Johnson Controls Offers GW-Scale AI Data Centre Blueprints

Johnson Controls has launched a new series of reference design guides aimed squarely at the next generation of hyperscale AI data centres, as operators prepare for facilities measured in gigawatts rather than megawatts. 

The Reference Design Guide Series for 1 Gigawatt AI Data Centers sets out detailed thermal management blueprints intended to help owners and designers manage the rising cooling demands of large scale AI workloads.

The guides are designed to map the full thermal chain across different compute densities, geographies and elevations, reflecting the diversity of environments in which AI infrastructure is now being deployed. 

Johnson Controls says the series responds to growing pressure on data centre operators to deliver facilities that balance performance, sustainability and long-term flexibility as AI clusters scale rapidly.

Thermal design for AI at scale

The first guide in the series focuses on water-cooled chiller plants, with future editions planned to address air-cooled and absorption chiller configurations.

Together, the guides are intended to support what Johnson Controls describes as AI factories – facilities built to deliver continuous, industrial-scale AI training and inference.

At gigawatt scales, thermal management becomes a defining constraint. AI accelerators generate significantly higher heat densities than traditional enterprise IT, pushing cooling systems beyond conventional design assumptions. 

Johnson Controls positions its reference designs as a way to standardise best practice while allowing operators to adapt to local constraints such as climate, water availability and grid conditions.

The company says the guides aim to help data centres achieve strong power usage effectiveness and water usage effectiveness metrics while retaining the ability to scale in phases as demand grows.

Mapping the full thermal chain

The initial guide outlines a complete thermal architecture capable of supporting both liquid-cooled and air-cooled IT loads.

This includes integrated computer room air handlers, fan coil walls, coolant distribution units and high-efficiency YORK centrifugal chillers.

Detailed sizing guidance is provided for 220MW compute quadrants, reflecting the modular approach many hyperscalers are taking as they roll out AI capacity in repeatable blocks.

The guide also defines temperature and operating conditions across all major facility loops, including Technology Cooling System loops designed to support next generation GPUs.

By addressing both current and future thermal requirements, the designs are intended to reduce the need for costly retrofits as chip architectures evolve and rack densities continue to rise.

Water, energy and density challenges

A central theme of the series is reducing reliance on water while maintaining performance.

One of the headline outcomes described in the guide is a zero water consumption heat rejection process using dry coolers. 

This approach is aimed at cutting operational costs and supporting sustainability targets, particularly in regions where water scarcity is already limiting data centre development.

The designs also emphasise readiness for higher temperature cooling loops, which are expected to become more common as GPU vendors push for greater efficiency at the chip level. 

Elevated condenser water temperatures, bifurcated loops and the use of high lift chillers are highlighted as ways to improve annualised efficiency while supporting high density AI deployments.

Johnson Controls also notes alignment with NVIDIA’s DSX reference architecture, positioning the guides as compatible with emerging standards for large scale AI factory design.

Supporting repeatable AI infrastructure

Austin Domenici, Vice President and General Manager of Johnson Controls Global Data Center Solutions, says the guides are intended to help operators move from bespoke designs to more repeatable models as AI capacity scales.

“AI Factories are production facilities – the places where intelligence is manufactured at an industrial scale,” Austin explains.

“By supporting the NVIDIA DSX reference architecture and improving water and energy efficiency in the cooling process while maintaining high temperature loop compatibility, our Reference Design Guide equips customers to deploy gigawatt-scale AI infrastructure that is scalable, repeatable, resilient and sustainable.”

Implications for data centre planning

As hyperscalers and sovereign cloud providers plan multi-gigawatt AI campuses, thermal design is becoming a board-level concern rather than an engineering afterthought.

Power availability, water constraints and sustainability regulation are increasingly shaping where and how data centres can be built.

By publishing detailed reference designs, Johnson Controls is aiming to influence early stage planning decisions and reduce uncertainty for operators facing unprecedented scale.

For data centre owners evaluating AI investments, the guides offer a structured view of how cooling architectures may need to evolve as facilities move towards gigawatt-class deployments.