Insider Brief
- Veeda AI has raised more than $90 million in seed funding to build world models that create simulated environments where robots can learn through trial and error, The Logic reported.
- Founded by former Nvidia AI researchers Sanja Fidler, Zan Gojcic and Huan Ling, the startup is backed by Radical Ventures and Khosla Ventures.
- Veeda plans to use multimodal world models trained on sensor and physical-world data to create scalable virtual environments for training and evaluating robots without relying solely on costly, slow and potentially unsafe real-world learning.
Veeda AI has raised more than $90 million in seed funding to build world models designed to give robots virtual environments where they can learn through trial and error, The Logic has reported.
The Canadian startup, founded by former Nvidia AI researchers Sanja Fidler, Zan Gojcic and Huan Ling, is backed by Radical Ventures and Khosla Ventures, The Logic reported, citing people familiar with the matter and corporate filings.
Fidler, Veeda’s CEO, said in a LinkedIn post announcing the company that robots will ultimately need to learn through interaction rather than relying primarily on imitation, where machines are trained to reproduce examples of human or robot behavior.
“For Physical AI, the central challenge is making this kind of interactive learning possible at scale,” Fidler wrote. “The real world is simply not a practical training ground for robots to learn through trial and error. It is unsafe, expensive, and time-consuming. To scale interactive learning, robots will need to learn in simulated reality.”
Veeda plans to move more of that learning into simulated environments, allowing robot AI systems to repeatedly perform tasks, fail and adjust without risking equipment or requiring fleets of physical machines.
“World models are the foundational technology that can make this possible,” the company notes. “These generative models learn from massive amounts of sensor and physical-world data to reach the quality, diversity, and physical realism of the simulated reality that robots require.”
The company compares its goal to building a Matrix-like virtual world for physical AI, according to its website. Within those environments, a robot’s AI could operate through a virtual version of its body and encounter many variations of objects, tasks and conditions.
“At Veeda, our sole mission is to build simulated reality for Physical AI,” Fidler added. “Our conviction is that this will become the critical infrastructure layer for all areas of robotics.”
