Physical AI that Moves the World — Qasar Younis & Peter Ludwig, Applied Intuition

In this episode of Latent Space, Alastio and Swix host Qasar Younis and Peter Ludwig, co-founders of Applied Intuition, to explore the frontier of physical AI—intelligence embedded in real-world machines beyond screens. Applied Intuition, a technology company with over 1,000 engineers and 40+ founders, builds foundational software for autonomous systems across automotive, defense, agriculture, construction, and mining. Unlike consumer-focused AI, their work centers on safety-critical, real-time systems where reliability, low latency, and hardware-software integration are paramount. The company operates as a 'Nvidia for physical AI'—providing operating systems, simulation tools, and AI models that enable OEMs and governments to deploy intelligent machines at scale. They emphasize a 'compounding technology' strategy, where each layer of their stack—simulation, OS, AI models—evolves over time and reuses prior work. The discussion dives into the challenges of deploying AI on embedded systems, the shift from binary testing to statistical validation in safety, and the critical role of simulation in closing the 'sim-to-real' gap. The hosts also reflect on the underrated difficulty of productionizing AI, the importance of human-in-the-loop systems, and the long-term vision of making physical AI as reliable and pervasive as modern software. Key takeaways include: 1) Physical AI demands a deep integration of hardware and software, with performance and safety as non-negotiables; 2) The future of autonomy lies in hybrid approaches combining world models, reinforcement learning, and traditional simulation; 3) Success in hard tech requires narrowing focus to a small, well-defined problem space and building a compounding technology stack; 4) The most valuable engineers are those who understand the full stack—from low-level systems to AI—especially at the hardware-software boundary; 5) Simulation is not a replacement for real-world testing but a powerful tool to accelerate development and close the gap between virtual and physical performance. The overall sentiment is highly positive, reflecting excitement about the tangible impact of physical AI and the founders' pragmatic, long-term vision.