Ex-Nvidia Researcher Raises $90M for World Model Startup

A prominent Toronto AI researcher just walked away from one of the most powerful positions in AI research to bet on a technology she says is about to have its breakthrough moment. Sanja Fidler, who led Nvidia's Spatial Intelligence Lab for eight years, has launched Veeda AI, a Toronto-based startup building world models for physical AI, raising more than US$90 million in seed financing backed by Radical Ventures and Khosla Ventures, according to The Logic's exclusive reporting on the launch.

Veeda describes itself as "building the next generation of multimodal foundation world models for physical AI," according to the company's own website, which remains in stealth mode and doesn't publicly identify its founding team. World models are AI systems typically trained on image and video data that simulate the consequences of actions in the real world, giving them a better grasp of physical mechanics than standard language models, technology with direct applications in robotics and what the industry calls embodied intelligence.

A Veeda job posting states the company is seeking to "push the frontier on generative world models," including advancing methods for creating image, video, and 3D content, while also recruiting staff to collect and prepare training data and manage the chip clusters needed to power that process.

Fidler joined Nvidia in May 2018 to lead a new Toronto research unit, later named the Spatial Intelligence Lab, and became a major figure in Canadian AI research along the way, serving as a founding member of the Vector Institute and holding a research chair position under Canada's federal AI strategy. She announced her departure from Nvidia in a LinkedIn post, framing the timing around a specific technical conviction: "World models is where the next breakthrough lies, and it is around the corner."

Fidler isn't leaving alone. Veeda's other co-founders are Huan Ling, who also worked at the Spatial Intelligence Lab, and Zan Gojcic, a Zurich-based former Nvidia research director. Fidler previously supervised Ling's doctorate at the University of Toronto, and the two were part of the team that created Nvidia's flagship world model, with the pair holding patents on joint work involving AI-generated images.

Veeda incorporated as Veeda Innovation in early June, and issued 60.6 million seed shares at US$1 each late last month, according to corporate filings reviewed by The Logic. On the same day, the company added two new directors, Radical partner Tomi Poutanen and Khosla partner Sven Strohband, changes typically associated with active fundraising. Khosla confirmed it has backed the firm; Veeda itself did not respond to questions about its fundraising or plans.

Veeda AI at a Glance

According to The Logic's analysis of PitchBook data, this financing would make Veeda one of Canada's three largest seed rounds ever recorded.

Fidler's move fits a genuinely notable pattern of prominent researchers leaving major tech companies to launch their own foundation model labs, often referred to as "neolabs," which require significant compute and typically raise mega-rounds from major investors almost immediately. Startups developing world models specifically have attracted substantial backing recently, including AMI Labs, started by former Meta AI chief Yann LeCun, and World Labs, founded by Radical scientific partner Fei-Fei Li, according to The Logic's reporting.

Toronto's world-model ecosystem specifically now includes at least one other major player: Waabi, led by fellow University of Toronto professor and Vector Institute co-founder Raquel Urtasun, which develops world models for autonomous vehicles and raised a US$750 million round in January co-led by Khosla with Radical participating. This connects to a broader pattern of Canadian AI investment we've tracked closely, including Anthropic's $10 million commitment to Canadian research institutions and the University of Toronto's own partnership with Cohere on sovereign AI infrastructure.

This launch is worth understanding as a genuine signal about where serious AI research talent and capital are flowing right now, away from purely conversational AI and toward systems that understand and simulate physical reality. For businesses in robotics, manufacturing, autonomous systems, or any physical operations context, world model startups like Veeda represent the technical foundation for a next wave of AI capable of genuinely understanding cause and effect in the real world, not just generating plausible text.

For Canadian businesses and investors specifically, Veeda's rapid, well-backed launch reinforces Toronto's position as a genuine hub for frontier AI talent, adding to a roster that already includes Cohere, Ideogram, and Waabi as homegrown foundation model companies commanding serious international investment.

What is Veeda AI?
Veeda AI is a Toronto-based startup founded by former Nvidia researcher Sanja Fidler, building multimodal foundation world models for physical AI applications including robotics.

How much funding has Veeda AI raised?
Veeda AI has raised more than US$90 million in seed financing, backed by Radical Ventures and Khosla Ventures, making it one of Canada's three largest seed rounds on record.

What is a world model in AI?
A world model is an AI system, typically trained on image and video data, that simulates the consequences of actions in the physical world, giving it a stronger grasp of real-world mechanics than standard language models, with applications in robotics and embodied AI.

Sanja Fidler left Nvidia after eight years leading its Spatial Intelligence Lab to launch Veeda AI, a Toronto-based startup building world models for physical AI, raising more than US$90 million in seed funding from Radical Ventures and Khosla Ventures. Co-founders Huan Ling and Zan Gojcic, both former Nvidia colleagues, join Fidler in building technology spanning language, image, and video with applications in robotics. The launch adds to a broader wave of star researchers leaving major tech companies to found their own foundation model labs, and strengthens Toronto's position as a hub for frontier AI research alongside Cohere and Waabi.