Genesis AI has raised a massive $105 million seed round to build a foundational AI model for robots. The funding is co-led by Eclipse Ventures and Khosla Ventures.
The startup launched out of stealth after forming in December. Founders Zhou Xian, a Carnegie Mellon robotics PhD, and Théophile Gervet, a former French AI lab Mistral research scientist, want their model to automate tons of repetitive tasks—from lab work to housekeeping.
Robotics AI needs real-world physical data, which is expensive to gather. Genesis AI uses synthetic data generated by its own physics engine. The engine, born from an academic project with 18 universities, simulates the physical world with high accuracy. Many team members came straight from that project. The company now has 20+ researchers across robotics, machine learning, and graphics. 
They say their simulator is faster than competitors relying on Nvidia’s tools. Other rivals in general robot AI models include Physical Intelligence, which raised $400 million, and Skild AI, valued at $4 billion.
Khosla Ventures partner Kanu Gulati told TechCrunch:
“It’s a big unknown: Will anybody have a large robotics foundation model that will generalize across tasks? That’s a bet we want to go after.”
“Of all the teams we have seen, we like [Genesis’s] approach for going after robotics foundation models.”
Genesis AI is building its model in Silicon Valley and Paris. The startup plans to release it to the robotics community by the end of 2024.
