Green data center practices

Over the last twenty years, digital technology has advanced rapidly, and data centers are now at the center of this progress. 

They represent crucial pieces of infrastructure that enable services such as cloud computing, AI, e-commerce, streaming services, and business systems. As the amount of data created worldwide continues to grow at an unprecedented rate, data centers have become increasingly crucial in keeping our connected world running.

Therefore, the concept of "green data centers" has evolved from a nice-to-have to a must-have for the industry.

Why go green? The environmental and regulatory drivers

There are two primary reasons why we are moving toward green data centers: the growing environmental impact of digital infrastructure and the increasing pressure from governments and international organizations to adopt this approach.

Growing energy footprint

Data centers consume a significant amount of electricity and water, and they also generate substantial amounts of waste heat and electronic waste (e-waste). Due to the considerable impact on the environment, we must rapidly and comprehensively transform the way we design, operate, and manage them.

According to the International Energy Agency (IEA), global data center electricity consumption is estimated to be around 415 TWh in 2024. Some estimates indicate that this figure could double to 945 TWh by 2030.

European data centers accounted for 46-65 TWh of electricity in 2022. To put this in scale, that is 1.8-2.6% of total EU electricity use.

According to Euro News, the electricity demand from data centers in Europe is expected to surge by up to 160% by 2030, potentially reaching 287 TWh, surpassing Spain's total electricity consumption in 2022.

If this increased electricity demand is primarily met by fossil fuels, annual emissions from newly added data centers in the EU could reach 39 million tons of CO2 in 2030, exceeding the combined annual emissions of Lithuania and Estonia in 2022. Cumulatively, new data centers could emit 121 million tons of CO2-equivalent over six years, a volume equivalent to half of Germany's planned 2030 CO2-equivalent emission cuts across all sectors. 

One of the primary reasons for this increased energy demand is the advancement and widespread adoption of AI. According to the data, training GPT-3 consumed around 1,300 MWh, while training GPT-4 required over 50,000 MWh.

To put those numbers into perspective, a single Google search uses around 0.04 Wh, while according to Sam Altman, the CEO of OpenAI, a single ChatGPT query uses 0.34 Wh.

The recent jump in AI demand is delivering a clear rebound effect across the data center landscape. What's happening is straightforward: every time a new chip or cooling method achieves a better performance-per-watt metric, the additional AI workloads rushing in push the total power draw beyond the efficiency gains, resulting in a larger, not smaller, energy footprint.

This situation suggests that future policy frameworks may need to move beyond solely focusing on improving efficiency and consider strategies for managing computational demand or, more critically, ensuring that all new data center energy requirements are met exclusively by additional, newly generated renewable energy capacity, rather than by diverting existing renewable supplies from other sectors or relying on fossil fuels. Failure to do so risks the profound societal benefits of AI being counterbalanced by an unsustainable environmental cost.

Regulatory momentum and policy drivers

Due to the large amount of energy consumption, countries around the world have started taking regulatory actions, with the European Union emerging as a leader in mandating data center sustainability.

The revised Energy Efficiency Directive (EU/2023/1791) serves as a central pillar of the European Green Deal. This directive establishes "energy efficiency first," ensuring that energy efficiency is regarded in all relevant policy and major investment decisions across energy and non-energy sectors.

To ensure compliance, the European Union introduced mandatory reporting. Effective from September 15, 2024, all data centers located in the European Union with an installed information technology power demand of at least 500 kW are required to report critical sustainability metrics to a dedicated European database.

These mandatory reporting metrics are: total energy consumption, Power Usage Effectiveness (PUE), water usage, renewable energy share, waste heat reuse, and data center location and floor area.

Collected data will be utilized for benchmarking, rating, and enforcing improvements across the industry, promoting greater transparency and sustainable development. Penalties for non-compliance with the new regulations include fines, delayed permits, and adverse public ESG ratings.

Lastly, the new regulations force a new shift in infrastructure design. They are now encouraging the adoption of advanced cooling technologies, such as liquid and immersion cooling, on-site solar and wind generation, and heat recovery systems that can be integrated into district heating networks. Finally, merely marketing a facility as "green" is no longer sufficient. With the new data available, verifiable metrics and transparent annual public reporting are now required, marking a shift from "greenwashing" to "green accountability."

With the new regulations, sustainability is no longer a discretionary corporate social responsibility initiative but a critical competitive differentiator and an emerging prerequisite for market access and growth in Europe.

The new regulations will accelerate innovation and investment in green technologies and practices within the EU, potentially establishing a "green premium" for compliant and high-performing facilities.

Core technologies of green data centers

With the new requirements for data centers, it is crucial to understand how to reach them.

Advanced cooling technologies

With the advancement of GPUs, traditional air-cooling methods are becoming obsolete —at least in most cases. Data centers in cooler climates, such as those in the Nordic region, can utilize low ambient temperature, reducing their reliance on air fans. For optimal performance, it can be combined with adiabatic cooling, which uses the natural process of water evaporation to cool the air. Adiabatic systems utilize a heat exchanger to cool the data center air without adding humidity, resulting in high efficiency while maintaining precise environmental control.

