Hayagreeva HEx is a direct-to-chip cold plate built for next-generation accelerators — rated for continuous heat removal above 5,000 W.
Rated performance
5,000+ W at the package
Continuous heat removal at 0.005 °C/W — capacity stated at the die, not averaged across the rack.
Energy path
Chip to facility rejection
Chip → HEx → CDU → rejection. One engineered path moves waste heat off the accelerator and out of the building.
Hayagreeva HEx
A liquid-cooling powerhouse
A cold plate mounts directly on a GPU, CPU, or accelerator package. Liquid flows through the plate and transports thermal energy to a coolant distribution unit, then onward to the facility heat-rejection system.
Direct-to-chip liquid cooling concentrates capacity at the highest-heat components in AI systems — where conventional air cooling cannot keep pace with rack power density.
Energy path
From chip heat to facility rejection
Hayagreeva HEx sits at the interface where semiconductor waste heat becomes manageable fluid energy. Thermal load moves off the die, through the rack, and out of the building — enabling sustained accelerator performance without thermal throttling.
Lower chip temperature
Reduces thermal throttling so accelerators sustain peak compute under load.
System-scale efficiency
Performance depends on pressure drop and pump energy across the full loop.
Facility integration
Designed to connect with CDUs, rack manifolds, and building-level rejection.
Copper-matrix technology
Engineered for extreme heat flux
Hayagreeva HEx uses a copper matrix with more than 4,000 m²/m³ of wetted surface area. At 5,000 W with 0.005 °C/W rated thermal resistance, exact results depend on coolant, flow rate, inlet temperature, and package geometry.
| Thermal capacity | More than 5,000 W continuous heat removal |
|---|---|
| Thermal resistance | 0.005 °C/W rated |
| Architecture | Copper matrix with 4,000+ m²/m³ wetted surface |
| Compatibility | NVIDIA, AMD, and Intel accelerator architectures |
| IP | 14 patent filings supporting the thermal design |
Built for high-density AI
Hayagreeva HEx targets data centers, OEM platforms, system integrators, hyperscalers, and sovereign AI programs where rack power density demands direct-to-chip liquid cooling at scale.
Request evaluation