The democratization of artificial intelligence (AI) is causing a tremendous revolution in the data center business, which is essential to our increasingly digital world. This transformation, the biggest since the industry’s founding, is changing everything from investment plans to data center layout and location selection. This and more are covered in JLL’s Data Centers 2024 Global Outlook Report.
Data centers have grown at an exponential rate over the last ten years due to the widespread usage of web-enabled devices and the growing demand for cloud services. In 2024, artificial intelligence (AI) and machine learning are driving a major change in data center architecture to meet the increasing power needs and deal with the shortage of data center colocation space, according to JLL’s report. Regional power restrictions and competition from other high-power consumers, such the infrastructure supporting electric vehicles and modern manufacturing, are contributing factors to this shortage.
According to European Commission projections, the demand for electricity is expected to rise by 60% by 2030, necessitating significant investments in power production capacity to satisfy requirements for renewable energy. This situation would encourags creativity and inventiveness since it offers the data center business both opportunities and threats.
Data Center Architecture & Construction
AI and the digital economy are having an increasing impact on data center architecture, according to the JLL report. Conventional facility designs are changing as a result of the need for bigger data centers outfitted with specialized technology, especially liquid cooling systems. These days, operators are looking for cutting-edge algorithms and software to boost efficiency improvements, with an emphasis on renewable energy sources. The size of new initiatives is also being driven by AI; needs are now ranging from 300 megawatts (MW) to over 500 MW, changing the criterion for site selection to give priority to power supply and delivery schedules.
Global data center construction is being accelerated by the continued move to the cloud, the expansion of the Internet of Things (IoT), laws governing data sovereignty and residency, and the growing volume of data produced by individuals and companies. The volume of data generated is predicted to quadruple over the next five years in comparison to the previous ten. With a five-year compound annual growth rate (CAGR) of 18.5%, the total storage capacity in data centers and endpoint devices is expected to increase from 10.1 zettabytes (ZB) in 2023 to 21.0 ZB in 2027.
The capacity of newly constructed hyperscale and colocation data centers has been gradually increased by developers. Structure Research projects that over the next five years, global colocation MW will expand at a 15.2% CAGR. Nowadays, most primary markets have a supply of at least 600 MW, and several even surpass 1,000 MW. Secondary markets are drawing more and more attention from lenders, investors, and developers looking for fresh prospects. These markets generally provide 100–600MW of supply. As edge deployments bring data centers closer to customers and comply with national data sovereignty rules, emerging markets are also anticipated to develop.
The Impact of GenAI
The data center industry is expected to be severely disrupted by generative AI (GenAI), not only in terms of the quantity of new facilities required, but also in terms of their layout and placement. Over the next 10 years, Bloomberg Intelligence projects that the generative AI business will reach a valuation of $1.3 trillion. These AI-focused data centers won’t look like traditional ones since they’ll need a more performance-intensive and tightly clustered IT infrastructure, which will result in increased heat and variable power usage, according to the JLL report.
Operators of data centers must carefully manage and distribute power resources according to the kind of data being processed and the stage at which generative AI models are being developed, added JLL in its report. When choosing a location for AI-specialized data centers, developers have more freedom, taking into account things like energy prices, transmission power availability, renewable energy sources, and temperatures that allow for free cooling. In 2024, it is anticipated that data center building would pick up speed due to the increasing demand and expanding needs for data. Smaller organizations must make plans ahead of time in order to get the capacity they need, while hyperscalers and other major companies are making commitments to space long in advance of delivery.
Rack power density will continue to rise as generative AI becomes more widely used. The biggest requirement for high-density infrastructure is seen in hyperscalers, who are at the forefront of AI and HPC. Large facilities now have an estimated average density of 36kW per rack; this density is predicted to increase to 50kW by 2027 at a 7.8% CAGR over the next several years. It is anticipated that many AI cluster needs would reach 80–100kW/rack.
Enterprise and colocation data centers will need to change to handle these sophisticated IT demands, stated JL in its report. Power density is rising as a result of growing demand for hyperconverged infrastructure, or HCI. Increased rack density demand is being met by the industry thanks to innovations in infrastructure. Specifically, liquid cooling systems mark a revolutionary development in operational efficiency that is consistent with the goals of sustainable high-performance computing.
To read the full JLL report, visit their website here.
