Nvidia has introduced DSX POD, a modular, containerised version of its DSX AI Factory Platform reference design for Vera Rubin powered data centres. The units are integrated and validated before delivery and connect to a prepared site’s power, cooling and networking, which Nvidia says reduces deployment time from six months to 14 days. Partners listed for worldwide delivery in early 2027 include Dell Technologies, HPE, Lenovo, Foxconn, Crusoe, Schneider Electric and Vertiv, among others.
The platform’s efficiency claims rest on cooling design. DSX connects liquid cooled compute to dry coolers that reject heat to the air without evaporation. Nvidia claims the approach can cut cooling water use by up to 100 percent, and that at a 2 gigawatt scale it could save 54 million litres of water a day compared with conventional cooling tower systems, while noting some climates may still need mechanical cooling in extreme heat.
Two companion products address power. DSX MaxLPS dynamically allocates power within a facility’s provisioned budget, which Nvidia says can produce the same token output with up to 30 percent less power. DSX Flex is designed to let data centres adjust workload power consumption to grid conditions, prices and utility requests. “When there is an energy spike and the grid needs an extra gigawatt, our data centers can be a resource to respond to that need,” Nvidia’s Ian Buck told media ahead of the GTC Berlin announcement.
All of the performance figures are Nvidia’s own claims from its announcement, reported by a single outlet in this run, and they have not been independently verified.