Mecka AI Raises $60M for Robot Training Data
Mecka AI has raised $60M to scale human-motion data collection for training robots.
Robotics foundation models need enormous quantities of real-world action data, creating a new infrastructure market around how that data is collected.
What happened
Mecka AI raised a $60 million Series B to scale its platform for gathering human-motion data used to train robots.
The company pays people to record everyday physical tasks using sensors and smartphones, producing datasets that help robotic systems learn how humans manipulate objects and move through the physical world.
Why it matters
Large language models benefited from huge text datasets that already existed online. Robotics does not have the same advantage.
High-quality physical-action data is expensive to collect because it must capture motion, context and interactions with real objects. That makes data collection itself a potential bottleneck for embodied AI.
The bigger picture
A specialised data layer may emerge around robotics just as annotation companies became important to the development of large language models.
Mecka is effectively betting that robot developers will need external infrastructure for collecting and organising physical-world training data rather than building every dataset themselves.