Humyn Labs, which supplies training data to robotics companies and AI labs, announced a $47 million Series A on Nov. 19. Only $26 million of it is equity, led by Accel. The other $21 million is venture debt led by Lighthouse Canton, according to the company’s release.
Humyn Labs says it has passed $50 million in annualised revenue within six months of operations, working with frontier AI labs, robotics companies and world-model developers, with capture and enrichment operations in more than 20 countries. The company was founded in 2026, and the revenue figure is its own and unaudited. In an interview with SiliconANGLE, co-founder Manish Agarwal described a video corpus of up to 200,000 hours of human motion, recorded with head-mounted cameras during everyday household tasks. The release says the money goes to deployment-readiness infrastructure in Brazil, South Korea and India, capture hardware and enrichment engines, and offices in San Francisco, London and Seoul.
Venture debt is repaid with interest, so about 45% of this round is borrowed against the revenue Humyn Labs reports. For a buying committee, the more useful detail is what the supplier sells: “labelled, training-ready datasets with verified provenance,” in the release’s words, recorded by workers in factories, warehouses, farms and homes. Agarwal said in the release: “Physical AI will only be as good as the real-world data it learns from.”
In our reading, a company piloting robots in its own plants will increasingly be asked to supply that data, or to buy it from a vendor like this one. Before signing, it should ask who consented to the recordings, where they were captured and who holds the rights to a model trained on them. We covered the tooling side in NVIDIA’s Cosmos data factory blueprints and its Flora robot labeling models.
Source: PR Newswire