AVEVA has moved deeper into NVIDIA’s push to standardise gigawatt-scale AI factory design, integrating engineering, simulation, and operations software into the Omniverse DSX Blueprint as hyperscale data centre projects begin to take on the structure of engineering, procurement, and construction programmes.
NVIDIA’s current DSX push is aimed at allowing developers to model, simulate, and optimise large-scale AI facilities before construction and during operations. AVEVA’s role is to bring in the industrial software stack more commonly associated with process plants, utilities, and capital projects, extending digital twin methods into a class of infrastructure now being treated less like conventional IT space and more like industrial capacity.
The practical effect is a tighter digital thread across design, build, and operations. AVEVA said customers will be able to bring OpenUSD SimReady assets into Unified Engineering, manage equipment changes through Asset Information Management, simulate advanced liquid-cooling networks in Process Simulation, and combine IT and OT data through PI System and the Omniverse DSX Exchange.
Power density, cooling performance, and commissioning speed are now shaping AI factory economics as much as server availability. Facilities built around dense GPU clusters leave less room for late redesign, while the need to coordinate electrical, mechanical, controls, and facility-management systems is pushing data centre delivery toward more industrial execution models.
Rob McGreevy, chief product officer at AVEVA, said AI factories are becoming “the industrial-scale engines of the global digital economy”, adding that domain-specific digital twins will be needed to design, build, and operate them at useful scale. Vladimir Troy, vice president of AI Infrastructure at NVIDIA, said the integration is intended to give developers a unified digital twin architecture spanning the full lifecycle of these sites.
The collaboration also sits alongside work involving Schneider Electric and ETAP on power-system and infrastructure modelling around AI factory design. That broadens the proposition beyond visualisation and into power, cooling, telemetry, and operations management, which is where some of the hardest constraints in AI facility deployment now sit.
For AVEVA, the move is also a clear sector extension. If AI campuses are heading toward utility-scale footprints, the software stack used to build and run them will increasingly overlap with the tools already used across energy, process, and industrial infrastructure.
