DSX POD integrates computing, networking and cooling before delivery, with deployment targeted in as little as 14 days.
NVIDIA has introduced DSX POD, a reference design for NVIDIA Vera Rubin AI factories that partners build and validate inside modular containers. The units are designed to connect to a prepared site’s power, cooling, networking and other supporting infrastructure, with deployment time estimated to decrease from six months to as little as 14 days.
DSX POD moves the integration and installation of computing, networking, cabling, rack-level power and liquid cooling to a manufacturing facility. This reduces the amount of onsite planning, construction, specialized labor and logistics required compared with a conventional build. Factory testing is intended to limit the effects of component delays and reduce the risk of errors during onsite installation, allowing customers to begin model training and production inference sooner.
Factory-built infrastructure
NVIDIA DSX POD combines Vera Rubin systems with the infrastructure required for operation in standardized, containerized designs that partners can manufacture and deliver.
The reference design follows the broader NVIDIA DSX infrastructure platform while using an architecture that integrates and validates systems in purpose-built containers before transport and deployment. Partners complete rack-to-rack cabling, wiring, piping and functional validation before shipment. This work can proceed separately from site preparation, reducing the time required to bring additional capacity online.
Earlier production availability connects deployment time with project economics. “Time to first token” refers to the point at which a new AI factory begins running models for production workloads. Earlier operation can allow infrastructure operators to provide services to customers sooner.
Global partner ecosystem and availability
NVIDIA is working with Anduril, Armada, Bull, Crusoe, DayOne, Dell Technologies, Delta, Eaton, Foxconn (Ingrasys), GIGABYTE, HPE, Lenovo, Pegatron, Quanta Cloud Technology (QCT), Schneider Electric, Vertiv and Wistron to deliver DSX POD AI factories worldwide beginning in early 2027.
For more information, visit nvidia.com.