Every Watt and Square Inch Matters: Solidigm’s Innovative SSD Cooling Solution for AI Datacenters
Tackling one of AI era’s most pressing challenges: thermal efficiency
California-based solid-state drive (SSD) company Solidigm’s recent collaboration with NVIDIA, unveiled at the GPU Technology Conference (GTC), drew significant attention for introducing a first-of-its-kind liquid cooling solution for SSDs. In an exclusive interview with TechSoda, Dave Sierra, Efficiency Evangelist at Solidigm, recently explored how the company is tackling one of the AI era’s most pressing challenges: energy and thermal efficiency in data center storage.
“AI is going to be more than 50% of worldwide data center energy,” said Sierra, “It's a huge, huge driver of energy usage and the problem that needs to be solved as data centers are built out going forward."
Solidigm, a leading enterprise SSD provider which was an Intel division and now a 100% subsidiary of SK Hynix, has positioned itself at the forefront of sustainable, high-performance storage solutions for AI workloads.
Sierra explained that while GPUs have seen advances in power efficiency, overall energy demands in AI computing continue to climb sharply.
Traditionally, SSDs were not designed for liquid cooling, unlike GPUs—creating a cooling mismatch in AI server systems and limiting full-system thermal efficiency.
To bridge that gap, Solidigm engineered a new SSD design featuring a specially built enclosure that enables cold plates to cool both sides of the drive. This innovation now allows for 100% liquid-cooled AI servers, marking a major industry milestone—and the first time SSDs have been fully integrated into liquid-cooled infrastructure.