Nvidia introduced Omniverse DSX during its GTC event in Washington, unveiling a transformative blueprint for modeling gigawatt-scale AI datacenters. This approach leverages digital twin technology to streamline the design and operation of AI facilities. At the heart of this initiative is Nvidia’s new AI Factory Research Center in Manassas, Virginia, designed to harness the capabilities of the Omniverse simulation environment in conjunction with open-source OpenUSD technology.

CEO Jensen Huang emphasized the significant shift in Nvidia’s strategy: “We commenced with chip design, progressed to systems, and advanced to AI supercomputers. Now, we’re engineering comprehensive AI factories designed to optimize productivity and intelligence at unprecedented scales.” The platform’s digital twin enables continuous performance and efficiency enhancements, acting as a datacenter’s operating system upon going live.

Partnering with Schneider Electric, the initiative highlights how the Omniverse-based platform redefines datacenter architecture. These AI-centric factories stand apart from traditional datacenters by focusing solely on AI-driven tasks. According to Huang, these facilities will underpin modern economies globally.

In addition to unveiling Omniverse DSX, Nvidia introduced the BlueField-4, its latest data processing unit engineered to enhance infrastructure operations within AI datacenters. The DPU, boasting significant computational advances, is primed to drive the next generation of AI data storage solutions as they prepare for widespread adoption by 2026.