Descriptions:
Applied Intuition, the $15 billion physical AI company, sits down with the Latent Space podcast for an in-depth founder conversation with CEO Kasser Ghosn and CTO Peter Ledwig. The discussion covers Applied Intuition’s mission to build what they describe as the ‘Android of physical AI’—a unified autonomy operating system addressing the extreme software fragmentation across vehicles, trucks, construction equipment, and defense systems. The company counts 18 of the top 20 global non-Chinese automakers as customers and currently runs driverless L4 trucks in Japan.
The founders explain Applied Intuition’s neural simulation work—a hybrid approach combining Gaussian splatting and diffusion methods—as the key enabler for reinforcement learning on end-to-end autonomy models. Unlike traditional simulation using physics formulas and binary pass/fail tests such as Euro NCAP’s automatic emergency braking scenarios, modern end-to-end models require full sensor data simulation and probabilistic reliability metrics measured in orders-of-magnitude ‘nines.’ Performance is everything: simulation that is not fast and cheap enough simply cannot produce competitive training results, a constraint that shapes the company’s entire infrastructure investment.
The interview also traces Applied Intuition’s origin story—initially compared to Scale AI for data infrastructure before expanding into the full autonomy stack—and discusses how the YC-backed startup grew to a $15 billion valuation by targeting safety-critical physical AI markets where software errors are not recoverable.
📺 Source: Latent Space · Published April 27, 2026
🏷️ Format: Interview
