Gm gVangrid Humanoids and Physical AI are moving from demos into factories. Swiss startup Embodied AI just showed its Air robot that learns factory tasks by watching workers. NVIDIA has been pushing vision-language-action models. Agility and others are scaling deployments. The missing piece is fresh, ground-level spatial data of the actual environments these systems will operate in stairwells, loading docks, night crossings, warehouse interiors, plant floors. Satellites and street-view cars don’t capture it at the density or recency required. @vangrid_io already supplies exactly that: phone captures turned into verified 3D models, settled in USDC, with on-chain attestations. The grid is the sensing layer Physical AI needs as it leaves the lab.