TechCrunch reports that the round was led by Sequoia, with Nvidia and Microsoft’s venture fund M12 among the participants. It also reports that Mecka, founded in 2024, collects motion data to train humanoid and other robots. These facts place the company upstream of robot behavior: its product is recorded examples, not a robot or a newly announced model.
The mechanism is human demonstrations gathered through wearable and phone-based capture, intended as training data for robots. The report describes people recording everyday tasks, including making coffee and repairing cars, with body sensors and smartphones. Why now: a $60 million Series B signals investor backing for this data-collection approach, although the announcement does not establish that it has become a standard or that the capital will produce a particular technical outcome.
The first implication is that robotics teams may treat access to varied, real-world human demonstrations as a distinct input to development, alongside models and hardware. The evidence boundary is that the supplied report gives no dataset size, coverage, quality metrics, customer list, deployment count, or breakdown of how the funding will be used. What to watch: whether Mecka discloses measurable data coverage, customer adoption, or robot-training results that connect collected demonstrations to performance in real tasks. This starts a distinct analytical record because no prior New Runtime record on Mecka is supplied; the related Gemini Robotics 2 post concerns a model release, not this company's financing and data-collection operation.