Nvidia’s Modular AI Data Centers Aim for 14-Day Rollouts
WHO:
CNBC - Ian Buck, Nvidia’s vice president of hyperscale computing, said the pod only needs to be placed, connected to other containers, and validated through final burning and testing upon arrival.
CNBC - Nvidia said it is working with manufacturing partners Foxconn, Gigabyte, Lenovo, HPE, Quanta, and Pegatron.
DatacenterDynamics - Nvidia unveiled the DSX POD, a modular version of its Vera Rubin reference architecture for AI data centers.
IJR - Nvidia introduced DSX POD, a factory-built modular design for Vera Rubin AI facilities.
WHAT:
CNBC - The DSX Pod holds seven Vera Rubin AI racks, weighs about 28,000 kilograms (nearly 62,000 pounds), and is designed to travel by truck on normal roads.
CNBC - Nvidia did not announce a price; analysts said a single Vera Rubin rack can cost more than $7 million.
CNBC - Jensen Huang said Nvidia’s systems offered per gigawatt rose from about $18 billion to more than $40 billion.
DatacenterDynamics - Nvidia says DSX POD could reduce data-center deployment from six months to 14 days.
IJR - Nvidia says DSX POD could cut deployment at prepared sites from six months to as little as 14 days.
IJR - The broader DSX platform is designed to reduce cooling-water use and power consumption; the efficiency claims depend on deployment conditions.
WHERE:
CNBC - Nvidia said the pod is intended for customers worldwide and could help address deployment challenges faced by European customers.
WHEN:
CNBC - Nvidia expects DSX Pod deliveries to start in early 2027.
DatacenterDynamics - Details about DSX POD were released at Nvidia’s GTC Berlin event, described as taking place “this week.”
WHY:
DatacenterDynamics - The modular design is intended to shorten AI data-center deployment.
IJR - The broader DSX platform is intended to reduce cooling-water use and power consumption.
HOW:
DatacenterDynamics - DSX POD uses a modular, containerized version of Nvidia’s Vera Rubin reference architecture.
IJR - DSX POD is factory-built and designed for deployment at prepared sites.
(UPDATED October 21, 2026 16:16 UTC)
Nvidia has unveiled DSX POD, a factory-built, modular system designed to speed up construction of AI data centers using its Vera Rubin architecture. The company says deployment at prepared sites could shrink from six months to as little as 14 days. Each pod contains seven AI racks, weighs about 28,000 kilograms, and can travel by truck on ordinary roads. On arrival, Nvidia says, it needs to be placed, connected to other containers, and tested. Manufacturing partners include Foxconn, Gigabyte, Lenovo, HPE, Quanta, and Pegatron. Nvidia says the system is intended for customers worldwide, including those facing deployment challenges in Europe, and expects deliveries to begin in early 2027. The company did not announce a price; analysts estimate a single Vera Rubin rack could cost more than $7 million. Nvidia also says its broader platform aims to reduce power use and cooling-water consumption, depending on conditions.
--
(UPDATED October 21, 2026 13:14 UTC)
Nvidia has unveiled DSX POD, a factory-built, modular version of its Vera Rubin architecture for AI data centers. The company says the design could cut deployment time at prepared sites from six months to as little as 14 days. Nvidia says its broader DSX platform is also designed to reduce cooling-water use and power consumption, though efficiency depends on deployment conditions.