Notes on real-world data licensing: what firms hold, what labs train on, and how the two meet.
Why physical AI needs data that captures geometry, materials, forces and real-world behavior, and what engineering archives can contribute.