Home/Data center/AI POD & liquid cooling

AI POD and rack-levelliquid cooling.

The busway, liquid cooling piping, cable trays and aisle containment that normally take months to build above the racks, prefabricated as one steel-framed pod and lifted into place. With WT-CDU rack-mounted units, liquid cooling deploys one rack at a time.

WincoTek AI POD prefabricated overhead infrastructure above a rack row
Racks / pod
42–46
Per rack
Up to 250 kW
Per pod
2.2–8.0 MW
WT-CDU
30–150 kW
AI POD · prefabricated overhead infrastructure

Move the work above the rack into the factory.

Installing the frame that supports IT, power and cooling often takes months, with schedule delays, complex multi-trade coordination and material waste of 10–15%. What remains on site with the AI POD is lifting, connection and acceptance testing.

01

Faster deployment

Factory prefabrication replaces months of overhead on-site construction.

02

Lower capex

Reduced by at least 15% versus conventional on-site build.

03

More usable space

Modular design practice unlocks up to 30% of space.

04

One purchase

One integrated system from one supplier, instead of sourcing busway, piping and containment separately.

AI POD end elevation with numbered components above a contained aisle

System architecture

The pod frame sits above the rack row and carries every overhead service. The contained aisle below keeps its double glass door and rack line unchanged.

  • 1  Intelligent busway system
  • 2  Technical water loop
  • 3  Steel framework
  • 4  Cable tray
  • 5  Double glass door
  • 6  Racks
AI POD specifications

At a glance.

Pod length, busway architecture (2Bus / 4Bus / cable), cooling layout and CDU interface are engineered per project.

ParameterSpecification
Racks per pod42–46 rack spaces (600–1200 mm racks)
Rack power densityUp to 250 kW per rack
Unit capacity2.2–8.0 MW per pod
BuswayUp to 4000A (2Bus) / 6300A (4Bus), up to 600V, 3Ph + PE
Liquid coolingUp to 5 MW thermal capacity per pod; DN250 main header; 250+ kW per branch
Frame load capacityUp to 1360 kg/m total unit weight capacity (4Bus)
Seismic designUp to Zone 4
DeploymentFactory pre-integrated; section-by-section delivery; lift-and-set on site

Subsystem detail

SubsystemParameterValue
Structural frameTotal unit weight capacity750 / 915 / 885 lbs/ft (1115 / 1360 / 1315 kg/m) for 2Bus / 4Bus / Cable
Infrastructure weight capacity465 / 605 / 555 lbs/ft (690 / 900 / 825 kg/m)
Load per unitUp to approximately 29,000 lbs of data center infrastructure
MountingPost-mounted or floor-standing; suspended possible; post height 8 ft (2.4 m)
SeismicDesigns up to Zone 4, pre-designed for 95% of global coverage
BuswayType / busbarsEnclosed channel track busway; 4–8 busbars per row, each run independent
RedundancyN or 2N (2Bus); N, 2N, 4N/3 (4Bus); 3Ph + PE
Tap boxesUp to 4 pcs (4Bus) per rack slot, up to 200A each; finger-safe and hot-swappable
Maintainable power density334 kW concurrently maintainable (standard); up to 345 kW (4Bus)
Liquid cooling (TCS)System / headerStainless steel Technology Cooling System; main header up to 10 in (DN250), ring or straight section
Rack branch1/2-inch interlocking ball valve quick-connect at 600 mm increments; rack isolation and energy control valves
Operating parameters90:10 liquid-to-air; 1.5 LPM/kW; 8 ft/sec; PG25 / EG25 / deionized water
CDU interfaceEngineered to order (ETO) per project
Hot aisle containmentConstructionPolycarbonate paneling with sliding doors; adjustable filler panels
Fit / width / ratingRack height up to 52U; hot aisle 6 ft (1.8 m); Category C standard, D and E by request
Monitoring & lightingProtocols / pointsModbus, BACnet, TCP, PLC; pressure, flow, temperature, leak detection
Lighting / trays40FC, 80CRI, 4000K; up to 3 tiers of cable tray per row
Configuration options

Three power architectures.

