As enterprises chasing AI success invest to build out the high-performance compute needed to drive innovation, a key challenge is emerging: the cost of real estate.
Enterprise AI ambitions remain a key priority. Figures from IDC suggest that worldwide spending on artificial intelligence is forecast to reach $632 billion in 2028.[1]
But further research in the 2025 State of the CIO study showed that IT modernisation was the reason why technology budgets were increasing for a quarter of respondents (26%). [2]
The race for AI success coupled with the need for powerful IT infrastructure means businesses must overcome the challenge of rising data centre costs to achieve their ambitions.
Barrier to growth
On premises deployments remain key in the dawning era of AI.
As Dave McCarthy, Research Vice President for cloud and edge services at IDC, told CIO.com: “The excitement and related fears surrounding AI only reinforces the need for private clouds, Enterprises need to ensure that private corporate data does not find itself inside a public AI model.”[3]
But new data centres are costly, complex, and slow to build. Meanwhile, real estate costs generally continue to rise, squeezing IT budgets further.
Nicola Tan, Director of Market Development for Enterprise Software at AMD, comments: “To stay ahead, enterprises must consolidate and optimise—maximising performance within existing space while keeping energy and property costs under control.”
“Consolidating ageing server infrastructure can deliver dramatic savings, freeing up real estate and cutting power use significantly. One AMD customer experienced a 10 times increase in capacity, all within its existing footprint—effectively unlocking the benefits of a new datacentre without building one.”[4]
According to Tan, many enterprises rely on servers that are 5 to 10 years old or more — which means relying on hardware that’s often slow, power-hungry, and inefficient.
While these systems may seem cost-effective because they’re already paid for, the reality is their poor performance can end up costing more in the long run compared to upgrading to modern systems.
A strategic infrastructure assessment helps identify consolidation opportunities. From there, businesses can define goals, prioritise workloads for upgrade, and begin realising benefits quickly.
“Crucially, consolidation isn’t a one-off project—it’s an ongoing journey, where regular modernisation ensures enterprises continue to maximise performance and efficiency as technology evolves,” says Tan.
Consolidating data centre infrastructure
AMD technology helps businesses consolidate their infrastructure in three ways. First, AMD EPYC™ CPUs deliver high levels of performance per core, letting businesses run more virtual machines with fewer cores. Second, a higher core density means those cores can fit into fewer servers, shrinking the enterprise datacentre footprint. Third, efficient AMD cores enable exceptional performance per watt for power efficiency. Essentially, AMD enables businesses to fit more servers on a single rack, allowing them to grow without having to spend on additional real estate.
However, a common concern with consolidation is the so-called “blast radius”—the risk that if one server fails, a large number of workloads go down with it.
According to Tan, there’s a smart way to consolidate while minimising that risk: “Instead of stacking up workloads on traditional dual-socket servers (like two 32-core CPUs), AMD offers high-core-count single-socket solutions.”
“For example, a 64-core AMD EPYC™ CPU delivers the same density but in a simpler, power-efficient package—cutting power use from 600 watts down to 400 watts per server.[5] This approach maintains performance and capacity, while reducing complexity, power consumption, and potential points of failure. It’s a great way to consolidate without compromising resilience.”
As real estate costs climb, every square foot of the datacentre must earn its keep. Enterprises that embrace smart infrastructure and high-efficiency compute will not only consolidate more effectively—they’ll unlock impressive performance, scalability, and control often with less space and effort.
Get more insights into the future of the datacentre.
[1] IDC, IDC Media Center, “Worldwide Spending on Artificial Intelligence Forecast to Reach $632 Billion in 2028, According to a New IDC Spending Guide,” August 2024
[3] CIO, 5 reasons the enterprise data center will never die, January 2025
[4] AMD, DBS transforms its data center with AMD EPYC™ CPUs
[5] 2x EPYC 9355 (32c) vs 1x EPYC 9555 (64c), both configured to the high end of the configurable TDP range