We use cookies to improve your experience. By continuing to use our website, you agree to our privacy policy.

How data centres are adapting to the needs of AI

How data centres are adapting to the needs of AI
4 December 2023

Artificial intelligence (AI) has been ramping up with the launch in late 2022 of ChatGPT. Most computer users will have heard of this AI-powered, large language model (LLM) chatbot, or had a go using it to brainstorm ideas, plan a holiday itinerary or generate responses to any topic in real-time.  

ChatGPT acquired a million users in just five days after launching – faster to hit the million-milestone than other popular online services such as Instagram, Spotify and Netflix.

By February this year, ChatGPT had attracted one billion visits, part of the AI ‘gold rush’ that has seen a vast number of applications coming out to the market.

With the significant uptake of AI, the need for data centres is surging. Data centres sit at the heart of many organisations, and are indispensable in supporting clients with their evolving product lines and the development of new technologies.

According to Synergy research group, the IT capacity of new built hyperscale data centres is expected to more than double over the next six years, and the impact of AI technology and services has created the need for much more powerful facilities.

Balancing power and efficiency

Charles Penny, Managing Director of AREA3, a BGIS company, highlights that with the uptake of AI, data centres face the challenges of power, density, cooling efficiency and connectivity.

And as data centres are notorious consumers of energy, the onus is on these assets to be more sustainable.

That is, AI is fuelling demand for data centres and computer power, at the same time that sustainability initiatives are focused on reducing the world’s carbon emissions.

“This can create a perfect storm,” explains Charles, about the balance between the demand for computer power with the speed at which our initiatives can keep up. He says the two main buzz words in the data centre community at the moment are sustainability and AI.

“Sustainability is the driving force that will move the dial and encourage data centre operators to implement not only AI solutions, but also sustainable power and cooling opportunities such as wind, batteries and liquid direct to the chip or fully immersed technologies. In some countries modular nuclear projects are being seriously considered.”   

AI is improving performance

AI is already helping data centres in three main areas, including optimising energy efficiency, predictive maintenance and enhanced security.

AI algorithms are being used to monitor and control data centre operations in real-time. By analysing data on temperature, server loads and energy consumption, AI can optimise cooling systems and server allocation, reducing energy waste and lowering operational costs.

The second area involves using AI-driven predictive analytics to anticipate equipment failures before they happen. By monitoring the health of servers, storage devices and networking equipment, data centre operators can schedule maintenance and replacement proactively. This minimises downtime and improves overall reliability.

Finally, AI-powered cybersecurity solutions are strengthening data centre defences.

Machine learning algorithms can detect and respond to security threats more efficiently than traditional methods. They analyse network traffic patterns, identify anomalies, and quickly respond to potential breaches, thereby enhancing data centre security.

How we cool data centres is changing

In response to how companies currently manage the risk of disruptions, temperature control in data centres plus power load, Charles explains that there are many systems that are needed to make a data centre resilient and operational – fire, communications, water, power and cooling to name a few.

These are designed to run as efficient as possible to allow for a low power usage efficiency (PUE), which is a ratio of total energy load over the IT load. The lower the PUE, the more energy efficient the data centre is.

Advanced monitoring systems continuously track the performance and health of data centre components, including servers, cooling systems and power distribution. Charles expects data centres and clients will become more innovative, in order to meet business objectives that focus on energy use and global sustainability initiatives.

In 2014, Google purchased British AI company DeepMind, and this system was implemented to run Google’s data centres and allow for constant adjustments to air temperature, pressure and humidity.

Google’s investment in more efficient ways to cool its data centres resulted in less energy used for cooling by up to 40 per cent, and the savings translated into a 15 per cent improvement in PUE.

Charles says this example shows how companies are now approaching data centres with innovation such as AI to assist with delivering a more economical product.

“We recognise the exciting growth underway, and so do our clients, who are well informed about the challenges ahead and the emergence of new technologies. Everyone is trying to do their part, and solutions are coming out all the time to meet the challenge of managing data centres.” 

The outcome will be more sustainable systems

While the average density of server racks today is about 15KW, Charles says that is expected to rise to 50-100KW+ to meet future computer requirements needed by AI.

AI can automate certain tasks and has the potential to create new job opportunities and enhance human productivity. The impact of AI on employment varies by sector and depends on how organisations implement and adapt to AI technologies.

“At BGIS, we work with a lot of AI software providers as well as companies who consistently look at ways to improve the data centre industry, from the initial design phase, through construction into operation,” says Charles.

“We can see there’s a wave coming. From a project delivery, operational and innovative point of view, BGIS has the expertise to be that partner that covers every aspect of critical environments – that’s project delivery with our AREA3 team, sustainability and asset management with our HFM team, and facilities management led by BGIS.

“Many people fear that AI will completely replace human jobs across various industries, but I believe it is better to embrace and understand the technology rather than turn a blind eye and get left behind.”

Power is going to be one of the biggest issues, and the demand on power and global sustainability objectives are out of sync. Enterprise, Hyperscale and Colocation clients need to also agree on how improvements can be made not to the determent of redundancy but to the benefit of sustainability.

The adaption of AI and liquid cooling technologies are two of the tools that can assist in the drive to a more sustainable data centre future.

Let’s all work together to see what else is possible.

Charles PennyManaging Director, AREA3