Genesis AI rises $105M to build a universal robotics model and automate physical labor through advanced simulation and real-world data.

Genesis AI, a full-stack robotics startup and international physical AI research center, came out of stealth today to enable limitless physical work. The company is investing $105 million to create a universal robotics core model and a horizontal robotics system with the assistance of Eclipse and Khosla Ventures. The initiative also includes backing from prominent figures like Xavier Niel and Eric Schmidt, along with support from Bpifrance and HSG.

Over 95% of physical work is still not mechanized because of the shortcomings of the automation solutions available today, despite the fact that it is predicted to contribute $30–40 trillion to the global GDP. Industrial arms and other modern robotic systems rely on software stacks that are stiff, brittle, and overfitted.

These systems are hard to scale, costly to implement, and have a narrow scope. Genesis employs a full-stack, data-driven approach to physical AI, combining high-fidelity physics simulations, multimodal generative models, and large-scale real-world robot data into a unified, modular data engine. Its fully in-house-developed simulation stack will produce large amounts of rich synthetic data in addition to a more effective and scalable real-world data collection method.

By bridging traditionally divided domains, their dual engine of synthetic and real data gathers the most extensive, varied, and superior quality data for RFM training. Mistral AI, Nvidia, Google, CMU, MIT, Stanford, Columbia, and UMD are among the top academic and industry technical talent that founded Genesis. With extensive experience in physics simulation, graphics, robotics, and large-scale AI model training and deployment, Genesis is well-positioned to quickly realize its vision through a unique approach to physical AI.