Genesis AI, a startup focused on creating a foundational model for robots, has recently emerged from stealth mode with an impressive $105 million seed funding round, co-led by Eclipse Ventures and Khosla Ventures. The company was founded last December by Zhou Xian, who holds a PhD in robotics from Carnegie Mellon University, and Théophile Gervet, a former research scientist at the French AI lab Mistral. Their goal is to build a general-purpose model that enables robots to automate various repetitive tasks, including lab work and housekeeping. Unlike large language models that are trained on extensive text datasets, robotics AI models require training on data from the physical world.
However, gathering this real-world data can be both expensive and time-consuming. To tackle this issue, Genesis is utilizing synthetic data generated by a proprietary physics engine, which they claim can accurately model the physical environment. The synthetic data engine that Genesis employs emerged from an academic collaboration led by Xian, involving researchers from 18 universities. This foundation has allowed Genesis to assemble a team of over 20 researchers specializing in robotics, machine learning, and graphics.
According to Genesis, their unique simulation engine enables them to develop models more rapidly than competitors who depend on Nvidia’s software. Other companies in the field include Physical Intelligence, which recently raised a $400 million funding round, and Skild AI, which was valued at $4 billion earlier this year. Kanu Gulati, a partner at Khosla Ventures, expressed optimism about Genesis, stating, “Of all the teams we have seen, we like Genesis’ approach for going after robotics foundation models.”
Currently, Genesis is advancing its synthetic data initiatives and foundational model development across its offices in Silicon Valley and Paris, with plans to release its model to the robotics community by the end of the year.