
Genesis AI Raises $500 Million to Train Humanoid Hands With Sensor Gloves
Genesis AI closes a $500M Series A to mass-produce its Eno humanoid in the U.S., training dexterous hands with sensor gloves worn by over 100 real workers.
- By
- Tom Whitfield
- Filed
- Channel
- Startups & Funding
- Read
- 4 min read
Genesis AI has raised a $500 million Series A round, led by Khosla Ventures, Premji Invest and Eclipse, to build humanoid robots whose defining skill is what their hands can do. The round follows a $105 million seed announced in July 2025 and arrives as the company prepares U.S.-based mass production of its flagship robot, Eno, starting next year through a chosen manufacturing contractor.
CEO and co-founder Zhou Xian claims the company has "cracked dexterous manipulation." At Genesis AI's Silicon Valley facility, its physical AI model ran a pair of off-the-shelf robotic arms through a packing workflow. The arms picked up two small cardboard boxes, scanned them, placed them inside a larger box, folded it, taped it shut, and applied a label peeled from printer tape.
That is precisely the kind of complex sequence today's robots struggle to complete. It is also, on the evidence of the demo itself, not yet solved. During a second run, the robot fumbled a box and had difficulty peeling the small adhesive label from the tape. A Genesis executive framed such failures as "hiccup data" — edge cases that can be used to improve the model.
Why dexterity is the differentiator
The gap between a lab demo and reliable automation is the gap Genesis, and the wider humanoid ecosystem, must close to avoid another robot winter. Genesis positions itself apart from competitors such as Figure and Agility through a narrow focus on hand skills and the data needed to teach them.
The mechanism is a sensor-enabled glove. More than 100 real workers around the world wear the gloves, along with headset cameras, and their activities feed a training dataset for the company's generative AI model. Force data gathered through the gloves appears to be the unlock for effective manipulation. The robots themselves wear the same sensor-equipped gloves, and their hands mimic the proportions of human hands — a fidelity Xian credits for capabilities in sorting and wire manipulation.
The second major investment is simulation. Adviser Minchen Li co-authored an influential physics algorithm that Genesis uses to create millions of high-fidelity models of the robot's experience.
What is the state of the robot itself?
Xian said the company's model performs better when operating Eno, the company's humanoid robot. Eno was not ready to demonstrate its abilities.
Visually, Eno is less a Terminator-styled metal skeleton and more, in Xian's description, "an iMac on wheels" — a design language that reflects "the feeling" the company wants to deliver to customers. Genesis has built 30 to 40 Enos so far and declined to say how many it will make in the year ahead. The company did confirm that the mass-produced version's design is finalized, that a mass-manufacturing contractor has been chosen, and that work starts next year in the U.S. A handful of undisclosed customers are piloting the machine.
Where does the bottleneck sit?
The founding story is academic: Xian and Genesis president Théophile Gervet met as graduate students in a Carnegie Mellon University lab and co-founded the company in 2024. Their strategy is vertical integration — generating the bulk of the company's own training data, developing its own generative AI model, and doing both alongside the design and construction of the humanoid robot itself.
"We wanted to ship the best robot, end-to-end product," Xian said.
The key bottleneck, in his view, is hardware capability, both in mechanical design and in the supply chains required for industrial-grade machines. Genesis's dexterous hands currently last only a few weeks of regular work before needing replacement.
On the software side, Xian dismisses the fine distinctions between architectures such as Vision-Language-Action models and World-Action models. The software is ultimately transformers and attention, he argues, so "what really matters is the whole infra — the training infrastructure, the data infrastructure, and the quality of the data." He estimates that "the model architecture contributes to maybe 10% of the final [result]."
On whether another breakthrough is needed, he is direct. "I don't think we expect or we need another fundamental scientific leap," he said. "No, we have all the recipes. It's just about execution."
The $500 million buys Genesis the runway to test that claim: whether durable hands, worker-gathered data and a U.S. manufacturing line can turn a taped-shut cardboard box into repeatable, revenue-generating automation.
Source: techcrunch.com
More from Tom Whitfield
Show full bio
Staff writer covering companies, deals and strategy in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,032 articles
Related articles
genesis-ai-raises-500m-series-a-at-3-5b-valuation-for-factory-robots
Genesis AI Raises $500M Series A at $3.5B Valuation for Factory Robots
genesis-ai-lands-500m-series-a-at-3-5b-valuation-for-factory-robots
Genesis AI Lands $500M Series A at $3.5B Valuation for Factory Robots
unconventional-ai-tapes-out-tsmc-chip-pursuing-1-000x-power-cut-2691ed31
Unconventional AI Tapes Out TSMC Chip Pursuing 1,000x Power Cut