Direct-to-Chip (DTC) liquid cooling technology utilizes specialized coolant that is applied directly to the heat-producing components. As liquids are more effective at heat transfer than air, depending on the type of liquid cooling, it can be dozens of times stronger than air cooling. Liquid cooling enables components to operate at maximum capacity for longer periods without thermal throttling.

Immersion cooling improves upon the DTC method by fully submerging hardware in a non-conductive, dielectric fluid. This method allows superior and uniform heat dissipation, allowing rack densities to exceed 100 kW.

Rear-door heat exchanger (RDHx) is frequently used as a hybrid solution alongside DTC cooling. This combination is important because DTC configurations always require a secondary air cooling system to handle residual heat from components not directly cooled by the liquid loop, such as memory modules, voltage regulators, storage devices, and network cards.

RDHx is one of the most widely adopted liquid-cooling solutions for high-density HPC and AI racks in modern data centers. RDHx units replace the standard rear door of a server rack with a water-based heat exchanger panel that captures heat immediately as air exits the servers.

The cooling process uses the server fans to pull ambient air through the IT equipment. Hot exhaust air is then directed through the heat exchanger, where thermal energy is transferred to chilled water or a cooling fluid circulating through the system. The cooled air returns to the data center space at near-ambient temperature, maintaining target conditions without relying on traditional CRAC or CRAH systems.

This approach enables very high rack densities, typically ranging from 50–100 kW per rack, with some manufacturers offering solutions capable of handling up to 200 kW per rack. In practice, RDHx has become increasingly popular in modern AI and HPC data centers because it offers an excellent balance of performance, flexibility, cost-effectiveness, and ease of installation.

Efficient power infrastructure

EU requirement for green data centers mandates that new data centers are met with new renewable capacity. That way, investments in data centers also directly lead to the creation of new clean energy added to the grid.

The most direct method involves on-site power generation with the installation of solar panels or wind turbines. However, for large-scale projects, the industry standard is a Power Purchase Agreement (PPA). PPA is a type of long-term contract in which a data center operator agrees to purchase power from a specific renewable energy project, often one that has not yet been constructed. This way, the power provider is guaranteed financial certainty, directly funding the green transition.

Waste heat reuse

Data centers produce significant amounts of low-grade heat, which can be reused in various ways. The most popular method is to utilize access heat as a district heating network for residential, office, or community buildings, reducing energy consumption and carbon emissions. In Stockholm Data Parks, 30 data centers are connected to the city's heating network, heating 10,000 residential units.

With the waste heat, there are practical examples where data centers are used in agricultural settings. One of the pioneers of this technological use is located in Boden, Sweden, where a 550 kW crypto-mining container provides heat to a nearby greenhouse, maintaining a temperature of 22-24 degrees Celsius throughout the year, even when outdoor temperatures drop to -30 degrees Celsius.

Future technologies in green data centers

As energy needs for data centers continue to rise daily, technology is struggling to meet the demands and maintain sustainable energy consumption.

Paradoxically, the largest user of data centers, AI, is both a problem and a solution for this issue.

Predictive cooling & thermal solutions based on AI are evolving cooling systems by forecasting thermal loads from historical and real-time sensor data. These algorithms automatically adjust cooling parameters based on real-time data, reducing energy use and saving up to 15% in cooling costs in trials and simulation studies.

Flexible task schedules also help reduce the carbon footprint. Google, among other companies, uses AI to schedule flexible, non-urgent tasks when the workload is at its lowest. The system utilizes analytical pipelines to gather the next day's carbon intensity forecasts, use models to predict demand, and generate Virtual Capacity Curves (VCCs) for all data center clusters across Google's fleet. Based on the collected data, VCCs impose an hourly limit on resources available for temporally flexible workloads, thereby preserving daily overall capacity while enabling all such workloads to complete within the day.

New overall metrics can also help with the carbon footprint. Water Usage Effectiveness (WUE) measures the amount of water consumed (in liters) per unit of energy (kWh). Industry averages around 1.8L/kWh while best-in-class facilities target <0.2 L/kW.

Carbon Usage Effectiveness (CUE) calculates total CO₂ emissions per kWh of IT energy use, combining PUE with grid emission factors. As the carbon intensity of energy shifts, CUE helps operators reduce the actual climate impact of data center operations.

The future of data centers is green and accountable

The future of data centers is unmistakably green. As demand surges, achieving genuine sustainability is crucial for both environmental reasons and business survival.

With comprehensive regulations in place in the EU, sustainability has shifted from voluntary good practice to a mandatory standard.

In the future, advanced cooling solutions, AI-driven optimization, waste heat reuse, and renewable energy integration will no longer be innovative exceptions but standard industry requirements. Companies that adopt these green practices now will position themselves as market leaders, benefiting from lower operating costs, an enhanced brand reputation, and compliance advantages.

These regulations will help ensure that a data center's true success will be measured not only in performance metrics but also in its accountability and commitment to our shared ecological future.

Hey, you! What do you think?

They say knowledge has power only if you pass it on - we hope our blog post gave you valuable insight.

If you want to share your opinion, feel free to contact us. 

We'd love to hear what you have to say!