NVIDIA said companies are using its DSX AI Factory Platform to plan, simulate and run power grids and AI factories as one system. The company shared the update in a blog post published on Oct. 22, 2026.
NVIDIA described the effort as a response to a growing power bottleneck in data center expansion. It said more than $300 billion in equipment is backlogged.
The company also said more than 200 gigawatts of projects are waiting in U.S. interconnection queues. These are lines of projects waiting to connect to the power grid.
How the DSX Platform Works
NVIDIA said part of the problem comes from siloed planning. The grid, the facility and the computing equipment have usually been designed separately.
DSX is meant to bring those layers together. NVIDIA said it uses DSX Sim and SimReady assets to model substations, generators, switchgear, power systems and cooling before construction begins.
The platform also includes DSX Flex. NVIDIA said this tool is designed to match AI factory demand with grid conditions and onsite power generation.
Two other tools, DSX MaxLPS and DSX Exchange, are meant to improve computing output per watt. They also share operating signals across compute, networking, energy, cooling and operations.
Partners and Early Results
NVIDIA named several partners working with the platform. GE Vernova is pairing its grid-to-facility and rack-power experience with DSX, according to the post.
Hitachi Energy’s battery energy storage system has qualified as NVIDIA DSX Ready. NVIDIA said this label helps customers find equipment that meets its reference design requirements for AI factories.
Schneider Electric, working with AVEVA and ETAP, is advancing validated power and cooling designs for NVIDIA’s Vera Rubin systems. The work also includes a digital twin powered by NVIDIA Omniverse.
Siemens Energy is developing a reference design for grid-aware power infrastructure. That design is aligned with the Vera Rubin AI factory layout.
NVIDIA said Phaidra added DSX Flex, DSX Max-Q and DSX Exchange to its self-learning software agents. The company said the setup delivers about 10% more compute within the same power limit.
Emerald AI added its Conductor platform to DSX Flex to help utilities use spare grid capacity. NVIDIA said the model could unlock up to 100 gigawatts on the existing U.S. grid.
That figure is a modeling claim made by the companies involved. It has not been independently verified.
NVIDIA describes DSX as a set of reference designs, software tools, libraries and partner technologies. The stated goals include faster deployment, lower token costs, better energy efficiency and more reliable operations.
The announcement follows several DSX updates in recent weeks. NVIDIA launched DSX Ready for power and cooling products on Sept. 21 and open sourced its Run:ai software on Oct. 13.
Most recently, on Oct. 21, the company introduced DSX POD AI factories for Europe, one day before this latest update.
