Johnson Controls has released its second AI Factory Reference Design Guide, focusing on air-cooled chiller systems designed for large-scale artificial intelligence data centres. The company stated that the blueprint can support facilities of up to 1 GW while improving thermal efficiency, reducing energy consumption, and eliminating water usage.
As artificial intelligence workloads continue to reshape global data centre infrastructure, cooling systems have emerged as one of the most critical engineering challenges. The latest guide builds on Johnson Controls’ earlier water-cooled chiller framework released in February 2026, with additional reference designs planned for absorption chillers and direct-to-chip liquid cooling systems.
The new architecture is designed to address the rapidly increasing power density of AI workloads, where traditional cooling systems face limitations due to higher rack densities, rising thermal loads and growing concerns over water scarcity.
The reference design integrates YORK centrifugal chillers, including YDAM and YVAM systems, along with fan coil walls and coolant distribution units. It supports both air-cooled and liquid-cooled IT environments and provides sizing guidelines for compute clusters of up to 220 MW, along with operational recommendations across the cooling lifecycle.
A key highlight of the design is its zero-water cooling approach, which eliminates the need for cooling towers. The company estimates this could save more than 12 million gallons of water per day in large-scale AI facilities, addressing increasing concerns around water stress in hyperscale infrastructure regions.
The guide also proposes strategies to reduce the heat island effect associated with large air-cooled systems, with potential peak power savings of up to 20 MW.
Johnson Controls further outlined expected efficiency gains from the system, including up to 50 MW of energy recovery through integrated air and liquid cooling loops, a 32 per cent improvement in annual energy consumption, a 30 per cent improvement in Coefficient of Performance (COP), and a 27 per cent reduction in chiller requirements through higher chilled water temperature operations.
The company noted that the shift toward AI-driven infrastructure is accelerating demand for scalable, energy-efficient cooling systems capable of handling extreme thermal loads without significantly increasing operational costs.
Austin Domenici, President of Global Data Center Solutions at Johnson Controls, said, “At the gigawatt scale, AI factories require a fundamentally different way of thinking about infrastructure. The future requires designing integrated systems that can scale predictably, perform efficiently, and adapt as technology evolves.”
The company added that its AI Factory Reference Design Guide series is intended to help operators build future-ready data centres capable of adapting to evolving workloads, regional climate conditions, and long-term expansion requirements, as demand for AI infrastructure continues to grow globally.
Send news announcements/press releases to:
editor@thefoundermedia.com
About The Author
