Daikin Applied Details 60 MW Water-Cooled Cooling Blueprint for NVIDIA DSX-Based AI Factories

Daikin Applied introduced Reference Architecture D60-W Water-Cooled Chillers with Dry Coolers, a 60-megawatt cooling design for AI factories built on the NVIDIA DSX AI Factory Platform with NVIDIA Vera Rubin NVL72.

Minneapolis announcement pairs chillers and dry coolers with Vera Rubin NVL72 systems

The blueprint is aimed at data center developers, engineers and operators who need a starting point for cooling design that spans from the rack to ambient heat rejection, tying cooling equipment and controls into a single connected system.

The company positioned the release around the growing complexity of cooling at AI factory scale. Coordinating rack requirements, facility equipment, heat rejection systems and controls often involves multiple teams and suppliers, and D60-W consolidates those elements into one design. According to Daikin Applied, that consolidation reduces integration complexity and lets customers move into project engineering sooner.

Oil-free Magnitude chillers cut standby maintenance and restart delays

At the core of the architecture are Daikin Applied's Magnitude WME magnetic bearing centrifugal chillers, which are oil-free and offer capacities up to 10 MW. In suitable climates, dry coolers carry most of the cooling load through the year, with the chillers supporting operation on the hottest days.

The company contrasted this with traditional large-scale chillers, whose oil lubrication demands maintenance even during extended standby, including oil management and periodic run cycles. Because the oil-free design removes oil heaters, pumps and standard servicing, standby energy use and maintenance drop. Daikin Applied also states that it eliminates oil-related startup delays when chillers come out of standby, allowing cooling to recover quickly and protecting GPU availability.

"AI factories are designed to turn energy into output as efficiently as possible, and cooling infrastructure sits right at the center of that challenge," said Yu Nishiwaki, president and CEO of Daikin Applied. "D60-W draws on the full depth of Daikin Applied's portfolio, giving customers a single partner as they scale AI operations.

" Karim Shalash, director of Data Center Solutions at Daikin Applied, described the design decisions as interconnected. "An AI factory is a system of systems," he said. "A cooling decision at the rack carries implications for water temperatures, flow requirements, plant operation and heat rejection. D60-W gives engineering teams a framework for working through those dependencies and developing a cooling architecture around the facility's operating requirements."

92 percent of the 60 MW load handled by liquid, split across two loops

D60-W supports a 60 MW IT load across six deployable units. Liquid cooling accounts for 92 percent of that load, with air cooling covering the remaining 8 percent. The design uses two temperature loops: a high-temperature liquid loop serves the coolant distribution units, while a dedicated low-temperature loop serves the air-cooling rack containment units, an arrangement the company says improves overall plant efficiency.

At the rack level, racks are primarily liquid cooled through CDUs, while Daikin Applied's DDC Solutions S-Series rack containment units capture residual heat and send it directly to the low-temperature loop. Every rack includes fire suppression and water-loop protection, and operators can run air-cooled, liquid-cooled and hybrid compute configurations side by side.

Redundancy layered from rack to dry cooler

Reliability is addressed at several levels. At the equipment level, Daikin Applied's RideThrough and RapidRestore chiller technologies handle brief power disturbances and restart after shutdown. At the system level, redundancy across the cooling system, from rack to dry cooler, is intended to maintain uptime through equipment outages and planned maintenance.

Dry coolers replace evaporative towers to cut water use

Water and energy consumption are managed through mode switching driven by outdoor conditions. The architecture moves between dry cooling, hybrid cooling and full mechanical cooling. In dry cooling mode, chiller compressors are off. In hybrid mode, waterside economizers precool return water while chillers take the remaining load. Heat rejection is handled by dry coolers rather than evaporative cooling towers, which removes cooling-tower make-up water and supports a low water usage effectiveness, or WUE.

iDCM supervisory layer runs the plant as one system

Daikin Applied's iDCM supervisory layer coordinates RCUs, CDUs, chillers and dry coolers as a single plant. It manages load sharing, balances equipment operating hours and controls transitions between the three operating modes.

Built-in redundancy and integration with site management systems add control resilience and centralized visibility, while energy optimization and predictive maintenance capabilities support plant efficiency and equipment reliability.

Capacity scales unit by unit, with new CDU and dry cooler due in late 2027

D60-W scales by deployable unit, so project teams can add capacity as demand grows, supported by Daikin Applied commissioning, planned maintenance and 24/7 emergency response

. The Daikin Applied CDU and data center dry cooler are under development, with availability targeted for the second half of 2027. The full reference architecture is available through Daikin Applied's website, where customers can also discuss applying it to specific projects. Daikin Applied, a member of Daikin Industries, named to the 2026 Fortune Global 500, designs and manufactures commercial and industrial HVAC systems sold through a global network of dedicated sales, service and parts offices.