AI and Data Centres: Ensuring the Next Era is Sustainable

As AI demands continue to increase, so does the pressure on data centres.

Businesses seeking to live up to the digital transformation potential are hungry for the power of AI. With demand skyrocketing, data centres are having to find ways to cater to the need for higher performance and infrastructure that is more scalable than ever before.

Data centres that are supporting AI are faced with the challenge of higher computational power, prompting greater changes in facility design, newer cooling technologies and greater innovation to reduce carbon footprint.

As the data centre sector is called upon to confront its sustainability challenges, Data Centre Magazine spoke with Tom Kingham, Design Lead Europe & Japan at CyrusOne. He explains how data centres are starting to evolve into renewable energy developers to confront the energy efficiency challenge, in addition to how facilities are looking to be more flexible - both digitally and environmentally.

The true impact of AI on the data centre industry

AI has changed the game for data centres around the world, with the technology prompting a need for higher performance and more scalable infrastructure. As Tom explains, companies are now adopting AI at a faster rate to handle complex tasks, which ultimately means that they are using more power to be just as efficient.

Consequences of this shift could be that data centres start to fall away from their emissions targets and fail to be more sustainable. A notable recent example of this is Google, which admitted its overall GHG emissions had surged by 13% within a year, as our sister publication Sustainability Magazine reported.

The cause, according to the tech giant, was due to its increased number of data centres powering AI.

To mitigate such events, Tom explains that data centres should continue to address the unique demands of AI workloads, including the “need for ultra-high-density power usage, advanced cooling technologies, and scalable designs.”

He explains: “This has been an ongoing priority for CyrusOne, guiding the design of current and future projects to effectively scale their AI infrastructure. Inevitably, we are seeing a stronger focus on sustainability, with innovations reducing physical space requirements and carbon footprints, reflecting a broader trend towards greener data centre operations.”

With AI workloads operating at a much higher density, Tom explains that CyrusOne has had to reconsider the design of its facilities.

“This consideration is both for how we accommodate existing facilities, or those under construction, and how we design for AI specific workloads,” he comments. “This is a huge challenge for third party developers and operators such as CyrusOne because our facilities need to remain as flexible as possible for each customer requirement and whether it will be cloud or AI.”

Adopting renewables for sustainable resilience

In addition to creating sustainable initiatives, the key challenge is staying sustainable - and not enabling disruptive technologies to drain increased levels of power. The data centre sector is still facing unprecedented demands for edge computing, which are set to continue as increased use of AI creates a greater need to process data closer to the source.

It is clear that AI is here to stay and will continue to significantly impact the sector. As Tom explains, the technology’s need for increased processing power is leading data centres to adopt more powerful GPUs and specialist hardware - but this needs to be balanced with energy efficiency.

“The industry's growth, driven by increased demand for AI-infused systems and enterprise digitisation, is prompting a shift towards more renewable energy solutions, with data centres adopting renewable energy sources, energy-efficient cooling systems, and designs aimed at lower power usage effectiveness (PUE),” Tom explains. “Additionally, there is stronger focus on hybrid and multi-cloud solutions, as companies seek to build more flexibility and resilience against the managing of intensive distributed workloads.”

Data centres have their work cut out for them to meet customer needs, creating new ways to support intensive workloads while balancing performance with energy consumption.

“To meet these demands, data centres will need advancements in several critical areas,” Tom notes. “The substantial heat generated by dense AI infrastructures will necessitate efficient cooling solutions such as liquid-to-chip cooling and closed-loop chillers, ensuring effective heat management without excessive water use. 

“AI workloads will also impact data centres’ space utilisation, which will need to be optimised to accommodate high-density workloads, thereby also reducing physical footprint and material usage.”

******

Make sure you check out the latest edition of Data Centre Magazine and also sign up to our global conference series - Tech & AI LIVE 2024

******

Data Centre Magazine is a BizClik brand