Veeda AI

Veeda AI is building multimodal foundation world models — AI that generates high-fidelity, scalable simulated environments where robots and other embodied agents can learn by interaction. The pitch is infrastructure: a training layer for the whole Physical AI sector, so machines can rehearse in simulation rather than risk the real world.

Physical AI and robotics are hitting a wall on real-world data, where training is slow, costly and physically risky. World models that simulate reality offer a way past that bottleneck, and Veeda AI is positioning itself as the layer everyone else builds on.

Founded in June 2026 in Toronto, the company grew out of work on world models and 3D data understanding at Nvidia's Spatial Intelligence Lab, where the founding team spent years before striking out on their own.

The team is led by Dr. Sanja Fidler, who ran Nvidia's Toronto AI research lab for eight years and teaches at the University of Toronto, joined by former Nvidia colleagues Zan Gojcic (CTO) and Huan Ling (Chief Scientist). That's a founding group with a track record specifically in world models and spatial intelligence — exactly the problem they're now tackling.