Amid reports of AI workloads demanding over 100kW per rack and widespread coverage of multi-gigawatt hyperscale campuses, many data center operators face a different reality: managing rising densities within legacy facilities.

According to Uptime Institute research, approximately 50 percent of data center facilities are more than ten years old, many of which were never designed for today’s densities, thermal loads, or dynamic workload profiles.

As AI and high-density compute reshape today’s infrastructure requirements, these facilities are increasingly being pushed beyond the limits of their original design and operating assumptions.

Traditional room-based cooling strategies, based on computer room air conditioners (CRAC) and computer room air handlers (CRAH), are struggling to cope with rising rack densities, constrained airflow, and uneven thermal distribution.

Refurbishment – rather than new build – is therefore the preferred path for many, driven by sustainability targets, cost pressures, and the need for speed to deployment.

For operators looking to modernize without a fundamental redesign, rear door heat exchangers (RDHx) – cooling units mounted at the back of server racks – are emerging as a practical, technically robust solution.

RDHx remove heat directly at the rack exhaust, reducing reliance on large-scale room airflow, improving thermal control, and enabling higher rack densities without extensive structural change.

Below are ten key reasons why RDHx should be considered as part of any data center refurbishment strategy.

1) Cooling closer to the heat source fundamentally improves control

Traditional CRAC and CRAH systems use room-level air conditioning to manage heat, which assumes predictable airflow paths and balanced loads – conditions rarely found in refurbished data halls.

RDHx remove heat directly at the rack exhaust, preventing it from mixing with room air. This significantly reduces recirculation and temperature stratification, resulting in more stable and predictable thermal conditions – especially important when legacy layouts restrict airflow optimisation.

For operators, this simplifies thermal management and improves rack-level monitoring when integrated with data center infrastructure management (DCIM) or building management system (BMS) platforms.

2) Reduced reliance on room airflow lowers energy consumption

As rack densities increase, room-based cooling demands disproportionately higher airflow and fan speed. This increases energy consumption, noise, and mechanical wear.

By extracting heat at source, RDHx reduce total airflow demand, enabling both facility and IT fans to operate at lower speeds. This improves overall efficiency, particularly in environments with physical airflow constraints such as constrained plenums or ceiling heights.

3) Efficiency gains extend beyond PUE

Power usage effectiveness (PUE) alone does not reflect inefficiencies caused by overcooling and uneven thermal distribution. In many facilities, CRAC and CRAH systems are typically sized to protect the hottest rack, resulting in widespread overcooling.

RDHx enable tighter thermal control at the rack level. This allows operators to raise ambient room temperatures, reduce overcooling, and better align cooling capacity with the actual IT load – particularly valuable in mixed-density environments.

4) Higher supply water temperatures improve system-level efficiency

Unlike traditional chilled-air systems, RDHx can operate effectively with higher supply water temperatures. This improves chiller efficiency, increases the number of hours where free cooling is possible, and reduces compressor runtime.

In refurbishment projects, this often enables the reuse of existing chilled or warm-water infrastructure with minimal adaptation. This avoids costly plant upgrades or replacements.

5) White space recovery improves data hall economics

CRAC and CRAH units occupy valuable white space, limiting layout flexibility and reducing available space for IT equipment.

Replacing room-based cooling with rack-level RDHx cooling can free up significant floor space, improving layout flexibility and capacity.

For colocation operators, this improves revenue density. For enterprise operators, it can defer the need for new builds or extensions.

6) Higher rack densities without rearchitecting the room

Supporting 20-40kW racks with room cooling typically requires containment, more airflow, and extensive redesign.

RDHx enable high-density deployments within existing rooms by removing heat at the source. This allows operators to introduce high-density zones alongside standard racks, without redesigning the entire data hall.

For refurbishment projects, this provides a viable pathway to deploy AI and HPC racks at ~30-40kW without major structural modifications, while also acting as a bridge to future liquid cooling.

7) Incremental deployment reduces refurbishment risk

Upgrading cooling systems in live data centers creates significant operational risk.

RDHx support a phased, modular approach. Operators can deploy systems rack by rack, allowing them to validate performance under real-world conditions before expanding.

This phased rollout reduces disruption, minimizes downtime risk, and aligns capital expenditure with actual demand growth.

8) Localized cooling improves operational resilience

In traditional room-based systems, cooling failures can have a widespread impact across the data hall.

RDHx localize cooling at the rack level, which simplifies fault isolation and allows operators to address issues without affecting the wider hall.

Operators can perform maintenance in a more controlled and predictable manner, while integration with DCIM or BMS platforms also improves visibility into thermal behaviour at the rack level.

9) RDHx provide a pragmatic bridge towards liquid cooling

Direct liquid cooling technologies continue to gain momentum, but many operators have not yet adopted them at scale.

RDHx offer a practical transition point. By introducing water-based cooling into the white space in a controlled manner, they help build operational confidence and prepare infrastructure for future liquid-cooling architectures.

This creates a clear migration pathway: starting with RDHx deployment, progressing to integration with existing water loops, and culminating in direct-to-chip cooling as requirements change.

10) Extending the life and value of existing data centers

Operators refurbishing legacy facilities face the challenge of not only improving efficiency but also enabling scalable growth without high cost or disruption.

RDHx address this challenge by offering a modular, scalable, technically robust, and low-risk cooling strategy that integrates with existing infrastructure.

By improving thermal control, recovering white space, supporting higher densities, and enabling a clear path towards liquid cooling, RDHx provide a practical pathway to modernizing legacy data centers. This supports today’s workloads while preparing for future AI, HPC, and next-generation compute demands.

To learn more about the capabilities of Legrand’s ColdLogik RDHx, visit this website