Home→Startups & Technology→Genesis AI raises $500 million to scale its glove-trained ro…
Startups & Technology

Genesis AI raises $500 million to scale its glove-trained robots

A robot at Genesis AI’s Silicon Valley facility packed two small boxes, sealed a larger one and applied a label—then struggled to repeat the task. The mixed demonstration captures the challenge facing the startup as it raises $500 million to turn dexterous manipulation into reliable automation.

Genesis AI raises $500 million to scale its glove-trained robots

The Series A round is led by Khosla Ventures, Premji Invest and Eclipse. CEO and co-founder Zhou Xian says Genesis has “cracked dexterous manipulation,” but a second run of the packaging demonstration exposed errors: the robot dropped a box and struggled to peel a small label from printer tape. A Genesis executive called such mistakes “hiccup data” that can help improve the model.

Founded in 2024 by Xian and company president Théophile Gervet, who met as graduate students in a Carnegie Mellon University lab, Genesis is pursuing vertical integration. It develops its own generative AI model, produces much of its training data and is designing and building its own humanoid robot. The company announced a $105 million seed round in July 2025. “We wanted to ship the best robot, end-to-end product,” Xian said.

Genesis collects force data through sensor-equipped gloves and headset cameras worn by more than 100 workers around the world. Its robots wear the same type of gloves, and their hands are designed to match human proportions. Xian says the approach supports tasks such as sorting and wire manipulation. The company also uses simulation, drawing on an influential physics algorithm co-authored by adviser Minchen Li to generate millions of high-fidelity models of robot experience.

The company says it has built 30 to 40 Eno humanoids and is piloting the machine with a handful of undisclosed customers. Eno was not ready for a demonstration during TechCrunch’s visit, though Xian said the model performed better on it than on the robotic arms. Genesis has chosen a mass-manufacturing contractor, which is due to begin work in the U.S. next year; the company said the production design is finalized but did not disclose planned output. Xian identified hardware and supply chains as key bottlenecks, adding that the robot’s hands currently need replacement after a few weeks of regular work. In his view, better data and training infrastructure matter more than a new model architecture: architecture contributes “maybe 10%” of the final result.

Comments (0)

Leave a comment

No comments yet. Be the first!