Johnson Controls Unlocks AI Capacity by Turning Waste Heat into Cooling

MILWAUKEE, WI – July 28, 2026 – As the artificial intelligence boom places unprecedented demands on global energy infrastructure, Johnson Controls (NYSE: JCI) today introduced a groundbreaking blueprint designed to turn a data center’s biggest byproduct—waste heat—into its most valuable asset. The company's new Absorption Chiller Reference Design Guide provides a roadmap for data center operators to convert heat from on-site power generation into productive cooling, a move that could significantly reshape the economics and sustainability of the AI industry.

This innovative approach promises to reduce the electrical power needed for cooling by up to 44%, freeing up that energy for critical AI computing tasks. For a gigawatt-scale AI Factory, this efficiency gain could unlock up to 97 megawatts (MW) of additional computing capacity without needing more power from the grid, potentially generating up to $18 billion in extra revenue over the facility's lifespan.

The AI Energy Paradox and the Cooling Conundrum

The rapid expansion of AI has created a paradox: the more powerful the computation, the more heat it generates, and the more power is needed to cool it down. This cycle is pushing data center infrastructure and local power grids to their limits. Industry reports show data center electricity demand is soaring, with some projections indicating it could account for over 10% of U.S. consumption by 2028. A single hyperscale facility can already consume as much electricity as 50,000 homes.

Cooling systems, which traditionally consume 30-40% of a data center's total energy, are struggling to keep pace. AI workloads create intense thermal spikes, with server rack power densities climbing from a standard 10kW to over 100kW, far exceeding the capabilities of traditional air cooling. This has forced a rapid industry shift towards advanced liquid cooling, but the core challenge of immense electricity consumption remains. Furthermore, with grid interconnection queues stretching for years in some regions, operators are increasingly turning to on-site power generation, making the efficiency of every megawatt generated a top priority.

A Blueprint for Efficiency: How Absorption Cooling Works

Johnson Controls' solution tackles this problem at its source by leveraging absorption chiller technology, a process the company has refined over 65 years with its YORK brand. Unlike conventional electric chillers that use a power-hungry mechanical compressor, absorption chillers use a thermal process. Waste heat—in this case, captured from the exhaust of on-site power generators—is used to boil a refrigerant out of an absorbent solution. The refrigerant then cycles through a system to produce chilled water, which is used to cool the data center's IT equipment.

The efficiency gains are dramatic. A single absorption chiller can deliver 2 MW of cooling capacity while drawing only about 25 kilowatts (kW) of electricity for pumps and controls. A conventional electric chiller providing the same cooling would require over 500 kW. By repurposing the 57% of energy typically lost as waste heat, data centers can drastically cut their cooling system's electrical draw.

"One of the biggest untapped opportunities in data centers is the heat they generate," said Austin Domenici, president, Global Data Center Solutions at Johnson Controls. "Johnson Controls helps transform recovered heat into useful work through Combined Heat and Power and absorption cooling solutions, enabling AI Factories to scale more efficiently while reducing strain on the grid and delivering value to local communities."

Unlocking Billions and Boosting Sustainability

The financial and environmental implications of this approach are substantial. The 97MW of additional computing power unlocked in a model 1GW AI Factory isn't just a technical achievement; it's a direct path to increased revenue. By reallocating power from cooling infrastructure to revenue-generating AI processors, operators can maximize the output and profitability of their existing power infrastructure.

Beyond the balance sheet, the reference design offers significant sustainability benefits. The blueprint can help facilities achieve a Power Usage Effectiveness (PUE) rating as low as 1.23. PUE is the industry's primary efficiency metric, and a score of 1.23 approaches the hyper-efficiency of the world's leading operators. Critically, this can be achieved with zero on-site water use by pairing the system with dry coolers, a major advantage in water-stressed regions. The reduced reliance on grid electricity also translates to a potential 43% reduction in CO₂ emissions from the cooling system.

"A vast amount of heat produced by on-site power generators is essentially thrown away — dissipated into the air. We see a huge opportunity when that energy is put to work instead," said Katie McGinty, vice president and Chief Sustainability and External Relations Officer at Johnson Controls. "By converting waste heat into useful cooling, we're turning a resource already bought and paid for into an asset rather than a disposal liability. Every megawatt we can shift from cooling to computing capacity helps customers increase the revenue potential of their facilities and accelerates time to value by significantly cutting pressure on the grid."

Scaling for the Gigawatt Era

To meet the voracious demand for AI, speed and scalability are paramount. Johnson Controls designed its blueprint as a repeatable, modular architecture that can scale from 100MW campuses to gigawatt-scale AI Factories without requiring a complete redesign for each deployment. This modularity helps operators bring new capacity online faster, a crucial competitive advantage in the fast-moving AI landscape.

By making on-site power generation more efficient, this technology directly addresses the grid-related bottlenecks slowing down data center construction. Instead of viewing on-site power as just a backup or a necessity due to grid delays, it becomes part of a hyper-efficient, integrated system. This shift turns a challenge into a strategic advantage, enabling the sustainable growth of AI by creating more computing power from the energy resources already available.