Skip to main content

Neural Concept

Category: Foundation Models / LLMs

An AI-native 'Engineering Intelligence' platform that leverages Geometric Deep Learning to predict physical properties of 3D designs in real-time, enabling engineers to optimize and generate complex products instantly. Neural Concept was founded in 2018. The company is led by Pierre Baqué. Based in Lausanne, Switzerland. Team size: 60-100. Total funding raised: $138.1M. Latest round: Series C ($100.0M, Dec 2025). Key investors include ["Goldman Sachs Alternatives","Forestay Capital","D. E. Shaw group","Alven","High-Tech Gründerfonds (HTGF)"].

Founded
2018
Headquarters
Lausanne, Switzerland
Team size
60-100
Total funding
$138.1M

Value proposition

Reduces simulation time from days to seconds (up to 1000x faster than traditional CFD/FEA) and enables 'Generative Design'—allowing engineers to explore thousands of design iterations and automatically generate geometries that meet specific performance criteria.

Products and solutions

["Neural Concept Shape (NCS) - Flagship Platform","Generative Design Copilot","3D Deep Learning API","Predictive Simulation Library"]

Unique value

Pioneered 'Geometric Deep Learning' (3D Deep Learning) which processes 3D data in its native non-Euclidean format (meshes/point clouds) rather than converting it to 2D images, retaining topological accuracy.

Target customer

Global OEMs and Tier 1 suppliers in Automotive, Aerospace, Consumer Electronics, and Microelectronics (e.g., Airbus, Bosch, Subaru, Safran)

Industries served

["Automotive (EVs, Aerodynamics)","Aerospace & Defense","Consumer Electronics (Thermal Management)","Energy & Microelectronics"]

Technology advantage

First-mover advantage in 'Physics-Informed AI' for CAE; the platform acts as a surrogate model that learns from historical simulation data to predict physics fields (stress, heat, fluid flow) with solver-level accuracy in real-time.

How they differentiate

Pioneered 'Geometric Deep Learning' (3D-native AI) that processes raw CAD/mesh data directly without 2D conversion, enabling real-time physics prediction with solver-level accuracy.

Main competitors

["PhysicsX","Monolith AI","SimScale"]

Key partnerships

["NVIDIA (Deep collaboration on Omniverse Blueprint & Inception Program)","Ansys (Software integration partner)","Siemens (Digital Industries Software partner)","Goldman Sachs Alternatives (Lead Investor/Strategic Partner)"]

Notable customers

["Subaru","Airbus","Bosch","Safran","Williams Racing"]

Major milestones

["Raised $100M Series C led by Goldman Sachs Alternatives (Dec 2025)","Secured $27M Series B led by Forestay Capital (June 2024)","Strategic collaboration with NVIDIA (Omniverse integration)"]

Growth metrics

Technology adopted by 40% of top European and Asian OEMs; reduces simulation time from days to seconds (up to 1000x faster).

Market positioning

Enterprise 'Engineering Intelligence' platform (moving beyond simple simulation acceleration to generative design)

Geographic focus

Europe (HQ in Switzerland), North America (US expansion focus), Asia-Pacific (strong traction in Japan)

Patents and IP

Holds patents related to 'Method for shape optimisation of a technical device' using Convolutional Neural Networks (CNNs) on surface meshes (Inventor: Pierre Baqué).

About Pierre Baqué

Pierre Baqué is the Founder and CEO of Neural Concept. He holds a PhD in Computer Vision (Deep Learning for 3D shape analysis) from EPFL and a Master's in Applied Mathematics from École Polytechnique. Prior to founding Neural Concept in 2018, he was a Visiting Research Scientist at Honeywell and a Quantitative Trading Intern at Credit Suisse.

Official website: