Indian startup Humyn Labs has raised $47 million from investors including Accel to supply the training data that U.S. robotics companies need to build capable machines. The company sits in the data layer of the robotics stack: models that control real hardware have to be trained on large volumes of recorded human and environment behaviour, and that data is increasingly scarce and expensive to produce.

The round is a bet on that bottleneck. U.S. robotics developers have the models and the capital but often lack the labour-intensive pipelines needed to generate structured, high-quality physical-world data; Humyn Labs is positioning itself as the supplier that fills that gap from India.

It also continues a funding streak across the robot-data corner of the market. Mecka AI recently closed a Sequoia-led round that valued it near $500 million, and XDOF reached a Series B at a $1.2 billion valuation just months after leaving stealth — a run of deals that has made training data one of the most actively financed parts of the robotics supply chain.