JLL famous that the arrival of AI in knowledge facilities has introduced important adjustments to the business, notably when it comes to energy density and facility dimension
In sum – what you’ll want to know:
AI reshaping knowledge middle design – The rise of AI is driving demand for smaller, extra power-dense services resulting from GPU prices reaching $30M per MW, JLL says.
Cooling and construction reimagined – AI {hardware}’s weight and warmth require new designs, together with liquid cooling and stronger flooring masses, plus hybrid HVAC techniques for blended tools.
Retrofits provide near-term reduction – With colocation emptiness at 2.6%, adaptive reuse and stranded energy capability in cloud-shifted websites are rising as scalable options for 1–3MW AI workloads.
As AI adoption accelerates globally, knowledge middle operators are grappling with unprecedented infrastructure and actual property calls for. From energy density to area constraints, the race to construct AI-ready services is reshaping the digital spine of the trendy financial system, Sean Farney, vice chairman of knowledge middle technique at JLL, advised RCR Wi-fi Information.
“The arrival of AI in knowledge facilities has introduced important adjustments to the business, notably when it comes to energy density and facility dimension,” the JLL govt stated.
Hyperscale suppliers proceed to develop large campuses to help conventional compute wants, however for AI-only operations, the economics and infrastructure look very totally different. “The extraordinarily excessive value of GPU server tools, which might attain $30 million per megawatt, makes it financially impractical to construct million-square-foot AI-only services apart from these with the deepest pockets,” he added.
This monetary actuality is driving a development towards “smaller, extra power-dense buildings.” AI infrastructure isn’t just dearer — it’s bodily totally different. “AI differs considerably from conventional servers, with {hardware} resembling large, heavy jet engines moderately than the simply manageable servers of the previous,” Farney defined. This shift is forcing operators to rethink flooring loading capacities and the bodily construction of buildings themselves.
Cooling infrastructure can be present process a change. “Liquid cooling has emerged as a brand new problem and alternative in AI knowledge facilities,” stated Farney. Whereas the know-how will be built-in with present chiller techniques — creating some retrofit potential — “air cooling continues to be essential for personnel, community gear and different non-liquid-cooled tools.” This hybrid requirement complicates facility design and HVAC planning, even in new builds.
On the identical time, useful resource constraints are piling up. “The info middle business is dealing with extra challenges resulting from energy and land shortages, in addition to restricted colocation availability,” the JLL govt warned. With colocation emptiness charges dropping to simply 2.6% by the tip of 2024 and rents up greater than 11% throughout the U.S., operators are on the lookout for inventive options.
Some of the viable choices is adaptive reuse — repurposing industrial or business property into AI-capable knowledge facilities. “This strategy harkens again to the early days of the web when many iconic knowledge facilities had been repurposed from present industrial services,” Farney famous. These conversions will be sooner and cheaper than greenfield developments, particularly in city areas the place energy and land are scarce.
Retrofits are additionally proving very best for smaller AI workloads. “Many services which have shifted their important masses to the cloud over the previous decade now have stranded energy capability. These areas may very well be appropriate for smaller-scale AI deployments of 1-3 megawatts, which are sometimes required for product growth and testing labs,” stated Farney.
Regardless of the constraints and complexity, he sees the business rising to the problem. “The info middle business is demonstrating flexibility and agility in adapting to those new applied sciences and their distinctive necessities,” he stated. “The business is constantly evolving its approaches to design, development and operations to accommodate the transformative potential of AI whereas navigating the challenges it presents.”