Stuart Sheehan

As data centers deploy liquid cooling systems to support high-density AI workloads, operators need to decide on coolant distribution unit (CDU) placement. The main question is whether to deploy in-rack CDUs or floor-mounted CDUs. Making a choice comes down to the specific requirements of the liquid cooling architecture in each implementation. The CDU acts as the hub of the cooling infrastructure, circulating liquid from high-density hardware in a closed loop to dissipate heat.

The AI shift: Why cooling is being redefined The growth of graphics processing unit (GPU)-based accelerated computing that powers AI workloads is changing the data center architecture. The power consumption of these chips is significantly higher than traditional central processing units or CPUs. For many years servers were designed for CPUs that consumed ~150 watts. Today’s Nvidia Blackwell chips are more in the 1000-1400 watt range.

Data center rack densities are rapidly increasing to support high-density AI workloads and high-performance computing (HPC). As a result, the data center power architecture is changing, with 800VDC (direct current) emerging as the new standard for rack power to support these workloads. Alongside this shift, the cooling implications of 800VDC architecture are becoming a critical consideration.