Veeda AI Raises $90M Seed: How Sanja Fidler’s World Model Startup Is Putting Toronto at the Center of Physical AI

Toronto’s AI ecosystem just recorded one of the largest seed rounds in Canadian history. Veeda AI, the world model startup founded by former Nvidia researcher Sanja Fidler alongside co-founders Huan Ling and Zan Gojcic, has raised over US$90 million in seed financing. The round is backed by Khosla Ventures and Radical Ventures, two of the most respected AI-focused investors in North America, and it places Canada firmly at the frontier of physical AI research.

What Is Veeda AI and What Does It Build

Veeda AI is building world models for physical AI. World models are a class of AI system that learns to simulate the physical world: how objects move, how forces interact, how environments change in response to actions taken within them. The goal is to create a training environment so rich and accurate that robots and autonomous systems can learn entirely within simulation, then transfer that learning to the real world.

The company’s core pitch rests on a specific technical insight. Robots cannot learn through trial and error in the real world because physical mistakes are dangerous, hardware does not scale like compute, and real-world experience cannot be parallelized. A robot can only be in one place at a time. A simulation can run millions of experiments simultaneously. Veeda AI’s platform creates training environments where robot intelligence can interact with a virtual world in countless configurations, learning from each interaction in ways that real-world deployment never allows.

The company is legally incorporated as Veeda Innovation and was founded in June 2026. It has offices in Toronto, Mountain View, Singapore, and Zurich, reflecting a global research ambition from the very start.

Who Is Sanja Fidler and Why This Matters

Sanja Fidler is one of the most accomplished AI researchers Canada has produced. Before founding Veeda AI, she served as the head of Nvidia’s Toronto AI Lab, one of the company’s most important research outposts outside the United States. She is also a professor at the University of Toronto, the institution that has produced a remarkable proportion of the world’s foundational AI research over the past two decades.

Her co-founders bring complementary depth. Huan Ling, serving as Chief Scientist, was a researcher at Nvidia with a focus on neural rendering and simulation. Zan Gojcic, the company’s CTO, previously led research at Nvidia’s Zurich lab, working on 3D scene understanding and robotic perception. Together the three bring exactly the technical background required to build what Veeda AI is attempting.

The founding team’s pedigree matters in a field where the technical difficulty of building reliable world models is genuinely high. Several well-funded startups have attempted versions of this problem. Veeda AI is approaching it with a team that spent years at the company that arguably understands AI hardware and software co-design better than anyone else in the world.

The $90M Seed Round: Investors, Structure, and What It Signals

The round, which closed in August 2026, was backed by Khosla Ventures and Radical Ventures. From Khosla, partner Sven Strohband joined Veeda AI’s board. From Radical Ventures, partner Tomi Poutanen took a board seat. Both are experienced investors in deep AI research companies.

Radical Ventures is the Toronto-based fund that has become the defining investor in Canada’s AI ecosystem, with a portfolio that includes some of the country’s most important AI companies. Khosla Ventures brings a global network and a long track record of backing technical founders working on hard problems in robotics and AI. Their joint investment in Veeda AI is a strong signal about how seriously the global investor community is taking the world model opportunity.

A seed round of over $90 million is exceptional by any standard, but especially in Canada. It ranks among the largest seed financings the country has ever seen and reflects investor conviction that Veeda AI is attacking a problem of fundamental importance to the future of robotics and automation, according to BetaKit.

The Technical Bet: Why World Models for Robots

World models have become one of the most discussed topics in AI research. The intuition behind them is straightforward: if a system can accurately predict the consequences of its own actions within a model of the world, it can plan better, learn faster, and generalize more reliably than systems trained on static datasets.

For robotics, the stakes are particularly high. The robots being deployed in factories, warehouses, construction sites, and eventually homes need to handle an enormous range of situations that no training dataset can fully anticipate. A world model that accurately simulates physical reality gives robotic systems a way to encounter and learn from novel situations before they happen in the real world.

Veeda AI’s multimodal approach processes complex sensor data alongside real-time action inputs, creating environments that are deeply reactive to the decisions an agent makes within them. The company is not building a game engine or a general simulation platform. It is building a specific class of AI-native environment designed to produce embodied intelligence, as reported by SiliconAngle.

Why Toronto and What This Means for Canada’s AI Ecosystem

Canada’s AI ecosystem has been building toward a moment like this for years. The country’s strength in foundational AI research, anchored by the University of Toronto, the Vector Institute, Mila in Montreal, and the Amii institute in Edmonton, has created a talent base that global companies increasingly cannot ignore. Nvidia recognized this when it built a major research presence in Toronto. Now that talent is spinning out into startups of global ambition.

Veeda AI is part of a broader wave. Canadian robotics and AI startups have been attracting serious capital across the country. TetraGen Robotics in Winnipeg recently closed a C$1.8M seed to apply AI to industrial welding automation, while Terminal raised $20 million in Toronto to accelerate commercial fleet AI infrastructure. Veeda AI operates at a different scale, but the directional story is the same: Canadian founders are building products, not just publishing papers.

For Toronto specifically, Veeda AI adds to a cluster of world-class AI companies that are choosing the city not for cost reasons but because the talent is genuinely among the best in the world. That is the strongest possible signal for the health of Canada’s technology ecosystem heading into the next decade.