Introducing DSX POD at GTC Berlin
NVIDIA introduced DSX POD on October 21, 2026, at GTC Berlin, marking a new milestone in AI infrastructure. The product is a partner-built, containerized reference design intended for complete Vera Rubin AI factories, assembled inside modular containers and validated prior to delivery.
According to NVIDIA's press release, the units arrive ready to integrate with prepared site utilities, with the company claiming it reduces deployment time from six months to as little as 14 days. Seventeen partners, including Dell, HPE, Lenovo, Foxconn, and Vertiv, plan to begin deliveries in early 2027.
What the 14-Day Deployment Actually Covers
The headline-grabbing 14-day window comes with a crucial qualifier: it applies exclusively to sites that have already been prepared. NVIDIA's documentation clarifies that the design does not eliminate onsite preparatory work, as customers still need to establish connections for power, facility water, cooling, and networking before the containers arrive. Months spent securing grid capacity and permits remain outside this advertising window.
The complex integration work, however, moves into the manufacturing facility. Compute, networking, cabling, rack-level power, and liquid cooling are installed offsite, where partners complete functional validation before shipment. Jensen Huang, NVIDIA's founder and CEO, stated that an AI factory creates value when it begins producing tokens rather than upon the completion of standard construction.
Seventeen Partners and Strategic Shifts
NVIDIA is not manufacturing the pods directly. Instead, the company designs and validates the cluster-system architecture, while a diverse roster of seventeen partners—including server makers, power specialists, and edge-computing firms like Anduril and Armada—handles engineering, manufacturing, and support. This collaborative approach aims to tackle bottlenecks and shift focus toward faster operational capacity.
Source: AI Daily News. This story was rewritten and translated for our readers.