Choose by redundancy topology and how much the row will change over its life.

2Bus busway
4000A
  • N or 2N redundancy
  • Up to 600V, 3Ph + PE
  • 1115 kg/m total unit weight capacity
  • Mixes storage, networking and liquid-cooled AI compute in one row
4Bus busway
6300A
  • N, 2N or 4N/3 redundancy
  • Up to 4 tap boxes per rack slot, 200A each
  • Up to 345 kW concurrently maintainable
  • 1360 kg/m total unit weight capacity
Cable option
6300A
  • 2.2–8.0 MW depending on voltage and redundancy
  • Protection centralized upstream in RPP, PDU or switchgear
  • Top of rack or back of rack connection
  • For static, fixed-load, ultra-dense deployments
WT-CDU Series · rack-mounted CDU

Liquid cooling, one rack at a time.

Localized, high-density liquid cooling delivered directly to the rack for AI servers, intelligent computing clusters and HPC. Available in 4U, 6U, 8U and 12U form factors, for individual cabinets or cabinet rows.

WincoTek WT-CDU rack-mounted coolant distribution unit, interior view

Features

  • Space-optimized: rack-level integration, no dedicated mechanical space
  • Localized distribution: short piping routes; faults contained at the rack
  • Modular & scalable: add cooling capacity alongside new cabinets
  • Carrier-grade reliability: N+N redundant pumps, dual power feeds, leak detection, anti-condensation control
  • Fluid integrity: high-purity stainless steel piping and precision filtration
  • Smart & connected: Modbus, TCP/IP and UDP for DCIM and BMS integration
Rack-mounted liquid-cooled CDU connected to racks and the system cooling source

Each rack carries its own CDU; the primary loop runs to the system cooling source.

Technical parameters

Five models, 30 to 150 kW.

Specifications are subject to project configuration. Model, interface and coolant are matched to the server and the site.

ParameterUnitWT-CDU-Y30WT-CDU-Y60WT-CDU-Y80WT-CDU-Y100WT-CDU-Y150
Rated heat transfer capacitykW306080100150
Primary: circulating medium/Cooling water
Primary: design inlet / outlet temperature°C35/45
Primary: circulating flow ratem³/h2.85.67.59.314
Primary: interfacemmDN25DN32DN40DN40DN50
Primary: water pressure dropkPa≤80
Secondary: circulating medium/Demineralized water / 25% propylene glycol (PG25)
Secondary: design inlet / outlet temperature°C50/40
Secondary: inlet / outlet temperature range°C20-50
Secondary: flow ratem³/h2.85.67.59.314
Secondary: interfacemmDN25DN32DN40DN40DN50
Secondary: total headm15
Rated electric powerkW0.71.51.82.23
Weight in operationkg708596116190
Maximum working pressureBar5
Dimensionsmm447 × 178 × 851447 × 267 × 920
Application scenarios

Sized for accelerated compute.

The AI POD serves full rows at hyperscale density; the WT-CDU brings liquid cooling to individual racks.

AI infrastructure

AI server deployments and intelligent computing clusters.

Hyperscale & colocation

Indoor hyperscale data centers and colocation facilities.

HPC & research

High-performance computing, research and supercomputing centers.

Cloud, edge & enterprise

Cloud and edge data centers and enterprise data centers adding liquid cooling.

Projects

Liquid-cooled builds in progress.

Containerized liquid-cooled data centers currently in progress.

In progress

7.2 MW liquid-cooled containerized

6× 1.2 MW B200/B300 GPU cluster blocks: 12 IT PODs, 14 electrical PODs and 4 auxiliary PODs, 72 liquid racks, 2N for IT and chillers, 10-minute Li-ion backup.

In progress

40 MW liquid-cooled containerized

16× 2.5 MW compute blocks at 50 kW per rack, N+1 power and N+1 cooling.

Send IT load, rack count, kW per rack, climate and power supply.

We return a matched configuration with layout and lead time.

WhatsApp