Why humanoid robots are stuck in their GPT-2 era
venture investment poured into 'physical ai' has hit a data bottleneck. specialized robots like excavators are deploying, but general humanoids remain stuck in labs.
the 'physical ai' sector is attracting billions in venture capital, highlighted by unitree's volatile ipo, but faces a critical 'data crisis' preventing commercial reliability. developers compare the current state of robotics to the gpt-2 era, requiring more data and specialized compute like ray-tracing gpus to advance. while autonomous vehicles benefit from human-driven data, general humanoids struggle to perform value-creating work. consequently, companies like tesla and wayve are leveraging av tooling to build humanoids, though successful deployments remain limited to specialized industrial machines like excavators.