japanma.net
Field TestedCapability Asset

Turnkey liquid cooling infrastructure for AI data centers

End-to-end liquid cooling infrastructure for AI data centers including 1MW-class cooling systems, containerized mechanical rooms, chillers/dry coolers, secondary-side piping, and 1MW-class load bank pre-validation testing. Modular design for rapid deployment.

Problem solved

Traditional data center cooling (air conditioning) becomes insufficient for high-density AI GPU racks (>40 kW/rack). Operators need a complete liquid cooling infrastructure — not just cold plates — including heat rejection, pumping, water treatment, and commissioning. Most lack in-house liquid cooling engineering expertise.

Application contexts

  • AI training data centers (40+ kW/rack density)
  • HPC facility upgrades from air-cooled to liquid-cooled
  • Containerized / edge data centers
  • Existing building retrofits for AI workloads

Technical principle

Kannetsu provides a modular liquid cooling infrastructure built around container-type mechanical rooms (40ft container housing pumps, water treatment, controls) coupled with chillers or dry coolers for heat rejection. The secondary-side piping delivers coolant to server racks. Pre-validation is performed using 1MW-class electrical load banks that simulate server heat output.

Typical use cases

  • 1MW liquid cooling system for 20-25 racks of NVIDIA GB200 NVL72 servers
  • Containerized cooling unit for edge data center (<2MW)
  • Secondary-side piping retrofit for existing building with available chilled water
  • Load bank testing of cooling system before server installation

Partnership route

How global engineering teams typically engage this asset.

01Technical Interview
02Feasibility Review
03Pilot Production
04Joint Development

Helpful to include: Total cooling capacity required (MW) · Rack density target (kW/rack) · Available space (indoor/outdoor, rooftop, container) · Coolant type preference (water, glycol, dielectric fluid) · Existing chilled water infrastructure (if any) · Project timeline and installation constraints

Have specific parameters to run against this asset?

Submit a Challenge Brief

Provider

Source & provenance
  • Source: AIデータセンター冷却設備 | カンネツ株式会社
  • Original language: JA
  • Summary AI-assisted; translation status: Unreviewed