Yes. Many AI deployments combine liquid and air cooling. Liquid cooling removes heat from the components connected to the liquid loop; other equipment still needs an appropriate cooling path. The balance depends on the server design and workload.

A rack drawing 120 kW releases roughly 120 kW of heat, equivalent to 80 space heaters rated at 1.5 kW each. NVIDIA describes a 120 kW cooling requirement for its GB200 NVL72 design. Actual rack heat depends on the equipment and operating load.

Some do over a year, while large evaporatively cooled facilities can use much more. Water use depends on facility size, climate, cooling technology and operating hours. Compare annual water consumption and energy use together, rather than assuming every liquid-cooled data center consumes large amounts of water.

Efficient cooling reduces the electricity needed to support a given IT load. It can also free capacity within a facility’s existing power allocation. The total effect depends on the site: adding computing can offset the reduction in cooling demand.

Power Usage Effectiveness, or PUE, is total facility energy divided by IT equipment energy over the same period. At 1.08, every 100 kWh used by IT requires 8 additional kWh for cooling, power conversion and other overhead. Compare PUE values only when measurement boundaries, IT load and reporting periods are consistent.

Compare the installed capacity with measured load, peak design conditions, growth plans and redundancy requirements. Spare capacity may be necessary for resilience. Annual weather and load analysis can identify opportunities to improve part-load operation without compromising peak cooling or failure protection.

Leak protection should combine detection, alarms, monitoring and a defined isolation response. The design must also explain how cooling is maintained during a fault. Factory testing and site commissioning should verify the specified behavior, including the effect of isolation on neighboring equipment.

A coolant distribution unit, or CDU, manages the delivery of coolant to liquid-cooled IT equipment. In a liquid-to-liquid design, a heat exchanger separates the server coolant loop from the facility loop. Pumps, sensors and controls manage conditions such as flow, pressure and temperature.

It depends on the architecture. A liquid-to-liquid CDU transfers heat to a facility liquid loop. A liquid-to-air CDU transfers heat to room air, so the room cooling system must remove that load. A recirculating liquid loop does not by itself mean continuous consumption of fresh water.

Often, yes, after a site assessment. Check the rack requirements, electrical service, floor loading, pipe routes, available space and heat rejection. Existing air cooling may still be needed. The installation plan must account for the live IT load and any required tie-ins or temporary cooling.

Indirect/direct evaporative cooling, or IDEC, uses two stages. The indirect stage cools supply air through a heat exchanger without adding moisture to that air stream. The direct stage provides further cooling by evaporating water into the supply air, which increases its moisture content.

Savings depend on climate, cooling load, supply conditions and the system being replaced. Air₂O’s DEN software estimates annual power and water consumption for evaporative cooling options. Compare alternatives using the same load and reliability requirements, with fans, pumps and any mechanical trim included.

It can, but humidity limits the cooling available from evaporation. Air₂O hybrid designs combine evaporative and mechanical cooling so the appropriate stages can operate as conditions change. The selected configuration and local weather determine performance; dry-climate savings should not be assumed for humid sites.

No. Recirculating and indirect designs can separate data hall air from the outdoor air used for heat rejection. Where outside air enters the data hall, the design must address filtration, humidity, contaminants and smoke or dust events. Air₂O offers both fresh-air and recirculating approaches.

In some facilities, upgrades can be phased while existing cooling supports the live load. Feasibility depends on spare capacity, isolation points, temporary cooling and the commissioning plan. A site assessment should establish whether planned interruptions are needed before a no-downtime commitment is made.

Dry cooling configurations can eliminate routine evaporative cooling water consumption. That does not eliminate every source of water use at a data center. Air₂O’s SMR offering includes a waterless configuration with optional evaporative assist; using that assist changes the water profile. Evaluate the exact configuration and annual operating modes.

