Tommy Lee Walker - stock.adobe.c

Datacentre operators face capacity planning challenges as AI use soars

A report by real estate consultancy JLL highlights the pressures datacentres operators are facing as they respond to the growing demand for artificial intelligence workloads

The impact the democratisation of artificial intelligence (AI) technologies is having on global demand for colocation capacity will force datacentre operators to rethink how they build and run their facilities.

That’s according to a 24-page report by commercial real estate and investment management company JLL, which predicts the take-off of AI will cause global datacentre storage capacities to increase from 10.1ZB (zettabytes) in 2023 to 21ZB by 2027.

“As the datacentre industry grapples with power challenges and the urgent need for sustainable energy, strategic site selection becomes paramount in ensuring operational scalability and meeting environmental goals,” said Jonathan Kinsey, EMEA lead and global chair for datacentre solutions at JLL.

“In many cases, existing grid infrastructure will struggle to support the global shift to electrification and the expansion of the critical digital infrastructure, making it increasingly important for real estate professionals to work hand in hand with partners to secure adequate future power,” he added.

All this has huge implications for operators when it comes to planning out the physical footprint of their facilities, as the amount of compute capacity they will need to house is set to soar.

“Most new datacentre builds 10 years ago had a critical IT load capacity of less than 10MW,” the report stated. “Today, it is not uncommon to hear developers announce news builds of 100MW or more.”

And with the take-off of generative AI (GenAI), the amount of space and power datacentres require will increase and create even more challenges for the sector. Especially as there are major differences in how a datacentre running more traditional, enterprise workloads is built versus one that is engineered specifically to accommodate AI workloads.

“AI-specialised datacentres look very different [to] conventional facilities and may require operators to plan, design and allocate power resources based on the type of data being processed or stage of generative AI development,” said the JLL statement.

“As the amount of computing equipment installed and operated is expected to continue increasing with AI demand, heat generation will surpass current standards.”

It added: “Since cooling typically accounts for roughly 40% of an average datacentre’s electricity use, operators are shifting from traditional air-based cooling methods to liquid cooling.”

AI workloads are energy-intensive, and this will also require operators to think about the best way to balance the energy needs of their facilities with their sustainability strategies too.  

“Generative AI’s greater energy requirements – ranging from 300 to 500-plus megawatts – will require datacentre operators to increase efficiency and work with local governments to find sustainable energy sources to support datacentre needs,” said JLL.

“The need for more power will require datacentre operators to increase efficiency and work with local governments to find sustainable energy sources to support datacentre needs.”

On this point, JLL said the situation should prompt governments around the world to consider investing in the respective electricity grids, with JLL making the point that around one-third of Europe’s grid infrastructure is more than 40 years old and requires an estimated €584bn of investment to meet the European Union’s green goals.

“The global energy conundrum presents both opportunities and challenges to commercial real estate leaders with a stake in the datacentre sector,” added JLL. “Generative AI will continue to fuel demand for specialised and redesigned datacentres, and developers and operators who can provide sustainable computing power will reap the rewards of the data-intense digital economy.”

Read more about generative AI and datacentres

Read more on Datacentre capacity planning

Data Center
Data Management