Physical Economy Picks & Shovels

The idea

I think the next decade of returns shifts from pure software into the physical world, with AI finally making machines smart enough to matter. Instead of picking which robot or defense name wins, I own the bottleneck suppliers everyone needs: motion control, machine vision, edge chips, connectors, and power gear. I size the hyped AI-power names smaller and lean heavier into out-of-favor automation names still below their old highs, with a small cash buffer. I'd rethink this if the physical-AI buildout stalls or the bottleneck names stop being bottlenecks.