Chip software maker Silvaco is joining forces with Nvidia to speed up the creation of physics-based "digital twins" for semiconductor design and manufacturing, according to announcements carried by Quiver Quantitative and The Manila Times.
The collaboration centers on using Nvidia's AI and accelerated computing to power Silvaco's simulation work, per The Manila Times, which framed the effort as accelerating physics-based digital twins across both the design and manufacturing sides of the chip business. Quiver Quantitative described the announcement as a collaboration to advance AI-powered semiconductor digital twins.
A digital twin, in plain terms, is a highly detailed virtual replica of a real object or process — in this case, the transistors, materials and fabrication steps that go into making a chip. "Physics-based" means the simulation tries to model the underlying physical behavior rather than merely approximating outcomes. Engineers can then test ideas on a computer instead of running expensive experiments in a fabrication plant.
That is where Nvidia's role comes in. Accelerated computing — running heavy calculations on graphics processors rather than conventional processors — is well suited to the enormous number-crunching that physical simulation requires, and layering AI on top can help predict results faster than brute-force math alone.
The available source items are brief announcement write-ups and do not disclose financial terms, a timeline, specific products, or named customers, so the practical scope of the work remains to be seen.
Why it matters: designing and manufacturing advanced chips is slow, costly and increasingly difficult, so anything that shifts trial-and-error out of the factory and into simulation could shorten the path from concept to working silicon — and it extends Nvidia's reach further into the tools used to build chips, not just the chips themselves.