MagPro has introduced AeroLev 1850, a 1.85 MW-class air-cooled magnetic-bearing chiller designed for AI data centers operating in hot, water-constrained environments. The platform combines oil-free magnetic-levitation compression, air cooling and intelligent controls, while supporting chilled-water temperatures of up to 45°C for emerging high-temperature liquid-cooling architectures.
AI data centers are creating a cooling problem that is increasingly difficult to separate from the broader questions of power availability, water use and site selection. As accelerator densities rise, operators need to remove more heat without necessarily adding proportional demands on local water resources.
MagPro, the joint venture between CIGU Technology Corp. Ltd. and EPG Data Technology (Shanghai) Co., Ltd., is targeting that challenge with AeroLev 1850, a 1.85 MW-class air-cooled magnetic-bearing chiller platform.
The company announced the system on September 28, positioning it for next-generation AI data centers in high-ambient and water-scarce environments. MagPro says AeroLev 1850 eliminates the need for cooling towers and evaporative cooling water while providing chilled water at temperatures of up to 45°C.
That combination is notable because AI infrastructure is increasingly pushing cooling systems beyond the assumptions of conventional data center designs.
AI increases the pressure on cooling infrastructure
The underlying issue is straightforward: electricity consumed by computing equipment ultimately becomes heat that has to be removed.
The International Energy Agency estimates that data centers consumed about 415 TWh of electricity in 2024, or roughly 1.5% of global electricity consumption. Its base case projects data center electricity consumption to reach approximately 945 TWh by 2030, with accelerated servers—largely driven by AI—accounting for a significant portion of the increase.
Gartner estimates that global data center electricity consumption will reach 565 TWh in 2026, up 26% from 2025. AI-optimized servers are expected to account for 31% of data center power consumption this year, with their consumption projected to exceed that of conventional servers in 2027.
That growth is pushing operators toward higher-density thermal architectures.
Gartner’s 2026 infrastructure research also identifies liquid cooling as an increasingly important consideration for facilities supporting AI and other high-density workloads, while highlighting water usage as a growing data center planning factor.
Air cooling meets liquid-cooling infrastructure
AeroLev 1850 takes a somewhat different approach from a conventional water-cooled chiller.
The platform uses an air-cooled, magnetic-levitation compressor, avoiding the cooling towers and evaporative water losses associated with many conventional cooling configurations. MagPro says its compressor can achieve up to 87% isentropic efficiency, while a proprietary condenser provides more than 3°C of subcooling.
The company claims these technologies can deliver up to 40% higher overall energy efficiency and up to 8% lower failure rates compared with conventional oil-lubricated compressors. It also claims maintenance costs can be up to 10% lower than conventional chillers.
Those performance figures are MagPro-reported claims, rather than independently verified benchmarks.
The magnetic-bearing architecture is central to the proposition. Removing oil lubrication and mechanical contact can reduce wear within the compressor, potentially lowering maintenance requirements while providing a different reliability profile from traditional oil-lubricated systems.
MagPro also incorporates EC fans and intelligent controls into the platform, while its product information describes AI-driven optimization and digital-twin capabilities for adjusting cooling performance as operating conditions change.
Water efficiency becomes a data center design issue
The water argument is becoming more significant as AI capacity expands into regions where water availability is already constrained.
The IEA estimates that data centers’ global water consumption was around 560 billion liters annually in 2023 and could reach approximately 1.2 trillion liters by 2030 in its base case. Water requirements vary substantially depending on cooling technology, climate and the electricity sources supplying a facility.
This makes water use an increasingly important variable when selecting sites for AI infrastructure.
An air-cooled chiller does not eliminate all water requirements associated with a data center. Water can still be involved indirectly through electricity generation, semiconductor manufacturing and other parts of the supply chain. But eliminating evaporative cooling at the facility can reduce one direct source of water consumption.
That is the market MagPro is targeting with AeroLev 1850.
