The $15B Physical AI Company: Simulation, Autonomy OS, Neural Sim, & 1K Engineers, Applied Intuition

In this talk

1. 2:56

   Building a technology stack for physical AI

   Applied Intuition sells software to machine makers and governments across automotive, construction, mining, agriculture, and defense. Ludwig and Younis describe three core areas: simulation and development tools, operating system technology, and AI models. The company says about 83% of its roughly 1,000 employees are engineers.

2. 16:32

   A shared operating system for machines

   Ludwig describes the vehicle OS as a safety-critical foundation for real-time control, sensor data, memory, networking, and reliable software updates, beyond the cabin display. Applied Intuition says its OS supports multiple chip architectures and can be licensed on its own or with the company’s autonomy software. The Android analogy is a way to explain how a common platform can help AI applications run across fragmented hardware.

3. 28:28

   Simulation, neural models, and safety evaluation

   The founders discuss simulation as part of a loop that must be checked against real-world behavior. They describe neural simulation as a fast way to generate sensor data for training end-to-end autonomy models, while emphasizing that physical testing remains necessary. For statistical AI systems, they say evaluation must measure reliability and failure rates, and give overheating robot actuators as an example of a physical constraint that simulation should represent.

4. 45:27

   Deploying models within real-world constraints

   Ludwig distinguishes large offboard models from onboard systems, which must produce results within tight limits on latency and power. Applied Intuition says it develops its autonomy models in-house, while general-purpose models may suit uses such as voice assistance. The conversation also covers production engineering, the difficulty of moving from demonstrations to dependable deployment, and the need to make models smaller and easier to evaluate.

5. 58:46

   Founder choices and engineering talent

   Younis advises founders to account for commercial constraints as well as technical ones, and to begin with a focused product problem that can grow over time. He cautions against copying advice from companies at different stages or in different markets. Ludwig says the company values engineers who understand hardware and software together, have production experience, and bring deep curiosity.