Liquid-Cooled SSDs Could Make AI Data Centers Faster

Liquid Cooling Takes on SSD Heat: AI workloads are pushing enterprise SSD power consumption beyond traditional cooling limits. Liquid-cooled SSDs use direct cooling to remove heat more efficiently, helping data centers maintain stable storage performance as AI systems handle increasingly demanding workloads.

Why Air Cooling Is Struggling: High-performance SSDs generate significant heat during AI training, data processing, and continuous data transfers. Traditional air cooling can struggle in densely packed servers, forcing drives to reduce speeds to control temperatures and potentially slowing down data delivery to GPUs.

How Cold-Plate Cooling Works: Liquid-cooled SSDs use a metal cold plate attached to the drive. A circulating coolant, often a water-glycol mixture, absorbs heat directly from the storage device and carries it away. This method reduces dependence on fans and helps SSDs maintain stable temperatures during demanding workloads.

Cooling Energy Could Drop by 98%: Liquid cooling could deliver major energy savings. Micron's analysis estimates that cooling a 32-drive NVMe SSD bank requires around 37–80 watts with air cooling, compared with just 0.42–1.35 watts using cold plates. This represents approximately 98% lower cooling power in the modeled setup.

More Storage in Less Rack Space: Liquid-cooled SSDs can help data centers fit more storage into compact server racks. By reducing the need for airflow gaps and storage-zone fans, these designs free up space and power for additional computing hardware. The result could be denser AI infrastructure without sacrificing thermal control.

Tech Companies Are Building Liquid-Cooled SSDs: Companies are developing storage designed for liquid-cooled AI servers. Solidigm's D7-PS1010 E1.S supports single-sided cold-plate cooling and hot-swapping, while Kioxia has introduced direct-liquid-cooled enterprise SSD options. Micron is also developing SSD architectures designed for efficient cold-plate contact.

Challenges Before Wider Adoption: Liquid-cooled SSDs could improve AI performance and reduce cooling costs, but adoption requires compatible servers, cold plates, coolant distribution systems, leak detection, and specialised maintenance. These investments may slow adoption in conventional data centers. AI and high-performance computing facilities are likely to benefit first, where dense hardware makes efficient cooling especially valuable.

Join our WhatsApp Channel to get the latest news, exclusives and videos on WhatsApp