Daikin D60-W Cools AI Data Centers Mostly With Liquid

Daikin Applied has drawn the cooling plant for the next AI data center as mostly liquid. The company announced the D60-W on November 14, 2026, a 60 MW cooling reference architecture for AI data centers built on the NVIDIA DSX AI Factory Platform with NVIDIA Vera Rubin NVL72 racks. It handles most of the heat with liquid, and it hasn't shipped yet.

The split is the specification that matters. The design assigns 92 percent of the load to liquid cooling and 8 percent to air, across a six-unit architecture. Two separate loops do the work: a high-temperature liquid loop serving coolant distribution units that carry heat away from the chips, and a low-temperature loop serving air-cooling rack containment units for the remaining load.

Heat leaves the building without evaporative towers. The D60-W rejects heat through dry coolers and pairs them with oil-free magnetic-bearing centrifugal chillers, the Magnitude WME line from Daikin's own portfolio, rated up to 10 MW. The system shifts between dry, hybrid and full mechanical cooling with outdoor conditions: in dry mode the chiller compressors switch off, and in hybrid mode waterside economizers pre-cool return water so the chillers handle only the leftover load.

One supervisory layer runs the plant. Daikin calls it iDCM, and it coordinates the containment units, distribution units, chillers and dry coolers as a single system rather than as separate machines with separate controls. That's the architectural claim: the cooling plant is designed as one coordinated unit sized to the rack platform it serves.

The limits deserve the same plain statement as the numbers. This is a manufacturer reference design, so the 60 MW capacity and the 92-to-8 split describe intended performance, not measured operation in a running facility. The announcement carries no availability date, no pricing, no customer deployment and no independent benchmark or efficiency comparison.

The next verifiable milestone is a name. Reference designs become meaningful when an operator builds one and runs it at scale, and neither Daikin nor NVIDIA has announced that customer. Until one appears, the D60-W is a blueprint that shows where AI cooling is heading: liquid first, air for the remainder, and dry coolers doing the work evaporative towers used to do.