PUE alone cannot capture.
Energy Reuse Factor (ERF) reflects Europe’s focus on circular energy use and the reuse of waste heat.
Renewable Energy Factor (REF) measures alignment with decarbonisation goals and the EU Taxonomy.
Together, these metrics allow policymakers, investors, and customers to evaluate datacentres holistically.
What the Borderstep Assessment Reveals About Europe’s Current Performance
The Borderstep–EY–AIT assessment provides the first consolidated view of how European datacentres perform against the EED’s reporting metrics. While participation in the initial reporting cycle was incomplete, several trends are already visible.
Recent national publications of first EED reporting results have further exposed implementation challenges. Member States such as Germany and the Netherlands have released aggregated data sets, drawing criticism due to inconsistencies in completeness, transparency, and interpretation.
Germany published the most comprehensive data set to date. The Dutch publication was widely regarded as incomplete, with significant data withheld or anonymised, particularly where organisations classified information as commercially sensitive or self-determined disclosure thresholds. Information related to non-EU operators was frequently excluded.
As a result, these early disclosures have not yet delivered the system-level insights the EED framework is intended to enable. This reinforces the Borderstep conclusion that harmonised implementation and clearer disclosure guidance are essential.
Why Immersion Cooling Aligns Well with the EED Metrics
The EED does not mandate specific cooling technologies. However, its metrics implicitly favour architectures that minimise trade-offs between energy efficiency, water use, heat reuse, and reporting clarity.
Immersion cooling aligns with these requirements due to its underlying thermodynamic characteristics rather than regulatory optimisation.
PUE Performance Independent of Climate
Immersion cooling reduces reliance on ambient conditions by transferring heat directly into a dielectric liquid. This enables low and stable PUE values across diverse climates, reducing regional performance disparities.
Near-Zero Water Consumption
Immersion cooling does not rely on evaporation, resulting in negligible operational water use and effectively zero WUE.
Higher-Quality Heat for Reuse
Heat extracted via liquid immersion is stable and higher-grade, improving feasibility for district heating and higher ERF performance.
Compatibility with Renewable Energy Integration
Predictable thermal behaviour supports renewable integration, warm-water loops, and alignment with REF objectives.
Energy Productivity and Future Metrics
Immersion cooling improves energy productivity by reducing parasitic losses, preventing throttling, and enabling higher compute density — aligning with future EU performance metrics.
Reporting Quality and Measurement Simplicity
Simplified thermal and water flows reduce reporting ambiguity, supporting consistent compliance as scrutiny increases.
Conclusion: Metrics and Architecture Are Converging
The revised EED signals a fundamental shift in how datacentres will be evaluated in Europe. Performance will be judged across energy, water, heat, renewables, and productivity — not a single score.
Immersion cooling aligns structurally with this multidimensional framework. While not mandated, it reduces trade-offs across all measured dimensions and simplifies compliance.
For operators planning new builds, upgrades, or high-density deployments, aligning design and technology decisions with the EED today reduces regulatory risk tomorrow.