Startup Mecka AI has raised $60 million to train robots on direct human data.
The New York City-based vendor’s CEO, Josh Gao, revealed the investment in an interview with Fortune and a post on X.
The vendor, founded in 2025, said it completed the funding round in two parts: a $25 million Series A that closed in November, and then $35 million in follow-on investments. The company’s valuation has not yet been disclosed.
Explaining the company’s vision, Gao said its aim is to become the “data and deployment layer for physical AI.”
Mecka plans to take robotics training beyond teleoperation and simulation by using actual human data, sourced using body sensors iPhones and other devices, from a variety of different environments worldwide, including homes, kitchens, laboratories and others.
It’s an unusual path the vendor chose after months of research, visits to labs and canvassing industry opinion.
The data is then run through the vendor’s internal labs, which build and train computer vision models for human motion tracking and 3D construction. According to Gao, the concept is gathering momentum, and the vendor is claiming an annual run rate of $100 million based on contracts it has already signed, although it has not yet named any customers.
Mecka plans to use the investment to expand its team of data engineers and researchers, bolstering the current staff number of 40. Gao says the company will ultimately be more than simply a data aggregator and will have a key role to play in bringing “physical AI into the real world … where robots dependably handle real tasks in commercial environments.”
Mecka plans to target businesses, and Gao said on X: “We believe that taking a data-first approach makes us hypercompetitive in the deployment race.”
Venture capital firm, Framework Ventures, led the funding, with other participants including Menlo Ventures, Kindred Ventures, SV Angel and Ted Xiao, one of the founding members of Jeff Bezos’s AI startup, Project Prometheus.
