Artificial Intelligence (AI) is oozing into every crevice of data centre operations. Although a hot talking point of 2024 so far, these conversations are only scratching the surface as the technology and implementations continue to escalate. Ramprakash Ramamoorthy, Director of Research at ManageEngine, touches on this and what’s to come.
“It’s no secret that the rapid growth of AI is placing sizeable demands on data centres. The massive computational power and data storage AI demands mean we will need to shift data centre design and operations to manage the increased energy consumption and address sustainability challenges.
However, AI itself holds the key to addressing these challenges. The technology can help optimise energy efficiency and can enable sustainable data centre design. For example, leveraging AI algorithms and Machine Learning can help data centres monitor and adjust cooling systems. The technology can also help maintain equipment and allocate resources based on real-time demands. This adaptive approach could minimise energy waste, reduce carbon footprints and extend infrastructure lifespan.
AI could also be used to predict potential equipment failures before they occur. It does this by analysing sensor data and historical patterns, which enables proactive maintenance and minimises downtime, ultimately improving operational efficiency and extending infrastructure lifespan.
In the realm of cybersecurity, AI-powered solutions can also help protect data centres. The technology’s advanced threat detection and response capabilities are integral to safeguarding data from increasingly sophisticated cyberthreats. By continuously monitoring network traffic, user activities and system logs, AI algorithms can identify anomalies and potential threats in real-time. This ensures business continuity and enables rapid response and the execution of mitigation strategies to protect against data breaches.
As the demand for AI technologies continues to soar, their impact on data centre design and operations will be transformative. Forward-thinking organisations are recognising this and leveraging AI’s capabilities to optimise their data centre operations whilst simultaneously positioning themselves at the forefront of the AI revolution. They can ensure high levels of performance, security and energy efficiency, which in turn enables them to better serve customers and stay ahead of the curve in an increasingly competitive landscape.
Companies that proactively embrace the shift toward AI to reimagine their data centre approach will be well-positioned to thrive in the digital age. They can deliver superior performance, efficiency and security to their customers, solidifying their position as industry leaders in the era of AI-driven data centre innovation.”
Scott Smyth, CEO and Founder, Soben
There’s no doubt that AI will have a profound influence on data centre design. As we navigate speeding technological advances, we envision a future where data centres are not only buildings for information but also a dynamic ecosystem optimised for AI-driven processes and analysis.
Amongst the most significant design impacts is the emergence of high-density racks and liquid cooling technologies. These stand at the forefront of innovation but come with tremendous challenges. In an AI environment, high-density racks over 100kW are no longer an exception, which forces us to move away from more traditional air-cooling solutions.
Liquid cooling is a good solution to this challenge, offering high-level thermal conductivity and energy efficiency compared to air-cooled systems. The question is whether this can be implemented fast enough to cater for the rapid growth of AI, and in what form. I’m excited to see which solution the market will embrace. Are we moving towards immersive cooling, direct-to-chip, or will we perhaps settle for an intermediate solution such as rear-door air cooling?
Let’s not forget that increased power density in white space areas presents a technical challenge for electrical design. There are opportunities for streamlining and eliminating electrical equipment previously focused on a more traditional density of 25kW and below. However, this is not without its risks, so it should be at the top of the agenda for every MEP designer around the globe.
All these only touch on what is happening inside the data centre. The elephant in the room is how AI data centres will fit into the current energy ecosystem. Power surges, but especially power drops, associated with AI require a completely new approach to power purchase agreements with grid providers. So, AI developers need to bring their own power, or at least be championing the discussion on how to redefine the ecosystem.
We’re at a pivotal moment in history. AI will be disruptive to our whole industry. Designers, developers, operators, builders and consultants must be dynamic, and ready to think on their feet.
Chris Coward, Director of Project Management, BCS
AI is already changing the way we design data centres as it requires far more processing power than standard computing, and so needs more power and cooling. In simple terms this means that data centres that are specifically designed to support AI’s high-density workloads will require larger area outside plant areas to house the larger and more advanced cooling equipment.
Data centre designs are also changing to ensure they can adapt to include liquid cooling to address the higher density. For existing facilities this may mean adapting cooling solution implementations and retro fitting air cooled, direct-to-chip and or immersed solutions.
One real positive is that these data centres could be located in secondary locations as there are less latency concerns – the data is just processing. This could unlock building in new regions that are less ‘saturated’ by the sector, these regions may have better power availability, driving employment and prosperity to a new area of the country. Existing empty buildings that have been left empty due to poor connectivity will become a stronger option bringing about additional sustainability benefits.
