Genesis AI was founded in December 2024 by Zhou Xian, a Carnegie Mellon robotics PhD, and Théophile Gervet, a former research scientist at Mistral who also holds an AI PhD from Carnegie Mellon. The company set out to solve one of robotics' hardest bottlenecks: the scarcity of high-quality, diverse training data needed to make robots generalize across tasks and environments. Rather than relying solely on expensive real-world teleoperation, Genesis built a proprietary physics engine capable of running thousands of accurate simulated trials in minutes, producing the synthetic data that powers its foundation model, GENE.
The GENE model family is designed to be embodiment-agnostic and horizontal, meaning the same underlying intelligence can drive different robot hardware on different tasks. Genesis has emphasized dexterity, demonstrating robotic hands shaped like human hands rather than the simpler two-finger grippers common in the industry, narrowing the gap between simulation and messy real-world conditions. The team describes its approach as 'full-stack,' spanning data generation, model training, and deployment, so customers can adopt a single platform rather than stitching together point solutions.
In July 2025 the company emerged from stealth with a $105 million seed round co-led by Eclipse Ventures and Khosla Ventures, with participation from Bpifrance and HSG and angel backing from notable technologists including Eric Schmidt, Xavier Niel, MIT roboticist Daniela Rus, and computer-vision researcher Vladlen Koltun. The capital funds expansion of the company's research and engineering teams and the scaling of its synthetic-data and model pipeline.
Genesis positions itself in the increasingly competitive 'robot foundation model' category alongside firms such as Physical Intelligence and Skild AI, but differentiates through its physics-engine-first data strategy and its ambition to be a horizontal platform for physical AI. The long-term vision is to 'unlock unlimited physical labor' by making robots cheap, flexible, and robust enough to operate broadly across industry and everyday settings.