Humyn Labs raised $47M in Series A funding led by Accel, with $26M in equity and $21M in venture debt from Lighthouse Canton and Trifecta Capital. The startup operates data capture programs across 20 countries, primarily in Latin America, Southeast Asia, South Asia, and East Asia. It works with frontier AI labs, robotics companies, and world-model developers to gather labeled, training-ready datasets that bridge the simulation-to-deployment gap.

The company captures data across four modalities: vision, sound, touch, and motion. Using proprietary hardware, Humyn records how humans perform specific tasks in target environments — residential, commercial (warehousing, retail, hospitality), and industrial (manufacturing, assembly). The data is embodiment-agnostic, meaning it supports both stationary manipulation (robot arms) and mobile platforms (humanoids, AMRs) depending on task requirements.

Humyn claims $50M in annualized revenue within six months of operations, a rare velocity for a Series A infrastructure play. The funding will support deployment readiness facilities in Brazil, Korea, and India, plus expanded presence in San Francisco, London, and Seoul. Co-founder Manish Agarwal frames the value proposition as letting labs focus on model research while Humyn handles the logistics of global data collection and enrichment.

The raise underscores a structural bottleneck in Physical AI: most robotics development still happens in simulation, and real-world transfer remains slow and expensive. Humyn's bet is that high-quality, geographically diverse, multimodal data will become a competitive moat as labs race to ship embodied systems. If adoption scales, it positions Humyn as critical infrastructure in the robotics supply chain — and a potential acquisition target for hyperscalers or large robotics platforms seeking vertical integration.