Wuhu Tier III Data Center

As AI training, inference and high-performance computing workloads continue to grow, data centers are under increasing pressure to support higher rack densities, maintain continuous uptime, and improve energy efficiency.

In Wuhu, one of China's national computing hubs, a Tier III data center faced exactly this challenge:

How to support growing AI computing demand without compromising uptime or energy performance?

3 Challenges

More Computing, More Heat, More Risk

Designed according to Tier III standards, the facility supports AI training, inference and high-performance computing workloads around the clock. Its cooling infrastructure faced three key challenges:

• Rising rack densities that push beyond the limits of traditional air cooling

• Zero tolerance for cooling interruptions that could impact business continuity

• Stable liquid cooling operation for large-scale GPU deployments

For operators, cooling is no longer just about removing heat—it is about protecting performance, uptime and efficiency.

MBT Solution

Reliable Liquid Cooling for AI Infrastructure

To support the facility's liquid cooling system, MBT supplied nearly 400 Cooling Distribution Units (CDUs).

Acting as the heart of the liquid cooling system, the CDU dynamically adjusts coolant flow according to server loads while ensuring efficient heat transfer and stable operation.

The results were clear:

Annual PUE ≤ 1.1

Equipped with high-precision control valves and variable-speed pumps, the CDU optimizes coolant flow according to server demand, helping the facility achieve an annual average PUE as low as 1.1.

Built for Continuous Operation

A redundant architecture with millisecond-level fault switching automatically activates standby pumps in the event of a failure, while power-loss protection and auto-restart functions help ensure uninterrupted cooling.

Simplified Operations

Supporting seamless integration with facility monitoring systems via the Modbus protocol, the CDU enables remote monitoring and intelligent flow management to maintain stable inlet temperatures with minimal manual intervention.

Beyond One Project: Scaling for the AI Era

The Wuhu project demonstrates how liquid cooling can help data centers meet the growing demands of AI while balancing efficiency and reliability.

Building on this expertise, MBT has developed a CDU portfolio ranging from 305 kW to 2600 kW to support the next generation of high-density AI infrastructure.

As AI computing continues to scale, cooling is becoming a critical foundation for sustainable data center growth—and MBT is helping drive that transition forward.