Designed around higher-temperature liquid cooling
The other significant element is the platform’s ability to provide chilled water at temperatures of up to 45°C, according to MagPro.
The company says this makes the system compatible with emerging direct-to-chip and immersion liquid-cooling architectures.
That is important because liquid cooling does not necessarily mean every component needs extremely cold water. Newer high-density cooling architectures can operate with higher coolant temperatures, potentially increasing opportunities for free cooling and reducing the amount of mechanical refrigeration required under favorable ambient conditions.
AeroLev 1850 incorporates a free-cooling mode that uses ambient air during cooler periods. The objective is to reduce compressor runtime and the associated electricity consumption when outdoor conditions make mechanical cooling less necessary.
This approach also reflects a broader industry shift toward designing cooling systems around the complete operating envelope rather than peak conditions alone.
Modularity matters as AI capacity changes
MagPro is also emphasizing standardized modular deployment.
AI data center capacity is being added in large increments, but operators still need infrastructure that can adapt as demand changes. A modular chiller platform can potentially allow cooling capacity to be added alongside new compute capacity instead of requiring an entirely new cooling architecture.
EPG’s broader data center infrastructure business includes modular systems, cooling solutions and liquid-cooling infrastructure. The company has also been promoting AeroLev 1850 as a 1.85 MW air-cooled magnetic-levitation chiller for high-density AI deployments.
The combination of modular deployment and higher-temperature chilled water could therefore make the platform relevant to both new AI facilities and infrastructure upgrades.
Cooling becomes part of the AI infrastructure stack
The emergence of systems such as AeroLev 1850 reflects a larger change in AI infrastructure. Cooling is no longer simply a facility-support function operating behind the IT equipment. At high rack densities, thermal design affects compute deployment, power capacity, operating costs and even where a data center can be built.
The IEA’s 2026 analysis says AI-focused data center electricity consumption increased 50% in 2025, while overall data center electricity demand rose 17%. Its updated outlook projects global data center electricity consumption will roughly double from 485 TWh in 2025 to 950 TWh by 2030.
That trajectory creates room for different cooling approaches rather than a single universal architecture.
Liquid cooling will remain important for high-density compute, but the infrastructure supporting those liquid loops can itself take different forms. MagPro’s strategy is to use air-cooled, oil-free magnetic-bearing chillers to support liquid-cooling systems while reducing dependence on evaporative water.
For AI data centers being developed in hot or water-stressed locations, that combination puts cooling efficiency, water consumption and deployment flexibility into the same infrastructure equation.
Market Landscape
AI is increasing both the power density and thermal requirements of data centers, while water availability is becoming another constraint on site development. Gartner says AI-optimized servers will account for 31% of global data center power consumption in 2026, while data center electricity use is projected to rise 26% year over year.
Meanwhile, the IEA estimates global data center water consumption could rise from around 560 billion liters in 2023 to approximately 1.2 trillion liters in 2030 in its base case.
The result is a growing market for liquid cooling, air-cooled chillers, coolant distribution systems, free cooling and higher-temperature thermal architectures. A June 2026 IEA 4E analysis found liquid cooling can offer server-level energy savings of around 8% and facility-level savings of 30–40% in applicable configurations, although adoption remains constrained by standardization, upfront costs and reliability considerations.
Top Insights
- MagPro’s AeroLev 1850 is a 1.85 MW-class air-cooled magnetic-bearing chiller designed for high-density AI data centers.
- The platform eliminates cooling towers and evaporative cooling water while supporting chilled-water temperatures up to 45°C.
- MagPro is targeting data centers in hot, water-constrained regions where cooling and resource availability can influence site economics.
- The system is designed to support direct-to-chip and immersion liquid-cooling architectures alongside free-cooling operation.
- Gartner and IEA forecasts show rapidly increasing AI data center power demand, strengthening the case for more efficient thermal infrastructure.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
