Vinci

Vinci builds a foundation model for physics that is meant to make engineering simulation fast enough to use throughout hardware design instead of only at the end. The company says conventional simulation tools take hours to set up and days to compute, and that they simplify the physics to fit the available compute, which widens the gap between a design and the part that is actually manufactured.

Its answer combines AI acceleration with accuracy checked against finite element analysis solvers. The site says the model works at design fidelity with no meshing, runs on GPUs, takes native design files such as OASIS, GDS and IPC-2581 layouts, and does not need customer data to train it. Thermal analysis of semiconductor packages is the example the pages develop, including a study presented at EPTC 2025 that covered a stacked package with ten layers and layout features below 7 nm.

The pages claim simulations up to 1000 times faster than solver runs, a worked comparison of 20 seconds against 2 hours at 117 million degrees of freedom, and benchmarking by more than half of the top 20 semiconductor companies. Buyers are hardware and semiconductor engineering teams, and a demo is requested through the site. The company announced a 250 million dollar Series B at a 1.5 billion dollar valuation on 6 October 2026.

Key facts

• Product: Physics AI foundation model for engineering simulation, with a thermal module
• Inputs: native layout and design files, no meshing
• Serves: semiconductor and hardware engineering teams
• Funding: 250 million dollar Series B, announced 6 October 2026
• Pricing: not published; demo on request