Building an AI data centre usually means months of construction, cabling and testing on site. Nvidia wants to move most of that work into a factory.
At GTC Berlin on 21 October, Nvidia introduced DSX POD, a reference design for complete Vera Rubin AI factories built by partners inside modular containers. The units are integrated and validated before they ship, then arrive ready to connect to a prepared site’s power, cooling and networking. Nvidia says that cuts deployment from six months to as little as 14 days.
The idea is straightforward. Computing, networking, cabling, rack-level power and liquid cooling are installed and tested at the manufacturing facility. Partners finish the rack-to-rack cabling, wiring, piping and functional checks before shipment, while the site itself is being prepared in parallel.
In his keynote, Jensen Huang described the building block: ‘It starts with Vera Rubin NVL72. Seven liquid-cooled compute racks connected with 17 km (11 miles) of Spectrum-X cables form one Vera Rubin DSX POD,’ as reported by DCD. Channel Insider adds that the core containers use a standardised 800-volt DC power input.
There’s an important catch, and Nvidia is clear about it. The 14-day figure assumes a prepared site. It doesn’t remove onsite work entirely. Customers still need the upstream electrical supply, facility water, cooling connections and external networking ready, and final commissioning happens after delivery. Channel Insider notes that delays in power access could keep factory-tested equipment from going live.
Nvidia says the initial focus is large, new, liquid-cooled AI facilities, including its cloud partners. It’s working with a long list of companies, among them Dell Technologies, HPE, Lenovo, Schneider Electric, Vertiv, Foxconn and Bull, to deliver DSX POD AI factories worldwide in early 2027. The company pitched it at GTC Berlin as a way to accelerate Europe’s AI infrastructure.
Why it matters
Compute is the bottleneck in AI right now. If factory-built pods really shorten the time from site readiness to running models, more countries and companies can get AI capacity online sooner. The lesson for planners is that the hard part shifts to site preparation: power and cooling now decide the schedule. That’s worth watching for fast-growing markets like India, where demand for local AI infrastructure keeps rising.
Sources: NVIDIA Newsroom, Channel Insider, DCD. Written by the xRapid Media LLP team.