companies

Genesis AI

Robotics startup building a universal foundation model for physical AI, trained largely on synthetic data from its own ultra-fast, open-source physics simulator.

Genesis AI was founded in December 2024 by Zhou Xian, a Carnegie Mellon robotics PhD, and Theophile Gervet, a former Mistral AI research scientist, to build what they call a universal robotics foundation model - a single AI system capable of controlling many different robot bodies and tasks rather than one model per robot. The company emerged from stealth in July 2025 with a $105 million seed round co-led by Eclipse Ventures and Khosla Ventures, with additional backing from Bpifrance, HSG and individual investors including Eric Schmidt and Xavier Niel. The company's approach centers on generating large volumes of synthetic training data through a proprietary physics simulation engine that its founders say runs physics far faster than real time, letting robot control policies be trained in simulation and transferred to physical hardware. That simulator builds on Genesis, an open-source physics engine for robotics and embodied-AI research that Zhou Xian helped create during his academic work, spanning an 18-university collaboration before the company's founding. Genesis AI is based in Palo Alto, California, with a research team drawn from Mistral AI, Nvidia, Google DeepMind and several robotics labs, and has since announced GENE, its robot foundation model, alongside custom dexterous-hand hardware built with manufacturing partner Wuji Tech.

Founded

2024

Headquarters

Palo Alto, United States

Sector

robotics