The Thinking Machine Podcast

We’re entering a new era in robotics. One where the bottleneck isn’t just algorithms, it’s the entire stack. The foundation models, the data pipelines, the simulation environments, the training infrastructure. All of it has to come together for robots to move from demos to deployment.

The Thinking Machine Podcast goes deep with the researchers, founders, and engineers working across this stack. That means conversations with teams building robotics foundation models like Groot and Gemini Robotics, architects of world models and neural simulators, and the people designing the data collection systems that make training possible at scale.

If you’re building in robotics, investing in the space, or trying to understand where this field is really headed, this podcast is for you.

Jonathan Stephens

Jonathan is Chief Evangelist at Lightwheel, a Physical AI infrastructure company building the data engine that powers robots to learn at scale. At Lightwheel, he works at the intersection of simulation, synthetic data, and embodied AI, helping the robotics industry understand the data-centric foundations of robot training.

Through The Thinking Machine Podcast, Jonathan explores these questions with the researchers, founders, and engineers building the field. On his Pixel Reconstruct platform, he breaks down the technologies shaping intelligent machines: world models, simulation environments, radiance fields, and the full data pyramid that determines what robots can learn and understand about the physical world.