Alvin Lang
Nov 18, 2025 03:18
NVIDIA unveils Apollo, a new family of open AI models, to enhance scientific simulations across industries like automotive and aerospace, offering significant speed and accuracy improvements.
NVIDIA has introduced a groundbreaking family of open models, known as NVIDIA Apollo, aimed at accelerating industrial and computational engineering. Announced at the SC25 conference in St. Louis, this initiative leverages NVIDIA’s AI infrastructure to enhance real-time capabilities in simulation software across various sectors, including automotive and aerospace, according to NVIDIA.
Revolutionizing Scientific Simulations
The Apollo models are designed with scalability, performance, and accuracy in mind, covering fields such as electronic device automation, structural mechanics, weather and climate, computational fluid dynamics, electromagnetics, and multiphysics. These models incorporate advanced AI techniques like neural operators, transformers, and diffusion methods, combined with domain-specific knowledge.
Industry Adoption and Impact
Several industry leaders, including Applied Materials, Cadence, LAM Research, and Siemens, have started adopting these models to refine their design processes. For instance, Applied Materials is utilizing NVIDIA AI physics to enhance semiconductor manufacturing processes, achieving up to 35x acceleration in its ACE+ multi-physics software.
Cadence has utilized its software, accelerated by NVIDIA’s Millennium M2000 Supercomputer, to produce datasets for AI physics models, enabling real-time digital twins of aircraft. LAM Research focuses on plasma reactor simulation, critical for semiconductor manufacturing, while KLA explores faster simulations to enhance semiconductor process control solutions.
Enhancements in Engineering and Design
Northrop Grumman and Luminary Cloud are leveraging NVIDIA AI physics to advance spacecraft thruster nozzle designs. The use of NVIDIA CUDA-X libraries allows rapid exploration of design options. Similarly, PhysicsX employs NVIDIA AI infrastructure to shorten product development cycles in automotive and aerospace industries.
Rescale integrates Apollo models into its AI physics operating system, enabling engineers to combine high-fidelity simulations with AI surrogates for faster design exploration. Siemens integrates these models into Simcenter STAR-CCM+, enhancing fluid simulation tools.
Synopsys reports significant speedups using NVIDIA AI physics models, with improvements up to 500x in computational engineering, particularly in fluid simulations using tools like Ansys Fluent.
The NVIDIA Apollo models will soon be available on platforms like build.nvidia.com and HuggingFace, offering pretrained checkpoints and workflows for training and benchmarking, thus allowing developers to tailor models to their specific requirements.
Image source: Shutterstock
Source: https://blockchain.news/news/nvidia-apollo-ai-physics-scientific-simulation
