Blogs
/
Nscale is developing digital twins to scale AI infrastructure
Company

Nscale is developing digital twins to scale AI infrastructure

October 21, 2026
•
2 min read

Nscale

Nscale is exploring how simulation can inform the way it designs, deploys, and operates AI data centers with NVIDIA Omniverse Libraries and NVIDIA DSX Sim.

Nscale today announced that it is developing digital twins to support the build-out of its AI infrastructure. The company intends to evaluate how physically accurate, virtual models of its facilities can inform decisions across the life of a site, from early design through day-to-day operations.

Nscale is using NVIDIA Omniverse Libraries and NVIDIA DSX Sim to understand how these technologies fit into its engineering and operations workflows.

Why digital twins, and why now

AI infrastructure is being built at a pace and density the industry has not seen before. Each new generation of AI infrastructure brings its own power, cooling, and layout requirements, and facilities need to accommodate them on agreed timelines. Decisions made early in a project shape how efficiently a site runs for years.

A digital twin is a virtual replica of a physical facility that combines design data, engineering models, and, over time, operational data. It gives teams a shared view of a site that they can test against before committing to changes in the real world. For AI data centers, that can mean simulating how air and heat move through a hall, how power is distributed across racks, or how a new hardware configuration fits within a facility’s limits.

What Nscale is using

NVIDIA DSX Sim is part of the NVIDIA DSX AI Factory Platform for designing, building, and operating AI factories. It brings together simulation-ready (SimReady) assets prepared with equipment properties and metadata for simulation, simulation technologies, and partner tools to help builders identify bottlenecks, evaluate trade-offs across systems, and compare expected and actual behavior after deployment.

Within DSX Sim, the NVIDIA Omniverse DSX Blueprint provides an OpenUSD-based framework for creating digital twins of AI factories and sharing 3D data across tools. It includes a reference site model and SimReady assets for modeling power, cooling, and compute configurations, helping engineering teams test and optimize facilities before construction and throughout operations.

Nscale is exploring where these tools could add value across three areas:

  • Design: Testing airflow, cooling, power, and rack layouts in simulation before hardware arrives on site.
  • Deployment: Using a shared model to help coordinate data center, hardware, and software teams as a site comes online.
  • Operations: Adding a spatial view of infrastructure, giving teams a clearer, more intuitive picture of their environment and how it’s performing.
“Digital twins can give us a way to test our decisions in simulation before they reach the data center floor. We’re looking forward to exploring what NVIDIA Omniverse and DSX Sim can bring to how we design and run our sites.”
David Power Chief Technology Officer, Nscale

‍

“Infrastructure design is becoming a defining factor in the scale and performance of AI factories. Nscale is leveraging NVIDIA Omniverse libraries and DSX Sim to bring a simulation-first approach to AI infrastructure — using digital twins to validate power, cooling, and facility layouts before deployment. This enables more informed engineering decisions, reduces operational risk, and accelerates the delivery of efficient, scalable AI infrastructure.”
Vladimir Troy Vice President, AI Infrastructure, NVIDIA

‍

By testing design, deployment, and operations decisions in simulation first, Nscale aims to shorten the path from site plan to live capacity and give customers infrastructure that performs as expected from day one. 

For more information on NVIDIA Omniverse and DSX Sim, visit nvidia.com/omniverse and nvidia.com/en-us/data-center/products/dsx/sim.