Summary
The author argues that Physical AI investing should start with the 'learning layer' — the data engineering, model evaluation, and safety infrastructure that enables physical systems to understand and act in the real world — rather than with the robots themselves. This positions the portfolio thesis around companies already paid for AI learning cycles today, with a credible path into Physical AI revenue as the technology transitions from digital to physical tasks.
-
•The author plans to build a Physical AI portfolio from scratch, starting with the learning layer (data, demonstrations, evaluation, safety, feedback) rather than the robot hardware.
-
•The specific company being recommended is not named in this article; the author only says it is 'already being paid by major AI customers' for data engineering and model evaluation, with early Physical AI revenue.
-
•The article serves as a preview of a forthcoming standalone piece that will explain the full portfolio construction, weightings, price targets, and positions.