WorldEngine AI, founded just ten months ago, is building infrastructure for Physical AGI and has reportedly reached a $100M annualized run rate, $1M to $100M ARR in just 3 months,with 8 frontier lab customers and 250K hours of humanoid robot data processed per month across ego, glove, and teleoperation data.

Its data infrastructure is 90% automated, enabling an estimated 80% cost reduction while performing quality assurance at the point of capture.

The company has also developed Psi-0, an open model for dexterous humanoid manipulation.

Its team brings experience from MIT, Stanford, USC, UC San Diego, Google, Waymo, Microsoft, NVIDIA, AI2, Meta, Amazon, and ByteDance.