Insights · Cooling

Liquid-cooling readiness for existing rooms

Published November 2026 · Mission Critical Sourcing & Logistics

Most server rooms and data halls were built for air. That is still the norm, but the next server refresh may include equipment that ships with cold plates, quick-disconnect hoses and a coolant spec instead of only an airflow rating. You do not need to convert a room to be ready for that. You do need to know what the room can support before the purchase order is written, because liquid-cooled equipment depends on parts that are rarely on a standard IT bill of materials.

Where adoption actually stands

Liquid cooling is growing from a small base. In Uptime Institute’s 2024 cooling systems survey, 22% of respondents said their organization makes some use of direct liquid cooling, and 61% said they do not use it yet but would consider it. Nearly half of the users said it covers less than 10% of their IT racks (Uptime Institute Cooling Systems Survey 2024). For most operators the near-term question is not a full conversion. It is how to support a few liquid-cooled racks next to air-cooled ones.

ASHRAE’s data center committee recommends that any future data center design include the capability to add liquid cooling (ASHRAE TC 9.9, Emergence and Expansion of Liquid Cooling in Mainstream Data Centers). For an existing room, readiness means checking six things.

1. Know the parts of a liquid-cooled system

A direct-to-chip system has three layers. Cold plates in the server capture the heat. A coolant distribution unit (CDU) circulates coolant through the racks and controls its temperature, pressure, flow and filtration. A heat rejection path moves the heat out of the building. Schneider Electric describes these three elements and six common ways of combining them in its white paper on liquid cooling architectures (Schneider Electric White Paper 133).

The loop between the CDU and the servers is called the technology cooling system (TCS). The building’s water loop, where one exists, is the facility water system (FWS). The CDU keeps the two separate.

2. Decide where the heat will go

This is the biggest fork in the road for an existing room.

  • Liquid-to-air CDUs release the heat into the room, and the existing air cooling removes it. Schneider calls these the easiest to add to a conventional air-cooled data center and often more cost-effective for deployments of about one to ten racks (Schneider Electric White Paper 210). The trade-off is that the room’s cooling must still have capacity for that heat.
  • Liquid-to-liquid CDUs transfer the heat to facility water. They scale to much larger loads, but they need a water loop at the right temperature near the racks.

ASHRAE describes liquid-to-air heat exchangers as a reasonable interim step for rooms that cannot yet bring water to the racks, with facility water headers added later. It also cautions that liquid-to-air exchangers have higher approach temperatures, which can push the room toward colder water sooner.

3. Check facility water temperature, if you have it

ASHRAE classifies facility water by its maximum supply temperature: W17, W27, W32, W40, W45 and W+, with the number giving the limit in degrees Celsius (ASHRAE Thermal Guidelines reference card). W17 and W27 typically mean chillers and cooling towers. W32 and above can often run without chillers in many locations.

Before you buy, compare the server manufacturer’s required water class with what your plant can deliver year-round. In Texas, summer conditions are the test. ASHRAE’s paper assumes facility water temperatures will need to fall over time as chip power rises, so a room that only just meets today’s spec may not meet the next generation’s. Schneider notes that a chiller plant shared between air and liquid cooling leaves a narrow range of setpoints, and suggests comparing a shared plant with a separate higher-temperature plant over ten years or more.

4. Walk the room: piping, space and floor load

  • Piping route: ASHRAE recommends keeping overhead piping above aisles, with drip pans, leak detection and drains where connections pass over critical equipment. Power, cooling and network routing need to be coordinated.
  • Space: floor-standing CDUs, manifolds and wider aisles for rear-door heat exchangers all take room.
  • Weight: dense liquid-cooled racks are heavy. ASHRAE notes that some older rooms will find floor loading, not floor space, is the limit, and that ramps, thresholds and elevators on the delivery path need ratings for rolling loads.
  • Power: the same paper notes that existing circuits may be enough for small systems, but larger ones may justify higher distribution voltages. Rack PDU sizing changes with them. See our rack PDU selection guide.

5. Treat coolant and materials as part of the spec

Cold plates have fine channels that can clog. Schneider’s white paper on direct liquid cooling challenges recommends following the IT manufacturer’s list of compatible wetted materials (every part the coolant touches, from CDU to hose to cold plate), avoiding aluminum in a copper and brass loop unless inhibitors are used, and keeping a registry of the materials in the loop. In Schneider’s words, these guidelines “should be strictly followed to prevent technical issues and maintain warranty coverage.” The paper also covers filtration, flushing the loop before connecting equipment, purging air, and assigning an owner for coolant monitoring.

For a buyer, this means fittings, hoses, filters and coolant are not interchangeable commodities. Mixing parts from different vendors without checking compatibility can affect warranties. Confirm early which vendor is responsible for each part of the loop.

6. Plan leak detection and failure scope

Schneider notes that leaks tend to occur at couplings and joints rather than along pipe walls. It recommends both direct sensors (cable and spot detectors at connections and drip pans) and indirect monitoring, all reporting to a DCIM or building management system. For operators new to liquid cooling, it suggests alerts with manual response rather than automatic shutdowns. It also recommends limiting how many servers share one loop, so a single CDU or contamination problem affects a smaller part of the room.

A readiness checklist

  • Server models planned, their cooling method and the required water or coolant class
  • Heat path: into the room (liquid-to-air) or to facility water (liquid-to-liquid), and remaining room cooling capacity
  • Facility water supply temperature through a Texas summer, if a water loop exists
  • Piping route, CDU location and space for manifolds and drip pans
  • Floor loading and the delivery path for heavy racks
  • Rack power and PDU capacity at the planned density
  • Wetted materials list, coolant spec and responsibility for each part of the loop
  • Leak detection, monitoring integration and how many servers share each loop

With those answers, a supplier can match CDUs, manifolds, hoses, rack PDUs and monitoring to the servers you plan to buy, and flag mismatches before equipment arrives.

This guide summarizes publicly available standards-body, manufacturer and industry guidance for planning purposes. It is not mechanical or electrical design advice. Have a qualified engineer confirm cooling, piping, structural and electrical requirements, and follow the IT and cooling equipment manufacturers’ instructions.

Send the server models and room details. We will source the fit.