However, power is even more of an issue. There has been much debate around how to get the necessary power to support these facilities. The challenge is not so much the power, as this is a universal problem, but more one maximising the use of available power – being a little creative in our approach. Failure to find a solution may well be a limiting factor as Amazon President and CEO, Andy Jassy, warned this year that there was ‘not enough energy right now’ to run new Generative AI services.
Finally, the industry will need to start engaging with planners and local community leaders to educate them as to why the designs and requirements have changed. This won’t happen overnight and as this unprecedented demand continues lead-times may start to affect the ‘speed’ of deployment as the sector struggles with this lack of awareness and both physical and human resources. Early engagement with the community and planning teams is therefore key as is the education message to the market.
Martin Coulthard, Global Vice President, Digital Customer Experience, Vertiv
In today’s fast-paced B2B environment, informed decision-making can be the difference between success and failure. Businesses that make speedy, evidence-based decisions – whether that’s embracing a new piece of technology or overhauling an entire facility – are in the best position to be successful. Augmented Reality (AR) tools can act as bridges, connecting complex products and solutions, providing immersive product exploration and an understanding that previously seemed out of reach.
AR is a powerful tool that can help to optimise data centre and IT planning and design. It allows users to see and explore equipment in a desired location before purchase and installation. By visualising critical infrastructure components within operational contexts, decision-makers can gain a deeper understanding of how solutions could seamlessly integrate into their operations.
The virtual representation provides a tool to help plan space and provide support after installation, with the intent of improving understanding of how the infrastructure will support their compute and impact the physical footprint. This streamlines the evaluation process, fosters better-informed choices and ultimately leads to more efficient operations.
Innovative tools like the Vertiv XR App empower IT managers, data centre operators and channel partners to visualise data centre solutions within their facilities. Just as consumers use AR apps to envision new furniture in their homes, this app enables users to position data centre equipment precisely where it will be installed, significantly aiding in effective planning and spatial comprehension. In addition, because Vertiv deals with mechanical and electrical equipment, the XR App enables users to look safely inside the equipment. The AR models are also animated with items such as airflow, which allows the user to see equipment simulations and things that are invisible to the naked eye.
The shift toward AR-driven customer experience (CX) in the data centre industry is an important example of how critical it is to be adaptable and flexible in today’s hyper-competitive landscape. Companies that embrace this transformation are better positioned to meet the evolving expectations of their customers and partners.
Nick Ewing, Managing Director, EfficiencyIT
In the evolving landscape of data centres, Artificial Intelligence (AI) is beginning to transform all aspects of design, efficiency and operations. Data centres are becoming hotter and denser, and as new advancements in GPU and CPU-powered technologies force an evolution in data centre power and cooling requirements, air cooling has no longer become viable for High-Performance Computing (HPC) and AI workloads.
At EfficiencyIT, we’re working with customers across a host of industries – life sciences, bioinformatics and motorsports – to design, install and specify liquid-cooled solutions within high-density data centres, regardless of our customers’ choice of vendor cooling technology.
The key benefits of liquid cooling include the ability to future-proof your data centre environment to accommodate workloads of over 100kW per rack.
Additionally, liquid cooling solutions can help businesses achieve leading levels of efficiency.
For example, customers utilising liquid-cooled infrastructure can benefit from up to 84% less cooling usage per kW of IT equipment (ITE), reducing energy consumption by up to 40%. From a sustainability standpoint, liquid cooling also decreases water usage by up to 96% and lowers CO2 emissions by up to 40% – providing a more sustainable and resilient solution for AI-powered workloads.
At EfficiencyIT, we’re also seeing the development of high-density data centres at the Edge. In particular, the adoption of prefabricated data centre solutions combines best-in-class engineering and flexible cooling technologies, to provide a safe and secure environment for next-generation AI.
In some instances, AI is also helping revolutionise facility management by using predictive analytics and Machine Learning to foresee potential system failures before they become critical. This allows organisations to undertake predictive maintenance, proactively reducing system downtime and extending the lifespan of critical equipment.
These AI-driven systems will also continuously monitor and adjust environmental controls, such as cooling temperatures and airflows, ensuring optimal performance and enhanced levels of energy efficiency for high-density applications.
Today, integrating AI into data centre design and operations marks a significant step change in efficiency, sustainability and security. However, with a focus on innovative engineering and by leveraging flexible cooling solutions, data centres provide a secure and resilient home for next-generation AI technologies.
