Why air cooled chillers make no sense on AI data centers….
If the entire cooling load is put on air cooled chillers, when 70% of the load only needs 32 deg C water is folly. Why cool all the chilled water to 20 C with mechanical cooling when it’s not needed?
With a hybrid system, the entire DLC load can be cooled using dry coolers with Adiabatic trim. This called “economizer. On a 50 MW POD, this reduces the chiller capacity from 50 MW to 15 MW.
And the increased efficiency of the hybrid system actually uses less water when the water used at the power plant is included.
Don’t believe me, run the numbers. We have, there is no such thing as a water less data center cooling system.
Why air cooled chillers make no sense on AI data centers….
R Stephen Spinazzola, PE
Verified Source
2026-03-03
1 min read
Source
**Key Insight:** Air-cooled chillers make no sense on AI data centers.
GasGx Editorial Insight
**Key Insight:** Air-cooled chillers make no sense on AI data centers.
[Body Paragraph 1: Analysis of the market/tech situation]
The article discusses the inefficiency of air-cooled chillers in cooling all the chilled water to a specific temperature, even when only 70% of the load needs it. This is due to the high cost of mechanical cooling and the need for large amounts of chilled water. The author suggests using a hybrid system that uses dry coolers with Adiabatic trim to reduce the chiller capacity and increase efficiency.
[Body Paragraph 2: The specific operational implication]
Using a hybrid system can significantly reduce the energy consumption and operating costs of an AI data center. It also reduces the amount of water needed for cooling, which is crucial in areas where water resources are limited.
[GasGx Take:] Our GasGx LCOE Calculator can help you accurately forecast the energy consumption and operating costs of your AI data center. By using a hybrid system, you can reduce the chiller capacity and increase efficiency, resulting in lower energy consumption and operating costs.
[Recommended SEO Tags:] "AI Data Center", "Hybrid System", "Energy Consumption", "Operating Costs", "GasGx LCOE Calculator"
[Body Paragraph 1: Analysis of the market/tech situation]
The article discusses the inefficiency of air-cooled chillers in cooling all the chilled water to a specific temperature, even when only 70% of the load needs it. This is due to the high cost of mechanical cooling and the need for large amounts of chilled water. The author suggests using a hybrid system that uses dry coolers with Adiabatic trim to reduce the chiller capacity and increase efficiency.
[Body Paragraph 2: The specific operational implication]
Using a hybrid system can significantly reduce the energy consumption and operating costs of an AI data center. It also reduces the amount of water needed for cooling, which is crucial in areas where water resources are limited.
[GasGx Take:] Our GasGx LCOE Calculator can help you accurately forecast the energy consumption and operating costs of your AI data center. By using a hybrid system, you can reduce the chiller capacity and increase efficiency, resulting in lower energy consumption and operating costs.
[Recommended SEO Tags:] "AI Data Center", "Hybrid System", "Energy Consumption", "Operating Costs", "GasGx LCOE Calculator"
Operational CTA
Recommended GasGx Navigation: air cooled
Based on the scraped content focus, this GasGx page best matches the current topic (air cooled). Open it to continue with related tools, rankings, products, or resources.
Original Source