Key Takeaways:
Johnson Controls' new blueprint turns wasted heat from on-site power generation into cooling, freeing up nearly 100MW of computing capacity at gigawatt-scale AI factories.
Johnson Controls released a reference design that converts waste heat from on-site power generation into cooling, cutting electrical demand by as much as 44% and freeing up 97MW of additional computing capacity at a 1GW-scale AI factory.
"One of the biggest untapped opportunities in data centers is the heat they generate," Austin Domenici, president of global data center solutions at Johnson Controls, said.
The absorption chiller design uses heat from natural gas turbines or fuel cells to drive cooling, requiring just 25kW of electricity for 2MW of cooling capacity — compared with roughly 500kW for a conventional electric chiller, according to the company. The modular blueprint scales from 100MW campuses to gigawatt-level facilities without redesign and can achieve a power usage effectiveness as low as 1.23 with zero on-site water consumption.
For operators racing to bring AI capacity online, the efficiency gains translate into as much as $18 billion in additional revenue over a facility's life by shifting power from cooling to computing. Johnson Controls, which has deployed absorption chillers in naval vessels and manufacturing plants for more than 65 years through its YORK brand, is offering the technology as a repeatable solution for the hyperscale buildout.
How Absorption Cooling Works in Data Centers
Absorption chillers use heat — typically steam or hot water — to evaporate a water-lithium bromide mixture in a vacuum. The vapor passes through condensers and cooling towers before returning to the cycle. The technology has long been used in industrial settings but has seen limited adoption in data centers because the heat expelled by IT equipment, even high-power GPUs, is not hot enough to drive the process.
The calculus changes when data centers generate their own power on site. Natural gas turbines and fuel cells produce high-grade exhaust heat that absorption chillers can capture and convert into chilled water for cooling server rooms. The approach effectively turns a waste stream into a productive asset, reducing strain on the grid and lowering the facility's carbon footprint.
Competitive Implications for the AI Buildout
The reference design arrives as hyperscalers and AI factory operators face mounting pressure to secure power and cooling for rapidly expanding workloads. Data center power demand is projected to more than double by 2030, driven largely by AI training and inference, according to industry estimates. Every megawatt shifted from cooling to computing improves the economics of new capacity.
Johnson Controls competes in the thermal management market against Vertiv, Schneider Electric, and Carrier Global. The absorption chiller approach offers a differentiated solution for facilities with on-site generation, a segment that is growing as operators turn to behind-the-meter natural gas and, potentially in the future, small modular nuclear reactors. The company said the design can reduce cooling system CO₂ emissions by up to 43% compared with conventional approaches.
For Johnson Controls, the reference design opens a new revenue channel tied directly to the AI infrastructure boom. JCI shares trade at roughly 22x forward earnings, in line with industrial peers. If the design gains adoption among hyperscale operators, the company could capture a meaningful share of the cooling equipment spend tied to the estimated $1 trillion in global data center capital expenditure expected over the next five years.
This article is for informational purposes only and does not constitute investment advice.
