An AI Superfactory (also called an AI Factory) is NVIDIA's term for a purpose-built computing facility designed specifically to manufacture artificial intelligence — training models, running inference, and generating tokens at industrial scale. Unlike traditional data centers optimized for storing and retrieving data, AI Superfactories are engineered for continuous, high-density compute workloads with specialized power, cooling, and networking infrastructure. 1)
Jensen Huang, NVIDIA's CEO, has described AI factories as the defining infrastructure of the AI industrial revolution — facilities that take in raw data and produce intelligence as their output, analogous to how traditional factories take in raw materials and produce physical goods. 2)
Key distinctions from traditional data centers:
The DGX SuperPOD is NVIDIA's reference architecture for AI factories, built from modular Scalable Units (SUs):
The Vera Rubin DSX AI Factory reference design (announced GTC 2026) provides a codesigned infrastructure guide for maximum tokens-per-watt and accelerated time to first production. It pairs with the Omniverse DSX Blueprint for digital twin simulation of the entire facility. 4)
The DGX SuperPOD architecture is designed as a “physical twin” of NVIDIA's own internal R&D systems, ensuring that software, applications, and support are pre-tested on identical infrastructure. Using Scalable Units, deployment times are reduced from months to weeks. 6)