Data Center Thermal Management

Liquid Cooling Components for AI and High-Density Computing

From GPU cold plates and manifolds to loop components and partner-sourced coolant distribution units, we help server OEMs, ODMs and cooling integrators source and specify direct-to-chip liquid cooling hardware.

Overview

Why liquid cooling for AI infrastructure

Accelerator and processor power continues to rise, and rack densities in AI and HPC environments are moving beyond what air cooling can practically support. Liquid carries heat far more effectively than air, allowing heat to be captured directly at the chip and removed from the rack.

Our data center work is structured in deliberate stages: we start with cooling components, build toward engineered assemblies, and support partner-integrated systems where a qualified manufacturing or integration partner is involved. We are clear about which stage each offering is in.

Our approach

Components → Engineered Assemblies → Partner-Integrated Systems

We build data center capability in deliberate stages and are clear about which stage each offering is in.

  1. Stage 1Cooling Components

    Cold plates, manifolds, fittings and loop components sourced and specified to your requirements.

  2. Stage 2Engineered Assemblies

    Manifold, cold plate and loop assemblies built to your drawings with manufacturing partners.

  3. Stage 3Partner-Integrated Systems

    CDUs and system-level solutions delivered with qualified manufacturing and integration partners.

Product categories

Products & Offerings

Specifications shown as "available upon request" are confirmed from supplier documentation for each inquiry. We do not publish unverified performance figures.

  • Component sourcing
  • Custom manufacturing coordination

GPU / CPU Cold Plates

Liquid cold plates mounted directly to GPUs, accelerators and CPUs, transferring heat from the processor package into the coolant loop.

Applications

  • AI accelerator servers
  • High-performance computing
  • GPU clusters
  • High-density server cooling
GPU / CPU Cold Plates — technical specifications
SpecificationValue
MaterialContact us for specifications
Supported hardware configurationDefined per project requirements
Thermal resistanceSupplier-verified value — available upon request
Coolant flow rateDefined per project requirements
Pressure dropSupplier-verified value — available upon request
Operating pressureAvailable upon request
Connection typeDefined per project requirements
Coolant compatibilityAvailable upon request
Leak testingTest requirements defined per project
  • Component sourcing
  • Engineered assemblies

Liquid Cooling Manifolds

Supply and return manifolds that distribute coolant between rack-level infrastructure and individual servers.

Applications

  • Rack-level coolant distribution
  • Server liquid cooling loops
  • High-density computing systems

Offerings

  • Supply manifolds
  • Return manifolds
  • Custom manifold assemblies
  • Connection and distribution components
Liquid Cooling Manifolds — technical specifications
SpecificationValue
MaterialContact us for specifications
Port count & spacingDefined per project requirements
Connection typeDefined per project requirements
Operating pressureAvailable upon request
Coolant compatibilityAvailable upon request
MountingDefined per project requirements
  • Component sourcing

Cooling Loop Components

The components that complete a liquid cooling loop. Availability and lead time are confirmed per inquiry — we do not list unverified inventory.

Component types

  • Quick disconnect fittings
  • Tubing and hoses
  • Pumps
  • Valves
  • Heat exchangers
  • Sensors
  • Coolant distribution components

Specifications by requirementSpecifications are defined against your application and confirmed with supplier documentation.

Contact us for specifications
  • Partner-sourced / project-specific

Coolant Distribution Units (CDUs)

CDUs separate the facility water loop from the technology cooling loop and control flow, temperature and pressure to the racks. IE offers CDUs as partner-sourced or project-specific solutions; we do not manufacture proprietary CDUs.

Applications

  • Rack-level cooling
  • Row-level cooling
  • Data center liquid cooling distribution
  • High-density AI computing environments
Coolant Distribution Units (CDUs) — technical specifications
SpecificationValue
Cooling capacityDefined per project requirements
Flow rateDefined per project requirements
Supply & return temperaturesDefined per project requirements
Pressure dropAvailable upon request
RedundancyDefined per project requirements
Monitoring & controlAvailable upon request
Facility-side interfacesDefined per project requirements

Direct-to-Chip Cooling

How direct-to-chip liquid cooling works

In direct-to-chip cooling, a cold plate is mounted directly on each GPU or CPU. Coolant flows through microchannels in the cold plate, absorbs heat from the processor and carries it through server tubing and rack manifolds to a coolant distribution unit, which transfers the heat to the facility water loop.

Because heat is captured at the source, direct-to-chip cooling supports higher rack densities than air cooling alone and is widely used for AI GPU servers, HPC clusters, high-power processors and high-density computing racks.

How IE supports direct-to-chip projects

  • Component sourcing
  • Cooling loop configuration support
  • Thermal requirement matching
  • Manufacturing partner coordination
  • System integration support through qualified partners, where available
  1. ProcessorGPU / CPU heat source
  2. Cold plateCaptures heat at the chip
  3. Server loop & manifoldDistributes coolant across servers
  4. CDUControls flow, temperature, pressure
  5. Facility waterHeat rejection outside the white space

Technical considerations

What to Specify — and Why It Matters

Heat load & thermal targets

Start from component TDP, allowable case temperature and how much heat must be captured by liquid versus air.

Flow rate & pressure drop

Cold plate and manifold pressure drop must be balanced against pump and CDU capability across every server in the loop.

Coolant chemistry

Wetted materials must be compatible with the selected coolant (e.g. treated water or glycol mixtures) to avoid corrosion and contamination.

Connections & serviceability

Quick disconnect type, sizing and placement determine serviceability and leak risk during maintenance.

Leak testing & quality

Pressure and leak test requirements should be defined up front and confirmed with supplier test reports.

Facility integration

Supply temperature, facility water quality and heat rejection strategy determine CDU selection and system architecture.

Application scenarios

Typical Use Cases

Representative application scenarios — illustrative examples, not customer case studies.

AI server OEM evaluating cold plates

A server manufacturer needs cold plates for a new accelerator platform and wants alternatives with documented thermal resistance and pressure drop data.

Integrator sourcing manifolds

A cooling integrator requires rack manifolds with a specific port count, spacing and connection type for a liquid-cooled rack design.

High-density retrofit

An operator converting high-density racks to direct-to-chip cooling needs loop components and a partner-sourced CDU scoped to its heat load.

Engagement

How a Project Moves Forward

  1. 01

    Thermal requirements

    Heat load, hardware platform, coolant, flow and pressure constraints.

  2. 02

    Component selection

    Cold plates, manifolds and loop components matched to your loop design.

  3. 03

    Partner coordination

    Supplier documentation, drawings and test requirements aligned.

  4. 04

    Samples & testing

    Samples, pressure and leak test reports for your validation.

  5. 05

    Supply & integration

    Production supply, and partner-integrated systems where applicable.

Request a Quote

Share your requirements for data center thermal management and our team will follow up with suitable options and next steps.