Scalable CDU Platform for AI, HPC, and Edge Workloads

As compute densities rise and GPUs push thermal limits, traditional single-phase cooling is falling short. Two-Phase direct-to-chip cooling is emerging as a more efficient solution—offering lower pumping power, higher heat transfer performance, and uniform chip temperatures using safe, dielectric fluids.

At the core of this architecture are ACT’s Coolant Distribution Units (CDUs), which precisely manage heat transfer between facility cooling systems and chip-level cold plates.

With decades of expertise, proven deployments, and leadership in ARPA-E and DARPA programs, ACT delivers intelligent, scalable CDUs that enable safe, energy-efficient cooling across rack, row, and edge environments—helping operators confidently scale for next-gen compute.

Why Two-Phase Direct-to-Chip Cooling Redefines Data Center Efficiency

Two-Phase direct-to-chip cooling leverages the latent heat of vaporization to absorb and transport heat far more efficiently than traditional liquid cooling methods. This enables higher performance with lower mass flow rates, reduced pumping power, and more uniform chip temperatures.

Unlike single-phase systems that depend on large volumes and high flow rates, ACT’s Two-Phase CDUs deliver coolant directly to cold plates mounted on processors and high-power components. The liquid absorbs heat, vaporizes, and returns to the CDU where it condenses and recirculates—creating a continuous, closed-loop cycle that maximizes thermal transfer while minimizing energy consumption.

Key Advantages

• Direct, efficient heat removal at the chip
• Significantly lower pumping power and energy consumption
• Compact infrastructure with smaller fluid loops
• Uniform thermal performance across high-density loads
• Dielectric refrigerants ensure electrical safety and system reliability
• Ideal for AI training, GPU clusters, HPC pods, and edge systems