Supercritical CO2 Experimental Cart for Cooling AI Data Center Chipsets

Supercritical CO₂ Experimental Cart for Cooling AI Data Center Chipsets

Introduction

The rapid growth of artificial intelligence workloads has pushed data centers to the limits of existing cooling technologies. High-performance GPUs and server racks generate massive amounts of heat that must be dissipated efficiently and reliably. Supercritical carbon dioxide is emerging as a promising fluid for next-generation cooling. A cart-mounted experimental test system provides a safe and modular platform for evaluating small-scale concepts before scaling up to production-ready designs.

Cooling Applications with Supercritical CO₂

Ejector Cooling

Supercritical CO₂ ejectors use pressure differentials to generate local zones of cooling. These systems can be compact, with no moving parts, making them attractive for immersion cooling or chip-level cooling where simplicity and reliability are essential.

Cascade Cooling Systems

By staging multiple CO₂ cooling loops, cascade systems can reach very low temperatures or provide distributed cooling across multiple rack zones. A test cart allows controlled trials of cascading pressure stages and heat exchangers.

Joule Thomson Effect Cooling

The Joule Thomson effect, in which expanding CO₂ through a throttling device causes rapid cooling, is well-suited for chip-level spot cooling. Experimental setups can test miniature throttling valves, orifices, and expansion paths to measure achievable cooling power.

CO₂ as a Traditional Chiller

Supercritical CO₂ can also function in conventional chiller-style systems, replacing water or refrigerants. On a small cart system, researchers can evaluate CO₂ condenser and evaporator designs, pressure management strategies, and overall energy efficiency before scaling to large data center chillers.

Additional Possibilities

Spray or jet impingement cooling directly onto chip surfaces.

Two-phase cooling loops that leverage CO₂ phase changes for high heat flux removal.

Hybrid CO₂ and liquid metal systems for extreme GPU cooling requirements.

Advantages of a Supercritical CO₂ Test Cart

Safe Small-Scale Testing: Engineers can trial miniature systems at manageable pressures and flow rates before investing in large prototypes.

Flexibility: The cart allows quick reconfiguration to test ejector setups one day, and Joule Thomson expansion the next.

Instrumentation: Precise pressure, temperature, and flow measurements make it possible to map performance curves and validate thermal models.

Scalability: Data collected from small test sections informs designs for full-scale data center rack systems.

Conclusion

As AI accelerators and GPUs continue to push thermal boundaries, traditional water and air cooling are reaching their limits. Supercritical CO₂ offers unique opportunities for compact, efficient, and versatile cooling technologies. A laboratory-scale test cart provides a vital stepping stone, enabling experimentation with ejectors, cascades, Joule Thomson systems, and chiller-like designs in a controlled environment. By refining these concepts at small scale, engineers can accelerate the development of practical cooling solutions for tomorrow’s data centers.


INFINITY TURBINE LLC We specialize in designs, plans, licensing, consulting, design services, and surplus spare parts. We no longer manufacture turbines or CO2 systems. More Info...

TEL: +1-608-238-6001 (Chicago Time Zone ) USA

Email: greg@infinityturbine.com

CO2 Ejectors for Data Center Waste Heat Cooling: A Technical Review Hyperscalers are racing to deploy gigawatts of AI compute, but the grid can't keep up and large gas turbines are backordered half a decade out. Infinity Turbine's Cluster Mesh Supercritical CO₂ system offers a radical alternative: modular, silent, trailer-deployable prime power that scales the way software does... More Info

Turning AI Data Center Waste Heat Into Electricity More Info

Cooling a Data Center Pod with Its Own Waste Heat: A 100 kW Ejector Design Study More Info

Portable Gas Turbine Prime Power AI data centers are pushing electrical infrastructure to its limits. The traditional AC power chain is no longer optimal for GPU-driven workloads. A DC-native architecture using Infinity Turbine’s Cluster Mesh system offers a path to higher efficiency, lower costs, and scalable modular power—potentially saving tens of millions per year at hyperscale... More Info

SMR and Cluster Mesh Supercritical CO2 Power System for Data Centers and AI Pairing Cluster Mesh Supercritical CO2 Power System with Small Modular Reactors enables hyperscalers to convert high-grade nuclear heat into ultra-efficient, dispatchable power with a compact, modular footprint tailored for AI-scale demand. More Info

ORC and Products Index Infinity Turbine ORC Index... More Info

AMD Helios Cooling Strategy Using Transcritical CO2 and a Cluster Mesh sCO2 Turbine Generator Cell More Info

CONTACT TEL: +1-608-238-6001 (Chicago Time Zone USA) Email: greg@infinityturbine.com (Standard Web Page) | PDF