Amidst the ever-evolving landscape of design, artificial intelligence (AI), and sustainability innovation, there lies a mysterious and intriguing world of green data centres that form the foundational pillars of our connected present and future.

In the digital age, the demand for data is skyrocketing, resulting in a surge in the need for data centres which will not subside anytime soon as AI-hungry workloads become widespread. As the custodians of this digital revolution, data centres shoulder a massive responsibility. Not only must they safeguard the world’s digital assets, but they must also do so sustainably. This transition to sustainability in data centres is not just a trend; it is a necessity. And it needs to happen now.

In the European data centre market, the rise of AI is causing a paradigm shift within the facilities themselves. AI-driven data centres offer numerous benefits, such as the ability to slash downtime, enhance overall system reliability, and translate into massive cost savings for organisations – something that proves more crucial in the current economic environment.

AI-driven data centres improve event detection and prediction abilities, offering fast failure detection and triggering self-healing mechanisms. AI also provides a clear picture of app resourcing levels, allowing for nimble scaling to meet demand in real time.

On the power front, AI applications consume significantly more than traditional set ups. For example, high-performance processors required for AI need more power than traditional processors.

This new AI data centre reality could lead to a spending rush to replace and install newer hardware. Tirias Research forecasts that generative AI data centre server infrastructure plus operating costs will exceed US$76 billion by 2028. This cost estimation takes into account the current course of generative AI, incorporating an aggressive 4X improvement in hardware compute performance. However, this gain is overrun by a 50X increase in processing workloads, even with a rapid rate of innovation around inference algorithms and their efficiency. Globally, AI data centres are set to inject a further US$500 billion in data centre development, according to Dell’Oro Group.

As AI demands more power, data centre operators must meet sustainability requirements and reduce greenhouse emissions – also with AI.

---

---

Building Fiercely

In Europe, and specifically, since 2020, there has been an explosion in the data centre market across south and central Europe – and this is not set to slow down any time soon, according to Marie Chabanon, chief technical officer of pan-European data centre operator Data4 Group whose portfolio spans 31 sites with a potential capacity of 850MW.

“Between 2022 and 2028, the Central and Eastern European data centre construction market is predicted to expand at a compound annual rate of 10.64%,” she explains. “This rapid and continual growth is leaving data centre operators in the region, including Data4, trying to catch up with a fierce level of demand.

The birth of the south and central European data centre market has been driven by the increasing demand for digital services and the need for sustainable infrastructure. Southern Europe, comprising countries such as Spain, Italy, Greece and Portugal, has witnessed a surge in data centre construction due to its favourable climate, reliable power supply, and strategic location.

These data centres are being built with a strong focus on sustainability, aiming to minimise their environmental impact. The integration of AI technology will also play a vital role in the sustainability-driven construction of new facilities.

“Data4 alone has increased production from around one to three data centres a year to five since 2020,“ Chabanon explains. “Today, our growth strategy is even more ambitious, with the new aim being to build ten new centres each year. Geographically, this has seen Data4 double the number of markets in which we operate – from three in 2016 to six in 2023.”

That growth is not only an enormous task, but it will need to be built on green AI-enabled pillars, something providers all around are taking significant strides towards by building green, scalable, and climate-friendly facilities.

“The reality is that if we look at the rate at which our industry is growing if we do not build in a sustainable manner, there is going to be nothing left to build with,” warns chief commercial officer Adam Levine.

He continues: “The raw materials to build data centres are not going to be there in 10 years. If we keep growing at this rate – and we keep growing -, can we keep doing it without a real focus on sustainability?”

Founded in 2006, the Paris-headquartered group’s commitment to sustainable development is deeply ingrained in its business plan with the aim to be a net zero organisation by 2030 – very much in line with the vast majority of operators globally.

Data4’s priorities include its commitment to Quality, Adaptability, and most notably, the Data4Good programme dedicated to Sustainable Development. The company’s focus on sustainability is neither an afterthought nor an add-on; it is an integral part of its ethos; its executives are quick to say.

This is built on a sustainability program designed to align with the ISO 26000 principles that encompass four key facets, including:

• Environment: Commitment to reducing the environmental footprint by employing energy-efficient technologies and practices;

• Social: Emphases on equity, inclusivity, and responsible operational practices;

• Innovation: Addition of innovative solutions to improve overall sustainability performance;

• Community engagement: Collaboration with various stakeholders, including municipalities, universities, and customers to foster a sustainable ecosystem.