Ask for the refrigerant designation, charge, global warming potential and applicable compliance information for the exact unit. Requirements vary by equipment, application, location and date. Evaporative cooling without a refrigerant circuit differs from hybrid or DX equipment that includes mechanical refrigeration.

Potentially, subject to approval for the selected equipment and application. Water chemistry, solids, biological control, corrosion and scaling must be assessed. Evaporative makeup water and server-loop coolant have different requirements. Confirm the permitted water quality and treatment plan before selecting a source.

Yes, where a nearby process or building can use heat at the available temperature. Feasibility depends on temperature, distance, seasonal demand and economics. Heat exchangers or heat pumps may be required. Air₂O’s custom DX range includes heat recovery options; the scope must be defined for each application.

Start with the project location, cooling or IT load, target air or coolant conditions and required commissioning date. Include available utilities, dimensional and weight limits, redundancy requirements, controls interfaces and whether the project is a new build or retrofit. For liquid cooling, include the server or rack thermal specification.

DEN is Air₂O’s performance-prediction software for evaporative cooling. It uses weather data and performance information for direct, IDEC and hybrid technologies to estimate annual power and water consumption. The analysis helps compare cooling options against a defined baseline for your project.

Yes. Air₂O engineers custom thermal management equipment around project requirements, including footprint, access and weight constraints. A custom solution can fit an application that standard equipment cannot. Any size or weight advantage should be compared at equivalent cooling duty and reliability requirements.

Share your required commissioning date early. The schedule should account for design approval, component availability, manufacturing, factory testing, shipping, installation and startup. Request a written project schedule for the selected configuration; factory location alone does not establish a delivery date.

Supported interfaces depend on the product and controller. Our custom DX documentation lists BACnet IP and MS/TP, plus Modbus IP and serial communications. EcoFlow documentation lists BACnet and Modbus serial communications with optional remote access. Confirm the network interface, points list and alarm requirements for your selected unit.

Certifications and listings are product-specific. For example, Air₂O’s LLB-670 specification sheet identifies U.S. and Canadian UL 508 certification and ISO 9001 manufacturing. Request the current certificate or listing, model scope and applicable standard for your unit. ISO 9001 addresses the quality management system, rather than cooling performance.

Phoenix manufacturing does not by itself establish protection from tariffs or material cost changes. Review the project quotation for price validity, adjustment terms and any agreed price protection. Domestic manufacturing location and component country of origin are separate considerations.

Yes. Air₂O serves international markets, including North America, EMEA and Asia Pacific. Product availability, local approvals, shipping terms, commissioning and ongoing service should be confirmed for the destination country and project.

Commissioning uses controlled tests to check cooling performance before the live IT load is introduced. Liquid load banks apply simulated heat to liquid cooling infrastructure. Air₂O’s LLB-670 is one option. A project test plan should define load conditions, controls, alarms, required failure scenarios and acceptance criteria.

Timing depends on the equipment, site readiness, pipework or ductwork, electrical and controls integration, flushing requirements and acceptance tests. A single-unit installation and a multi-rack deployment have different schedules. Establish responsibilities, prerequisites and test milestones before agreeing a completion date.

Maintenance depends on the configuration and site conditions. The equipment manual and service plan should define filter inspection and replacement, coil or heat-exchanger cleaning, water or coolant checks, media servicing and controls verification. Confirm access, service responsibilities and critical spares during design.

We design and manufacture thermal management equipment for data centers and other demanding facilities. Our portfolio includes air handling, evaporative and hybrid cooling, dehumidification and liquid load banks. We also serve semiconductor and battery manufacturing, industrial facilities and indoor agriculture.

Air₂O has engineering and manufacturing operations in Phoenix, Arizona. Our manufacturing capabilities include welding, assembly, electrical controls and testing. Equipment is selected for the operating conditions and requirements of each project.

Our manufacturing operation brings engineering, fabrication, assembly, electrical controls and testing capabilities together. These teams turn project requirements into thermal management equipment, with product-specific manufacturing and test requirements.