Nvidia announced the DSX Pod AI Factory, a reference design that packs seven Vera Rubin racks into a unit roughly the size of a standard shipping container.
Each pod weighs about 28,000 kilograms, is built to travel by truck on ordinary roads, and arrives with all cabling and liquid cooling pre-assembled, according to Nvidia.
"The only thing that has to happen when it gets to the site is it's placed in its location, connected to the rest of the other containers, and validated through final burning and testing," Ian Buck, Nvidia's vice president of hyperscale computing, said on a call with reporters.
Nvidia said the design could cut the time needed to bring a new AI data centre online from six months to 14 days, a shift aimed partly at easing deployment constraints faced by European customers.
The move extends a progression Nvidia has followed from single GPUs to multi-GPU servers, then full racks, and now factory-integrated containers; each Vera Rubin NVL72 rack pairs 72 Rubin GPUs with 36 Vera CPUs, putting roughly 504 GPUs and 252 CPUs inside a single pod. Analysts estimate a single Vera Rubin rack can cost more than $7 million, though Nvidia did not disclose pricing for the pod itself. CEO Jensen Huang recently noted that the cost of outfitting one gigawatt of data centre capacity has climbed from about $18 billion to over $40 billion, and said he expects that figure to keep rising.
Nvidia is working with manufacturing partners including Foxconn, Gigabyte, Lenovo, HPE, Quanta and Pegatron, with deliveries expected to begin in early 2027.
Read more about Chipmakers
