Neural Concept was founded in 2019 as a spin-out from the Computer Vision Lab at EPFL in Lausanne, built on research into geometric deep learning that lets neural networks reason directly over 3D shapes rather than flattened images. The company's core insight is that traditional engineering relies on slow, expensive physics simulation loops, and that AI surrogate models trained on a company's own simulation and test data can predict performance almost instantly. This turns design exploration from a multi-week process into an interactive one, where an engineer can sweep through thousands of geometry variants and watch predicted drag, stress or temperature update in real time.
The platform is explicitly CAD-native: it understands geometry, constraints and design intent, and integrates with the CAD systems (Siemens NX, CATIA, SolidWorks) and simulation tools (Ansys, Abaqus, StarCCM+) engineering teams already use. On top of prediction, Neural Concept layers generative CAD capabilities and an AI co-pilot that proposes optimized shapes against multi-physics objectives, helping teams converge on designs that balance competing requirements across aerodynamics, structures, thermal and electromagnetics.
The results the company reports are concrete: customers save an estimated $50 million annually, cut late-stage redesigns by 30 to 50 percent, and accelerate time-to-market by up to two years. More than 50 global OEMs and suppliers rely on the platform, including General Motors, GE Vernova, Leonardo Aerospace, Eaton, Safran, Renault Group and several Formula 1 outfits.
In December 2025 Neural Concept closed a $100 million Series C led by Growth Equity at Goldman Sachs Alternatives, with existing backers Forestay Capital, Alven, HTGF, D.E. Shaw Ventures and Aster Capital participating. The round followed a $27 million Series B in 2024. The capital funds product development, including a generative CAD breakthrough, global go-to-market expansion, and deeper partnerships with Nvidia, Siemens, Ansys, Microsoft and AWS.