Genesis AI raises $500M Series A for its Eno humanoid
Cofounder Zhou Xian says the company has cracked dexterous manipulation, and that no further scientific leap is needed.
Genesis AI has raised a $500M Series A led by Khosla Ventures, Premji Invest and Eclipse, according to Lukas Ziegler. That is up from a $105M seed in July 2025. The company was founded in 2024 by @zhou_xian_ and @theo_gervet, who met as grad students in a CMU lab.
I don’t think we expect or we need another fundamental scientific leap. We have all the recipes. It’s just about execution.
That is a strong claim, and it will be tested on the factory floor. $500M buys a lot of execution.
The headline claim is bigger than the number. Ziegler reports that Xian says Genesis has "cracked dexterous manipulation."
The vertical integration bet
Genesis generates its own training data, builds its own model and designs the humanoid alongside both. Force data comes from sensor-enabled gloves worn, along with headset cameras, by more than 100 real workers around the world. The robots wear the same gloves, and their hands copy human proportions. Millions of high-fidelity simulated experiences come from a physics algorithm co-authored by adviser Minchen Li.
The flagship humanoid is called Eno. More than 30 have been built so far, and they are piloting with a handful of undisclosed customers. A US mass-manufacturing contractor has been chosen to start next year, and the production design is finalized. Ziegler says Eno looks nothing like a metal skeleton, describing it as closer to an iMac on wheels.
Architecture is 10%, says Xian
The part Ziegler finds most interesting is Xian's view of models. Xian says architecture, whether VLA (a model that turns camera images and text into robot actions) or world-action, contributes maybe 10% of the final result. What matters instead is the training infrastructure, the data infrastructure and data quality.
That is a direct shot at a debate the field has been having all month, including the frontier-models-versus-specialists argument that got personal at CoRL, covered earlier. Xian's position is that the recipes already exist and the remaining work is execution.
Shown versus claimed
Nothing here has been shown publicly. The customers are undisclosed, the pilots have not been demonstrated outside the company, and the claim that dexterous manipulation is cracked is so far one cofounder's claim. Ziegler is blunt about it himself, writing that it is a strong claim and that it will be tested on the factory floor. The mass-manufacturing contract starts next year, so the first outside check on any of this is some way off.
Sound like you were at the demo
One email a day. Under three minutes. The demos, the deployments and the part the press release left out.
