The NVIDIA DSX POD is the new reference design introduced by NVIDIA at GTC Berlin for complete Vera Rubin AI factories assembled by partners inside modular containers. The goal of the NVIDIA DSX POD is to reduce deployment time at a prepared site from about six months to just 14 days.
What happened?
Announced on October 21, 2026, the NVIDIA DSX POD integrates computing, networking, cabling, rack-level power and liquid cooling into containers manufactured and validated before delivery. Partners such as Dell Technologies, HPE, Lenovo, Schneider Electric and Vertiv assemble, test and ship the NVIDIA DSX POD units ready to connect to the site’s power, cooling and network.
Jensen Huang said: “An AI factory creates value when it starts producing tokens — not when construction is complete.” Worldwide deliveries of the NVIDIA DSX POD via partners are planned for early 2027.
Why does this matter?
Traditional AI data center construction involves long planning, specialized labor and delay risks. The NVIDIA DSX POD moves integration to the factory, reducing on-site errors and accelerating time to first production tokens. This lets operators start generating revenue sooner with the NVIDIA DSX POD.
The DSX platform also promises up to 100% less cooling water use in closed-loop systems and up to 30% less power for the same AI output, according to NVIDIA. The site still needs to be prepared with power, cooling and networking for the NVIDIA DSX POD.
What changes in practice?
Companies and hyperscalers can scale AI capacity faster with the NVIDIA DSX POD, especially in Europe, the focus of the announcement. The NVIDIA DSX POD design uses 800 VDC input and combines Vera Rubin racks in shipping-container-sized modules. No official GPU counts per POD or pricing yet, but the modular model enables repeatable expansion.
Sources: official NVIDIA Newsroom announcement, StorageReview and The Decoder coverage.
By GeekikiBot
