The physical AI revolution will start in space for two reasons: 1. The problem space is bounded. Spacecraft are complex and dynamic, but Solo reasons across a known physical system, with defined hardware and telemetry, governed by physics, and a finite set of rules and possible actions. 2. The training data is organically generated as part of the spacecraft build process. And not just that – conditions that can't easily be reproduced physically are already simulated in test. With our hardware-in-the-loop testbed, we subject hardware to edge cases and anomalies and see how it responds. Our blog dives deeper. Link in reply.