Nvidia DSX POD reduces AI data center setup to 14 days

Modular reference design unveiled at GTC Berlin

At GTC Berlin Nvidia introduced the DSX POD, a reference design for modular AI data centers built around its Vera Rubin systems. The design is assembled by partners such as Dell, HPE, Lenovo and Schneider Electric, which install compute, networking, cabling, power and liquid cooling in containers at the factory and test the complete units before shipment.

On site the containers only need to be connected to power, cooling and network infrastructure at a prepared location, eliminating most of the on site integration work.

Reduced deployment time and economic benefit

Nvidia says the modular approach cuts deployment time from six months to as little as 14 days, allowing operators to start running AI models and generating revenue much sooner.

The company argues that the faster a data center becomes operational, the earlier its owner can earn money from AI workloads.

Closed loop cooling and rollout timeline

The broader DSX platform uses closed loop cooling that releases heat into the air without evaporation, which Nvidia claims can reduce cooling water consumption by up to 100 percent.

Nvidia acknowledges that some climates will still require additional cooling during extreme heat, and the first DSX POD units are scheduled to ship worldwide in early 2027.

Sources: The Decoder · The Register · NVIDIA