Nidec Shifts Data Centre Cooling to In-Row CDUs
The Japanese manufacturer is pushing modular, stackable cooling units as AI computing platforms drive higher thermal loads.
Updated on Nov. 20, 2026 in Data Centers

Nidec Corporation is moving its data centre cooling strategy toward in-row cooling distribution units and modular stackable configurations. The company is also engaging data centre operators directly more often while continuing to work with server manufacturers.
Why it matters
Newer AI platforms such as the Vera Rubin generate heat that conventional air cooling struggles to remove, increasing demand for liquid cooling infrastructure. Changing requirements are expected to affect how cooling equipment is specified, purchased and installed.
Nidec's stackable configuration combines multiple 250 kW CDUs to reach capacities of 1 MW or 1.25 MW, compared with roughly 250 kW for a single in-rack unit and two megawatts for its larger in-row system. Adoption rates and pricing for these configurations remain unreported.
The players
Nidec Corporation
Nidec Corporation is a Japanese manufacturer of electric motors and precision components that has expanded into data centre cooling systems.
Nozawa
Nozawa is Deputy Head of the AI & IT Division SPMS at Nidec Corporation, and he expects changing requirements to shape how cooling equipment is specified, purchased and installed.
The details
Nidec is seeing a shift toward in-row cooling distribution units, which can deliver substantially greater capacity than in-rack systems, historically the primary liquid cooling form factor. To serve customers needing intermediate capacity, the company is developing modular configurations that combine smaller units, and it is now speaking more frequently with data centre customers directly.
Timeline
Historically, in-rack CDUs served as a primary form factor for liquid cooling systems.
Next-generation AI computing platforms are expected to change how cooling equipment is specified and purchased.
The Tech Race
The move reflects the broader pressure on infrastructure suppliers as Nvidia's Vera Rubin platform raises heat output beyond what air cooling can handle. Suppliers offering both large in-row systems and stackable modular units are positioning for that transition.
Data centre operators planning AI capacity may need to specify different cooling equipment, with configurations ranging from 250 kW units to multi-megawatt in-row systems. Operators evaluating these options should account for how modular units can be combined to meet intermediate capacity needs.
The takeaway
Cooling capacity is becoming a central design constraint for AI infrastructure, and modular liquid cooling offers a way to scale capacity in measured steps. Planners should treat cooling form factor as a purchasing decision alongside compute hardware.
Further reading
For broader context on infrastructure demands, explore the Data Centers section.
Source note: This article includes information reported by SecurityBrief Asia.



