There are a range of issues signalling a pivotal shift in the world of data centres – among them rapidly evolving customer requirements, sustainability, skills enhancement and the custodianship of data, says Rennie Dalrymple, Head of Data Centres at Ridge.
The emergence of Artificial Intelligence (AI) looms large and it’s clear that it is not just a trend but an approaching tsunami that is reshaping the future of data processing and management. Rather than ringing alarm bells, this should be beneficial in many ways.
Enterprise clients’ requirements are evolving due to AI and High-Performance Computing (HPC) requirements. It’s no surprise that a great deal of time and energy is devoted to adopting appropriate technology and adaptations to infrastructure to support rapidly evolving AI models.
It is anticipated the utilisation of AI in our future projects will offer insights and optimisations to enable the creation of improved infrastructures. Through AI-driven analytics, data centres can identify energy-efficient practices, optimise cooling systems and minimise environmental impact, therefore taking a significant step closer to action net zero targets.
Looking ahead and examining the operational potential for data centres further, AI-enhanced algorithms will seek to drive predictive maintenance and resource allocation. There are unparalleled performance and reliability gains to be made if the industry is willing to leverage AI in this way.
The envisaged proliferation of AI is only going to increase the demand for fully optimised data centres. For instance, ChatGPT, which launched in November 2022, generates new data instead of just processing existing data. This development highlights the increasing demands on data centres to handle these advanced capabilities.
To give some idea of the ‘gold rush’ of data centres, according to research expert Thomas Alsop, in 2024 the global Graphics Processing Unit (GPU) market, was valued at US$65.3 billion, with forecasts suggesting that by 2029 this is likely to rise to US$274.2 billion, growing at a Compound Annual Growth Rate (CAGR) of around a third in just five years, from 2024 to 2029.
And looking specifically at the financial performance of AI chip giant, NVIDIA, has recently overtaken Microsoft, to be the world’s most valuable company with a recent stock valuation of over US$3 trillion (£2.64 trillion).
The evolution of hardware such as GPUs which can transform the handling of complex computations by processing thousands of programmes simultaneously, allows computing capacity within data centres to increase dramatically. And a good job too – our clients are reporting that customer enquiries are seeking for rack outputs to be 5-7x higher than cloud applications.
The physical infrastructure within data centres is envisaged to go through a significant evolution as the building structure and services installation evolve to accommodate the envisaged increases in power density. In addition to power availability, the issues of emissions, noise and water consumption remain challenges for those involved in the planning, building and running of data centres to tackle.
We have always been acutely aware of their uniqueness as a building typology. This starts at the planning stage, where a lack of clarity on the categorisation and their impact on the urban environment and neighbouring communities, sometimes leads to early-stage challenges to overcome.
It’s important to stress the collaboration needed within all the different disciplines from engineering to architecture through to the technical support required to get a data centre planned, built, commissioned and ready for operation. Increased coordination between developers, IT customers, data centre engineers and the construction supply chain are all integral in designing and building these facilities.
As technology within the data centre evolves, several key considerations will dominate the sector, such as scalability to accommodate future growth without compromising on performance, incurring excessive time delay and cost escalation.
Once operational, data centre heat emissions and the subsequent impact on the need for cooling present another issue to be addressed, with the greater capacity enabled using GPUs bringing this into sharper relief.
Cooling systems are going to evolve to support increased densities and the optimisation of water use and sustainable energy sources will continue to see continued focus from IT companies, developers, operators and investors alike. And that is reflected in the global sales revenue of the data centre liquid cooling market. According to Persistence Market Research, it was around US$2.52 billion in 2021 with a predicted to surge at a CAGR of over 25% to reach US$31.07 billion by 2032.
Of course, sustainability is a key issue in data centres as with other building typologies. As part of the EU’s latest Energy Efficiency Directive, there is an EU-wide scheme for rating the sustainability of data centres, to encourage transparency and address issues such as using renewable energy and waste heat reuse.
There is doubt from some quarters, however, on regulators’ expertise in this area. Germany’s new Energy Efficiency Act has a stringent Power Usage Effectiveness (PUE) of 1.2 for new data centres by 2026, a challenging target with the current average PUE being 1.45. As PUE targets are lowered this could result in data centre closures, cloud migration and a greater need for refurbishments. Regulation from central governments, the EU and other legislative bodies is pushing for greater efficiency despite potential cost and operational challenges.
Consumers also play a key role in the sustainability of data centres, which they aren’t perhaps currently aware of. The link between an individual’s own digital footprint and the fight against climate change isn’t clear in the minds of most people in the way that driving a car versus getting public transport.
Pip Squire, Head of Energy, Design and Innovation at Ark Data Centres gave a useful analogy here at a recent roundtable: ‘If you are live streaming a movie on a train travelling at 70mph you are consuming as much energy – end to end from the handheld device, through the networks to the servers in data centres – as cooking your porridge in a microwave.”
Returning to the issue of AI, if we harness it appropriately, we can look forward to a future where automation and intelligence streamline operations, making data management more agile and responsive.
