Mecka AI secures $60M to build human‑motion data layer for robots

Embodied AI has long been constrained less by algorithms than by data: there simply isn’t much high‑quality, task‑rich human motion at scale. Mecka AI is explicitly trying to become the “data layer for physical AI,” and a $60 million war chest validates that thesis. By industrializing collection via body sensors, iPhones and custom capture rigs, Mecka is betting it can build a proprietary corpus that general‑purpose humanoid and manipulation models will eventually depend on. That is directly relevant to AGI trajectories where physical interaction with the world is as important as language or code.

Strategically, this round cements a new niche alongside GPU clouds and model labs: specialized data providers whose moats are logistics and IP rather than parameters. If Mecka becomes the go‑to source for robotics training data, labs may face a choice between licensing from them or trying to recreate years of capture operations. Either way, abundant high‑quality motion data should accelerate progress on dexterous, generalist robots. That pushes us closer to agents that can not only reason in text but also execute complex sequences in the real world, a core ingredient of many AGI visions.