Silvaco Group and NVIDIA to push digital twins for semiconductor design and manufacturing to the next level. The companies announced a collaboration that combines Silvaco’s decades of physics-based modeling expertise with NVIDIA’s accelerated computing and AI stack, including CUDA-X™ libraries, PhysicsNeMo, Omniverse libraries, and Nemotron open models.
Together, they aim to help customers build, train, and deploy high-fidelity digital twins that can predict, optimize, and validate complex semiconductor systems with far greater speed and accuracy.
At the core of the partnership is GPU-accelerated physics simulation. Silvaco plans to use NVIDIA accelerated computing and CUDA-X™ libraries to turbocharge its device, process, photonics, and multiphysics simulation tools, cutting runtimes and boosting engineering productivity.
As an early proof point, Silvaco ran a fully scaled 3D FDTD simulation of a photonic edge coupler with 3.2 billion mesh nodes on 32 NVIDIA GPUs connected via NVLink in under four hours after the same workload failed to converge on CPUs achieving less than 0.15 dB difference between measurement and simulation.
The collaboration also targets AI-driven surrogate modeling. By leveraging NVIDIA PhysicsNeMo, Silvaco intends to develop customizable AI surrogate models that sit alongside high-fidelity physics simulations, enabling faster exploration of design alternatives without sacrificing accuracy.
On the visualization side, the company plans to link its digital twin environment with NVIDIA Omniverse™ libraries and NVIDIA Cosmos™ to deliver collaborative, real-time visualization and simulation across semiconductor fabs, manufacturing systems, robotics platforms, and infrastructure creating interactive environments that span the full design and manufacturing lifecycle.
To support modern engineering practices, Silvaco aims to establish cloud-native workflows that allow design, testing, and validation to run across distributed teams and compute environments.
By combining their technologies, Silvaco and NVIDIA expect to help customers cut simulation cycles from weeks to days through GPU-accelerated simulation and AI-driven modeling, improve accuracy and insight via high-fidelity digital twins, and scale engineering and collaboration using cloud-based visualization and AI-enabled workflows.
The joint solutions are expected to unlock advanced digital twin applications, including semiconductor process, device, packaging and photonics simulation; AI-assisted development of next-generation chips and advanced nodes; and factory optimization and predictive manufacturing.
This collaboration brings together Silvaco’s semiconductor modeling strengths with NVIDIA’s accelerated computing and AI platform to advance high-fidelity simulation, AI surrogate models, and digital twins across the semiconductor design and manufacturing value chain.
Leadership Comments
“The convergence of physics-based simulation, accelerated computing, and artificial intelligence is transforming design and manufacturing,” said Walden C. Rhines, President and Chief Executive Officer of Silvaco. “By combining Silvaco’s deep expertise in semiconductor and multiphysics digital twins with NVIDIA’s industry-leading computing and AI platforms, we can help customers model increasingly complex systems with greater speed, fidelity, and confidence. Together, we are positioning the industry for a future where AI-powered digital twins can fundamentally transform how semiconductor technologies are designed, validated, and optimized.”
“Digital twins are becoming essential tools for engineering and manufacturing innovation,” said Da Yang, senior director of product, semiconductor and EDA at NVIDIA. “By using NVIDIA AI, open models, libraries and accelerated computing, Silvaco is connecting high-fidelity simulation, helping customers move faster from modeling to insight across semiconductor design and manufacturing.”
To Know More: CLICK HERE
