Johnson Controls, a global leader in smart, healthy, and sustainable building solutions, has released its second AI Factory Reference Design Guide, this one focused on air-cooled chillers, as the Milwaukee-based company works to build out what it describes as the industry's most comprehensive series of thermal design blueprints for large-scale artificial intelligence infrastructure.

The water-cooled chiller guide, released in February, is intended to give data center designers and operators a globally repeatable framework for cooling facilities up to one gigawatt in scale. Johnson Controls has said additional guides covering absorption chillers and direct-to-chip liquid cooling are planned.

What the Guide Covers

The new reference design outlines a thermal cooling architecture built around high-efficiency air-cooled YORK centrifugal chillers, specifically the company's YDAM (YORK YVAM Magnetic Bearing Air-Cooled Centrifugal Chiller) and YVA (YORK YDAM Air-Cooled Direct-Drive Optimization Centrifugal Chiller) models, alongside fan coil walls and coolant distribution units.

The system is designed to manage both air-cooled and liquid-cooled information technology loads within a single integrated framework.

Johnson Controls says the guide includes sizing references for 220 megawatt compute clusters and provides recommended design temperatures and operating conditions across each stage of the thermal chain.

The company positions the guide as adaptable to different climates, workloads, and growth trajectories, which it says is a necessary feature given the variability in how and where AI infrastructure is being deployed globally.

Energy and Efficiency Claims

The company outlines several specific performance outcomes it says the design enables. Chief among them is a 32 percent improvement in annual energy consumption, which Johnson Controls attributes to the intelligent use of redundant chillers within the system architecture.

The design also claims to return up to 50 megawatts of power to the AI factory itself through the implementation of bifurcated loops that separately handle air-cooled and liquid-cooled systems.

Johnson Controls additionally states the design achieves peak power savings of 20 megawatts by quantifying and mitigating heat island effects associated with air-cooled chiller plants, a known challenge when dense arrays of cooling equipment are packed into limited footprints.

Raising the chilled water temperature to support warm-water Technology Cooling System loops, the company says, produces a 30 percent improvement in Coefficient of Performance and allows operators to use 27 percent fewer chillers compared to conventional approaches.

Water Elimination as a Design Priority

One of the more prominent claims in the guide concerns water usage. Johnson Controls says the air-cooled design supports zero water consumption for cooling by eliminating the need for cooling towers entirely.

The company states this approach saves more than 12 million gallons of water per day, a figure it frames as increasingly significant as data center operators face regulatory and community pressure around water availability in many regions.

The elimination of cooling towers is presented not only as an environmental consideration but also as a factor in reducing noise impact on surrounding communities, which Johnson Controls identifies as one of the mounting challenges operators face as AI facilities expand in scale and number.

The Broader Context of AI Factory Design

Johnson Controls frames the guide series against a backdrop of what it describes as a fundamental shift in how data center infrastructure must be conceived and built.

The company points to the power requirements of cooling systems, rising cooling-loop temperatures, efficiency losses from heat islands, and limited water availability as the central problems operators are trying to solve.

Austin Domenici, president of Johnson Controls Global Data Center Solutions, said in the announcement that gigawatt-scale AI factories require a different way of thinking about infrastructure altogether.

He described the guide as reflecting the company's effort to help customers plan holistically for AI growth, from design through operations, across geographies.

The scale referenced throughout the guide is notable. A one-gigawatt AI factory represents a level of energy consumption that would have been considered extraordinary for a data center just a few years ago, but is now being discussed as a realistic planning horizon as hyperscale operators and AI companies expand their compute footprints aggressively.

Where the Series Is Headed

Johnson Controls has not announced a timeline for the remaining guides in the series covering absorption chillers and direct-to-chip liquid cooling, but the company has positioned the overall series as an evolving resource. The first guide, covering water-cooled chillers, was released in February, making the air-cooled chiller guide the second installment in roughly three months.

The full series is being positioned as a resource for designers and operators who need to plan infrastructure that can scale predictably while meeting efficiency and sustainability targets. Johnson Controls says the guides are available through its website under its data center reference designs section.

Power Your Data Center Pipeline with Intelligence That Spans Markets Worldwide

As hyperscale investment accelerates and colocation demand reaches new heights across markets worldwide, the organizations that win are those with the clearest view of what is being built, where, and by whom. Are you working from complete intelligence, or leaving critical opportunities to chance?

The Global Project Tracking (GPT) platform by Blackridge Research was built to give data center developers, contractors, equipment suppliers, and investors a single, authoritative source for project activity across every major region. From land acquisition announcements to ribbon-cutting completions, every stage of the data center lifecycle is captured and continuously updated.

Whether you are targeting new campus developments in emerging markets or tracking the expansion plans of established cloud giants, the platform delivers the structured, timely data your team needs to prioritize outreach, benchmark the competition, and close more business.

• Upcoming Projects

• Tender Notices

• Contract Awards

• Projects Under Construction

• Completed Projects

See the full depth of global data center project coverage for yourself. Book a Free Demo with the Blackridge Research team today and discover how the GPT platform can sharpen your market strategy.