Physical AI Training Data Platform Raises $30M as Data Infrastructure Gains Focus

The funding round puts a spotlight on a less visible layer of embodied AI: collecting and processing real-world interaction data. Instead of relying only on algorithms, the company pays ordinary participants to record everyday physical interactions in genuine environments, then refines those recordings into structured datasets for robot learning. The $30 million raise suggests investors view data infrastructure—not just models—as a bottleneck and an opportunity in physical intelligence. As robots move from controlled tests toward open-ended settings, diverse human-recorded experience may become as important as compute and model design. The round also highlights a shift in competitive focus: as model architectures converge, proprietary datasets may determine which embodied systems perform reliably outside labs. The company’s model points to a distributed human data supply chain, though quality control, privacy, and scalability remain open questions.