NVIDIA DSX POD: the AI factory that ships in a box — 14 days from delivery to tokens

• Who / what: NVIDIA introduced DSX POD, a reference design for complete Vera Rubin AI factories built by partners inside modular containers.
• When / where: announced October 21, 2026, at the GTC Berlin keynote by CEO Jensen Huang; partner deliveries begin worldwide in early 2027.
• Deployment claim: from about six months to as little as 14 days — the on-site phase at a prepared site with power, cooling, and networking already in place.
• What's inside: full Vera Rubin systems plus computing, networking, cabling, rack-level power, and liquid cooling — integrated and factory-tested before shipment.
• Ecosystem: 17 partners, including Dell Technologies, HPE, Lenovo, Foxconn (Ingrasys), Schneider Electric, Vertiv, Anduril, Crusoe, GIGABYTE, Pegatron, and Quanta Cloud Technology (QCT).
• The economics: Huang said a one-gigawatt AI data center costs about $50 billion and should generate about $40 billion in annual revenue — so a three-month delay postpones roughly $10 billion in potential revenue.

AI factories are about to get the prefab treatment. At its GTC Berlin keynote on October 21, NVIDIA CEO Jensen Huang introduced DSX POD, a reference design for complete Vera Rubin AI factories that partners build inside modular containers — integrating and validating the whole stack at the factory, then shipping it as a finished product. At a prepared site, NVIDIA says the pods cut deployment from about six months to as little as 14 days.

The timing is deliberate. Europe’s AI build-out is the announced target — the press release is titled around accelerating Europe’s infrastructure deployment — but the container playbook is global, and it reframes NVIDIA from a chip vendor into a factory vendor.

What a DSX POD actually is

Each pod is a complete Vera Rubin AI factory in a container. Partners complete the rack-to-rack cabling, wiring, piping, and functional validation at a manufacturing facility — computing, networking, cabling, rack-level power, and liquid cooling, all factory-tested — so the units arrive ready to connect out of the box to a prepared site’s power, cooling, and networking. Factory testing is the quiet differentiator: it reduces exposure to component delays and the error-prone onsite work that stretches traditional builds.

“An AI factory creates value when it starts producing tokens — not when construction is complete,” Huang said in the announcement. “With NVIDIA DSX POD, we’re bringing factory-built precision and speed to AI infrastructure construction — reducing deployment time from months to weeks so customers can scale capacity and bring revenue-generating production online faster.”

The partners and the timeline

NVIDIA named 17 partners that will deliver DSX POD AI factories worldwide beginning in early 2027: Anduril, Armada, Bull, Crusoe, DayOne, Dell Technologies, Delta, Eaton, Foxconn (Ingrasys), GIGABYTE, HPE, Lenovo, Pegatron, Quanta Cloud Technology (QCT), Schneider Electric, Vertiv, and Wistron. Note the structure: NVIDIA publishes the reference design; partners manufacture and deliver it at scale.

Why 14 days is worth billions

On stage, Huang put the stakes in dollars: a one-gigawatt AI data center costs about $50 billion and should generate about $40 billion in annual revenue — so a three-month deployment delay postpones roughly $10 billion in potential revenue. When the bottleneck is installation and commissioning, shaving months off the on-site phase is not a logistics win; it is a revenue event. That is the number DSX POD is aimed at.

The power and water angle

Efficiency rides along with speed. NVIDIA says its DSX designs can cut cooling water use by up to 100% through closed-loop liquid cooling and dry coolers, and that its DSX MaxLPS software can deliver the same AI output with up to 30% less power, according to the company’s GTC Berlin keynote coverage. CoreWeave, Google Cloud, Lambda, Nebius, and Nscale are among those planning to use the designs — a reminder that the same companies racing to build capacity are also the ones most exposed to power and water scrutiny.

What to watch

Three things will decide whether DSX POD is a milestone or a marketing number: whether “as little as 14 days” holds anywhere but the ideal prepared site, how much factory capacity the 17 partners can actually line up by early 2027, and whether the container playbook stays a European story or becomes the default way AI factories get built everywhere.