AI data centers are evolving from general-purpose computing facilities into hybrid environments that support both AI and conventional workloads. Chips with different architectures, precision levels and generations increasingly coexist in the same facility. Meanwhile, rack power density is rising from several kilowatts to tens of kilowatts—and even approaching the 100 kW class.

This shift is challenging the traditional model of building the facility first and configuring the equipment later. Power systems must withstand fluctuations in AI loads, cooling must be controlled precisely rather than applied uniformly, and the overall architecture must support flexible deployment.

Kehua Tech WiseAisle 8.0 Modular Data Center Solution aligns power and cooling resources with computing demand, enabling compute, power and cooling to operate as a coordinated system.

Integrated Infrastructure for High-Density Deployment

To support high-density, efficient and flexible AI data centers, WiseAisle 8.0 centralizes power capacity while bringing cooling closer to the equipment it serves.

Power and Compute: More Space for IT

For high-density computing zones, WiseAisle 8.0 can be configured with the MR33 Series Modular UPS. Each 200 kW high-density power module is only 4U high, and each cabinet occupies approximately 0.8 m². Compared with typical industry configurations, the design reduces the footprint by approximately 39%, freeing more space for IT equipment.

For larger-scale deployments, including power enclosures or full-floor power distribution systems, Kehua’s prefabricated power module integrates the MR33 Series Modular UPS. At this scale, the integrated design reduces the overall footprint by approximately 24% compared with a conventional solution.

Cooling and Compute: Matching Cooling to the Workload

AI data centers do not always benefit from a single cooling method. Applying liquid cooling throughout a facility may not suit every workload, while relying only on air cooling may not meet higher-density requirements.

WiseAisle 8.0 supports different cooling configurations for AI inference, medium- and high-density training, and mixed AI and general-purpose computing. The solution categorizes racks by power density and matches each category with an appropriate cooling approach. 

The available options include high-density air cooling, hybrid air and liquid cooling, and retrofit configurations that convert existing air-cooled deployments to liquid cooling, covering rack power densities from 10 kW to 70 kW.

Coordinated Infrastructure for Mixed Workloads

The main challenge of mixed computing is not simply higher power consumption, but greater diversity. Training, inference and storage workloads have different power-density and cooling requirements. A fixed configuration can easily result in resource underutilization or overloading in parts of the system. WiseAisle 8.0 addresses this through modular power zoning and cabinet-level temperature control.

Power and Compute: Modular Zoning

WiseAisle 8.0 uses modular zoning to organize IT space and power distribution around different workload requirements. Areas serving AI workloads and general-purpose computing can be configured at cabinet or room level.

It also applies a modular design to precision power distribution cabinets. Each cabinet supports up to six power distribution modules. Each module provides up to 150 kW in a height of 6U and supports both single-phase and three-phase outputs.

Cooling and Compute: Cabinet-Level Precision Control

Designed around low-carbon operation, WiseAisle 8.0 integrates AI and smart energy expertise throughout the system, enabling precise temperature control for individual cabinets. This creates a closed loop of sensing, control and action, making mixed-workload deployment easier.

When AI workloads fluctuate, the system can provide feedforward response within seconds, anticipate hotspots and issue the corresponding cooling strategy. Once conditions stabilize, the system dynamically adjusts cooling to precisely control the temperature of each cabinet, rather than cooling the entire data hall uniformly. By coordinating power and cooling under different operating conditions throughout the year, this mechanism can deliver approximately 20% energy savings.

For AI data centers, the goal is not simply to add more equipment, but to coordinate compute, power and cooling as one system. Drawing on the solution experience across countries and regions, Kehua Tech applies this coordinated approach from system design to digital infrastructure deployment, enabling every watt of power and every unit of cooling capacity to serve computing hardware more precisely.