Why data centre adaptability beats capacity for AI
TechRadar says AI demand is pushing data centres to plan for higher rack density, faster cooling upgrades and long-term flexibility rather than simply adding floor space.

Key takeaways
- TechRadar says AI is making rack density, cooling and grid capacity more important than simple floor space.
- The article describes racks moving from five to 10 kVA toward 15 kVA and then 25 to 30 kVA.
- Modular POD designs and in-row cooling can allow capacity to be added in stages.
- TechRadar says adaptable facilities can reduce upgrade time and support local sustainability goals.
In this article
AI changes the capacity question
According to TechRadar, data centre adaptability is becoming more important than raw capacity as AI workloads demand more power in each rack and faster changes to facility design. The article, published on 17 November 2026, says operators must plan for higher density, better cooling and long-term flexibility rather than simply offering available space.
Two years ago, many customers in regulated sectors such as finance and healthcare asked mainly about uptime, certification and cost. TechRadar says those questions still matter, but AI has added a harder one: whether a facility can support rising workloads on a much shorter timeline.
Rack density is the main constraint
The article says the infrastructure story is less about headline capacity and more about power density, cooling and the capacity of the grid. Racks that once supported standard enterprise systems at five to 10 kVA may need to support 15 kVA within 18 months. The kVA measure describes power capacity. Some facilities are already planning for 25 to 30 kVA in the next few years.
This is not a small upgrade. TechRadar says it requires a different physical design, often using modular POD architecture and in-row cooling that can be added as demand grows. POD is a modular data centre architecture that can add power and cooling in stages. That approach can control investment and improve efficiency as demand grows.
Cooling must scale with demand
Higher density creates more heat, so cooling has to be part of the core design rather than an afterthought. TechRadar says modular in-row cooling can be activated as IT demand rises, allowing operators to match cooling to actual need. This can avoid building more cooling capacity than is required on day one.
The article also says refrigerant-based cooling can keep water consumption low compared with water-based systems, and that Ilkari uses POD-based design and in-row cooling to scale cooling as IT demand rises. That matters because regulators and local communities are paying closer attention to energy and water use. Efficient cooling is therefore not only a technical issue. It is also part of how facilities maintain trust with regulators and communities.
Adaptable sites can upgrade faster
TechRadar gives an example of a POD reconfigured in 30 days to meet changing customer needs. The project changed 40 racks at five kVA into 20 racks at 10 kVA to support higher-density demand. The article says this kind of turnaround is possible when a facility is designed for change from the start.
By contrast, a site designed for older power needs may become a risk if changes require a long rebuild. The article says several established markets, including some European hubs, are struggling to add capacity quickly enough for demand. The lack of suitable density is a major reason.
Community impact and sustainability

The article acknowledges recent critical coverage over data centres using land, water and power that communities also need. It argues that the economic role of a site extends beyond the building. Data centres can support direct jobs and wider work in security, maintenance and cleaning, as well as logistics and transport. They can also encourage investment in energy and connectivity.
TechRadar says sustainability should be part of daily operations, not a future promise. In practice, that means efficient cooling, transparent community engagement and investment in skilled engineering staff. It also means working with local governments to ensure power and energy infrastructure can support higher-density facilities.
What technology leaders should ask
For CIOs and technology leaders, the article suggests a different question from the usual capacity check. Instead of asking only whether space is available today, they should ask whether a design can handle the density likely to be needed in three years. A facility that cannot adapt may become a constraint on AI, cloud, banking and public services.
TechRadar says AI-driven demand is compressing decisions that once took years into months. That makes flexibility a practical requirement, not a bonus. Facilities designed for modular growth, efficient cooling and local cooperation may be better placed to handle the next wave of computing demand.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



