Silvaco, NVIDIA Partner on AI-Powered Chip Design Digital Twins

On July 26, 2026, Silvaco Group, Inc. (Nasdaq: SVCO), a provider of TCAD, EDA software, and semiconductor IP solutions, and NVIDIA, announced a strategic collaboration to build next-generation digital twins for semiconductor design and manufacturing. The partnership combines Silvaco's decades of physics-based modeling expertise with NVIDIA's accelerated computing, CUDA-X libraries, PhysicsNeMo, Omniverse libraries, and Nemotron open models.

The collaboration unfolds across four focus areas. First, GPU-accelerated physics simulation: Silvaco will leverage NVIDIA accelerated computing and CUDA-X libraries to speed up its semiconductor device, process, photonics, and multiphysics simulation solutions, reducing simulation runtimes and increasing design productivity. As proof, Silvaco completed a fully scaled 3D FDTD (finite-difference time-domain) simulation of a photonic edge coupler with 3.2 billion mesh nodes on 32 NVIDIA GPUs connected with NVLink in under four hours. The workload did not converge on CPUs, and the result achieved less than 0.15 dB difference between measurement and simulation—a level of accuracy that validates the approach for production use. Second, AI-driven surrogate modeling: Silvaco intends to leverage NVIDIA PhysicsNeMo to develop customizable AI surrogate models that complement high-fidelity physics simulation and accelerate exploration of design alternatives. Third, digital twin visualization and collaboration: Silvaco plans to connect its digital twin environment with NVIDIA Omniverse libraries and NVIDIA Cosmos to deliver collaborative, real-time visualization and simulation environments spanning semiconductor fabs, manufacturing systems, robotics platforms, and infrastructure applications. Fourth, scaled engineering workflows: Silvaco aims to establish cloud-native workflows that support design, testing, and validation across distributed teams and compute environments, reflecting the reality that modern semiconductor design teams span multiple locations.

Together, the two companies expect to help customers design, simulate, and optimize increasingly complex semiconductor technologies. The partnership leverages Silvaco's two decades of physics-based modeling expertise—critical for maintaining accuracy in semiconductor simulation—combined with NVIDIA's leadership in accelerated computing and its growing portfolio of AI tools. By integrating these capabilities, Silvaco customers will have access to digital twins that can predict, optimize, and validate complex semiconductor systems with unprecedented speed and accuracy, directly addressing the challenge of shrinking design cycles in an era of rising chip complexity.

Silvaco, a provider of semiconductor design and simulation software, has long relied on physics-based modeling to help chip makers design and validate complex systems. However, traditional CPU-based simulation has struggled with the computational demands of modern semiconductor designs—as evidenced by the photonic simulation that could not converge on CPUs. By partnering with NVIDIA, Silvaco gains access to GPU acceleration and AI tools that can dramatically speed up these simulations and introduce new approaches like AI surrogate modeling. The partnership addresses a real bottleneck: semiconductor design cycles are lengthening as chips become more complex, and simulation time is a critical constraint. NVIDIA's accelerated computing and AI frameworks, combined with Silvaco's physics expertise, position both companies to serve customers racing to iterate faster on advanced chip designs.

The collaboration spans four distinct areas that together form a complete digital twin ecosystem. The GPU-accelerated physics simulation directly tackles runtime—Silvaco's proof point shows the impact is substantial (under four hours for a previously unconvergent workload). AI surrogate modeling adds a new capability: complementing full physics simulation with faster, trainable models that let design teams explore alternatives without running expensive full simulations each time. The visualization and collaboration layer, powered by Omniverse and Cosmos, acknowledges that modern chip design is a team activity spanning multiple sites and time zones. Finally, the push toward cloud-native workflows recognizes that semiconductor companies now operate distributed engineering teams. Together, these investments signal that Silvaco and NVIDIA see the future of semiconductor design as a blend of accelerated physics simulation, AI optimization, real-time collaboration, and cloud-scale compute.

No comments yet. Be the first to share your thoughts!