Mecka AI raises $60M for human-motion robot training

Mecka AI announced a $60 million Series B, TechCrunch reports. Founded in 2024, the startup is building a supply of human-motion data for teams training humanoid and other robots: people demonstrate physical tasks, and Mecka collects and analyzes their movements so machines have examples of how humans do the work.

Sequoia led the financing, with Nvidia and Microsoft’s M12 among the participants. TechCrunch had previously reported that Mecka was nearing a round at a $500 million valuation. That earlier figure described reported talks, not an announced price for the completed financing.

The raw input begins with ordinary work. Contributors wear body sensors and use smartphones while completing activities ranging from brewing coffee to repairing cars; Mecka records and analyzes those demonstrations for use in robot training.

The approach adapts a familiar AI-data model to physical movement. TechCrunch says Mecka intends to play a role in robotics similar to the one Scale AI, Mercor and Surge play for language models by supplying human-generated examples that systems can learn from. Mecka pays the people producing those recordings.

Competition is already forming around that supply chain. TechCrunch identifies XDOF as another collector of real-world robot-training data and previously reported that it was discussing a Series B at a $1.2 billion valuation. Human-data platforms Scale AI and Micro1 are also expanding from language-model work into robotics, putting Mecka alongside both specialists and established data providers.