Leading aerospace and automotive companies are accelerating engineering workflows with Nvidia’s DoMINO NIM microservice, part of the PhysicsNeMo Artificial Intelligence physics framework. By combining GPU-accelerated computing, PhysicsNeMo and interactive digital twin technologies, enterprises report speedups of up to 500x versus traditional methods. The approach enables near real-time simulation of complex physical systems such as automobiles and aircraft, allowing teams to explore larger design spaces with higher accuracy and faster time to market.
Simulation providers including Ansys, part of Synopsys, are using PhysicsNeMo to deliver these gains in computational engineering. A key advantage is initializing fluid simulations with a highly accurate starting state, a step that typically requires many iterations and significant compute time. Nvidia notes that GPU-accelerated tools like Ansys Fluent can already run fluid simulations up to 50x faster than traditional CPU-based methods. Applying PhysicsNeMo’s pretrained models to generate the initial solution compounds that benefit by another 10x, yielding up to a 500x overall acceleration with improved accuracy and lower runtime cost.
In aerospace, Northrop Grumman and Luminary Cloud are applying accelerated compute and Artificial Intelligence-driven physics to spacecraft thruster nozzle design. Using Luminary’s Nvidia CUDA-X-accelerated computational fluid dynamics solver, Northrop generated a large training dataset to build a surrogate nozzle model on Luminary’s cloud platform powered by PhysicsNeMo. This enabled engineers to quickly test thousands of design options and converge on an optimal configuration, highlighting how surrogate modeling can unlock rapid iteration on mission-critical components.
Blue Origin is also tapping PhysicsNeMo and advanced Artificial Intelligence modeling to design next-generation space vehicles. By training models on existing and augmented datasets, the company can rapidly explore candidate designs and then validate promising options with high-fidelity, CUDA-X-accelerated solvers. This workflow blends data-driven insights with traditional simulation to move more efficiently from concept to validated design.
Beyond aerospace and automotive, the momentum extends across computational engineering. Cadence is advancing real-time simulation through its Fidelity platform using Nvidia CUDA-X libraries, enabling large-scale Artificial Intelligence training datasets on its Millennium M2000 supercomputer for interactive design optimization. A global energy solutions leader used Cadence Fidelity LES Solver with Nvidia Grace Blackwell-accelerated simulation platforms to iterate quickly on high-fidelity multiphysics models. The result was shorter design cycles and optimized turbine performance focused on efficiency, emissions management and reliability for next-generation energy systems.