---

---

Energy Efficiency and AI

The energy consumption of data centres has seen a significant surge with the rise in dependence on these facilities. To put things into perspective, as an example and according to researchers from Google and California’s University of Berkeley, it is estimated that the carbon emissions due to training ChatGPT are 552 tCO2e and its energy consumption is 1287 MWh. That would be enough to power, on average, between 200 and 300 European homes for a year. And this is just the tip of the iceberg.

Recognising the gravity of this situation, operators are placing a strong emphasis on energy efficiency. One of the key techniques employed to enhance energy efficiency is the use of none other than AI which is in its own turn redesigning grid and electric networks.

AI plays a crucial role in optimising various operational aspects of data centres, which today account for roughly 4% of the total greenhouse emissions worldwide, according to global real estate consultant house JLL. From managing energy consumption to predictive maintenance, AI has proven to be a game-changer. By analysing power consumption data and cooling system parameters, AI can dynamically adjust these parameters, thereby reducing energy consumption while ensuring optimal functioning of the equipment.

In addition to energy management, AI also aids in predictive maintenance. By examining data from sensors and monitoring equipment in real time, AI can detect early signs of potential failures, thus enabling optimal planning of maintenance operations. This reduces the risk of breakdowns and the associated costs.

Advanced robots equipped with AI and ML capabilities will be implemented by half of the cloud data centres by 2025, leading to a remarkable increase of 30% in operating efficiency, according to a recent report by Gartner.

Elsewhere, professional services firm EY has recently raised the novel concept of digital twins as one of the major near-future advancements in data centre management.

“Digital twins (real-time virtual representations) are now becoming key to data centre efficiency improvement,” says Alexy Thomas, EY Technology Consulting Partner.

“They allow the collection of data from all sources and help data centres to operate more sustainably, not only from a cost perspective but also from an environmental standpoint. From facility design to space utilisation, digital twin technology reduces carbon footprint. As a data centre becomes bigger and handles more data, the operations become more complex. Digital twins with AI and ML platforms analyse silos of data generated and track all the components within the facility to make real-time adjustments. This can also mean predicting behaviours, which in turn helps in predictive maintenance, cutting energy, time and costs.”

Leveraging complex data models, AI enables real-time analysis and decision-making, greatly enhancing the overall efficiency of the facility. However, training AI models with real data is a long and intricate process, requiring extensive data collection, rigorous data cleaning, and significant computational resources.

AI in Data4’s data centres relies on various data models, including customer energy requirements, environmental conditions, maintenance data, and equipment data. By processing these data, AI can adjust energy parameters in real-time, thus optimizing the energy performance of the facility.

---

---

Environmental Stewardship

It is widely acknowledged by most data centre companies that their infrastructure is a high consumer of electricity. As such, businesses are committed to reducing their environmental footprint through integrated management systems. For Data4, this commitment is reflected in the company’s long-term engagement in continuous improvement processes, adherence to international standards, and the search for innovative solutions.

One of the ways to reduce the environmental impact of hosting facilities is through the design of one’s own campuses. By owning its data centres on campuses, a company can better control energy consumption – and again, aided by AI platforms. The campus environment also offers a lower air temperature than dense urban environments, reducing the energy requirements for cooling. This not only benefits the data centre infrastructure and surrounding areas, but it also helps in reducing heat-producing sources in large urban centres.

Moreover, being located on average 30 minutes from large cities, such as Data4’s footprint, minimises social and environmental disturbances.

In addition, being located outside cities makes it also easier to help safeguard and promote biodiversity. To that end, for example, Data4 protects 70 hectares of listed woodland on its flagship Paris-Marcoussis campus, making a significant contribution to the preservation of local wildlife.

With numerous tree species and abundant biodiversity, the site is a valuable natural carbon sink. Forests, just like oceans or bogs, are major carbon sinks and play a vital role in reducing the greenhouse effect and therefore global warming. Nearly 45,000 tonnes of CO2 are currently stored in this forest. That is the equivalent of 45,000 eight-hour flights – individual seats – between Paris and New York.

In part two of this special deep dive into Data4’s green AI data centres, we explore further into the group’s use of renewable energy, why quality should always win over the need for speed, and other factors driving its expansion amidst a rapid boom in demand.