Supermicro Computer (SMCI) said on Wednesday it will work with Nvidia to expand manufacturing capacity and cooling technology for liquid-cooled AI infrastructure solutions for the next-generation Vera Rubin and Rubin platforms.

The collaboration will make Supermicro the fastest supplier of Nvidia's Vera Rubin NVL72 and HGX Rubin NVL8 solutions. Supermicro adopted a building-block-based modular design method for its data centre building block solutions (DCBBS). It said this streamlines production and supports a wide range of configuration options and rapid deployment.

DCBBS includes both air-cooled and liquid-cooled systems. It offers multiple CPU options and includes data centre management software. It enables rack-level integration, including networking and cabling and cluster-level L12 validation. It also provides global delivery, support and services.

Supermicro will accelerate the buildout of next-generation liquid-cooled AI infrastructure based on its design and manufacturing capabilities in the United States. The company also expanded rack-scale manufacturing capacity for liquid-cooled AI solutions. It plans to support hyperscalers and enterprise customers in deploying AI infrastructure at scale based on the Vera Rubin and Rubin platforms.

Charles Liang, president and CEO of Supermicro, said, "Through our long partnership with Nvidia and Supermicro's agile building block solutions, we can bring industry-leading AI platforms to market faster than anyone." He added, "With expanded manufacturing capacity and industry-leading liquid-cooling technology, we will support hyperscalers and enterprise customers in building large-scale AI infrastructure quickly and efficiently based on the Vera Rubin and Rubin platforms."