The new offering pairs each NVL72 rack – which houses 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs – with a complete in‑house cooling loop that runs from cold plates through manifolds to facility‑side cooling towers, according to the company’s press release.
Each 1U compute tray in the rack holds four Rubin GPUs and two Vera CPUs; nine sixth‑generation NVLink switch trays provide 216 TB/s of bandwidth, while the rack delivers 20.7 TB of HBM4 memory and up to 54 TB of LPDDR5X storage.
Supermicro’s DCBBS portfolio supplies every component of the liquid‑cooling chain, including cold plates, hose kits, power shelves, in‑row and in‑rack distribution units (rated at 1.8 MW per CDU with N+1 redundancy), side‑car CDUs, rear‑door heat exchangers and cooling towers, all manufactured in the United States, Taiwan and the Netherlands.
The company’s DCBBS Blueprint defines a balanced bill‑of‑materials for deployments ranging from 5 MW to gigawatt scale. A single “Scalable Unit” based on one NVL72 rack aggregates 1,152 Rubin GPUs and 331 TB of HBM4 memory across 16 compute racks, with coordinated power, cooling, storage and networking resources.
"We have spent years building the liquid‑cooling stack, the manufacturing capacity, and the deployment teams for exactly this moment," said Charles Liang, president and CEO of Supermicro. "Our customers can now order a Scalable Unit and receive production‑ready systems with end‑to‑end integration, because we design and build every layer between the cold plate and the cooling tower."
The Vera Rubin platform was co‑designed with direct liquid cooling to achieve AI‑throughput‑per‑watt levels that air‑cooled systems cannot match. Supermicro says its testing and validation of each rack with the full cooling loop shortens the time required to bring AI clusters online after site installation.
Supermicro, listed on NASDAQ as SMCI, describes itself as a total‑IT‑solutions provider for AI, enterprise, storage and 5G/edge markets. The company emphasizes that its in‑house design and manufacturing approach aims to improve total cost of ownership and reduce environmental impact through what it calls “green computing.”