Johnson Controls Launches Thermal Designs for Gigawatt-Scale AI Data Centres

Johnson Controls has released reference designs for thermal management in AI data centres, addressing water-cooled, air-cooled, and absorption chiller use cases.

As AI systems push data centres towards gigawatt scale, Johnson Controls (JCI) has released a series of Reference Design Guides for 1 Gigawatt AI data centres, mapping the full thermal chain and offering cooling architectures tailored to compute densities and geographical factors. The initial guide details water-cooled chiller plants, with subsequent guides slated to cover air-cooled and absorption chiller solutions.

These designs align with the NVIDIA DSX reference architecture, aiming for scalable deployment of 1-GW-class AI factories.

The reference designs target several key performance indicators:

• Zero Water Consumption: through the use of dry coolers.
• Future-proofed thermal flexibility: via high-temperature TCS loop readiness.
• Optimised AI performance: achieved through alignment with the NVIDIA DSX architecture.

Elevated condenser water temperatures, bifurcated loops, and YORK high-lift chillers contribute to industry-leading Power Usage Effectiveness (PUE) and improved annualised efficiency.

AI Factories are production facilities – the places where intelligence is manufactured at an industrial scale – said Austin Domenici, vice president & general manager, Johnson Controls Global Data Center Solutions.

These reference designs arrive as the demand for high-density computing infrastructure intensifies, particularly for AI applications.

Learn more about Johnson Controls's Reference Guide Series at www.johnsoncontrols.com/industries/data-centers/reference-designs.