Nvidia Modular AI Design Targets Deployment in as Little as 14 Days

At GTC Berlin on Oct. 21, Nvidia introduced a modular design for Vera Rubin AI factories that could reduce deployment at prepared sites from six months to as little as 14 days.

The modular DSX POD design moves system integration and validation to partner factories before delivery. At a prepared destination, the modules are designed to connect to power, cooling and networking already in place. The design gives partners a blueprint for building Vera Rubin systems around Nvidia’s platform.

The 14-day figure is a potential deployment timeline for a prepared site, not a result from a completed installation. It compares with a six-month timeline, and requires the site’s supporting connections to be ready.

Nvidia named 17 partners, including Dell Technologies, HPE, Lenovo, Schneider Electric and Vertiv. Partners are set to deliver DSX POD factories worldwide in early 2027.