Veeda AI Raises $90M Seed to Build World Models

Veeda AI, a startup founded by former Nvidia computer vision researcher Sanja Fidler and her colleagues, has secured more than $90 million in seed financing. Legally incorporated as Veeda Innovation in early June 2026, the Toronto-headquartered company issued 60.6 million seed shares priced at $1 each in late July. The massive round was backed by Khosla Ventures and Radical Ventures, with Khosla partner Sven Strohband and Radical partner Tomi Poutanen joining Veeda's board. This transaction represents one of the largest seed rounds in Canadian history.

Fidler, who spent eight years at Nvidia and rose to vice president of AI research, co-founded the company alongside Huan Ling and Zan Gojcic. Ling previously worked with Fidler at Nvidia's Spatial Intelligence Lab, while Gojcic served as a Zurich-based research director for the chipmaker. Veeda is leveraging this expertise to build multimodal foundation world models. These models simulate physical reality to create highly scalable virtual environments where embodied AI agents, such as humanoid robots and autonomous vehicles, can learn through interaction.

For AI practitioners and robotics developers, Veeda's approach addresses a fundamental bottleneck in physical AI. Training robots in the real world is notoriously slow, expensive, and unsafe, as physical hardware cannot scale at the pace of compute. By generating high-fidelity simulated environments, Veeda aims to let developers parallelize training and test agents safely. Fidler has noted that "world models is where the next breakthrough lies," positioning these simulations as a virtual training ground to bypass physical constraints.

This capital-intensive bet places Veeda alongside other heavily funded players in the spatial intelligence sector. In early 2026, Yann LeCun's AMI Labs raised $1.03 billion, Fei-Fei Li's World Labs secured $230 million, and autonomous trucking firm Waabi closed a $750 million round. Veeda, which operates offices in Toronto, Mountain View, Singapore, and Zurich, is already hiring staff to manage its chip clusters and prepare training data to push the boundaries of generative world models.