Sanja Fidler, the computer vision researcher who spent eight years leading Nvidia's Toronto AI lab, has raised more than $90 million in seed financing for her new startup Veeda AI, according to a report by The Logic on August 19, 2026, citing corporate filings and people familiar with the matter. The round, backed by Khosla Ventures and Radical Ventures, ranks among the largest seed financings ever raised by a Canadian company — and it lands less than three months after the company was incorporated.
The bet behind the money is specific: Veeda is building multimodal foundation world models that simulate physical reality, creating scalable simulated environments where robots and other embodied agents can learn through interaction. For readers tracking the AI industry's shift from chatbots to physical intelligence, it is one of the clearest signals yet that world models have become the hottest frontier in AI funding. For more context on this story, see our ongoing more AI stories.
From Nvidia's Spatial Intelligence Lab to a New Company
Fidler joined Nvidia in May 2018 to establish the chipmaker's Toronto research unit, which later became the Spatial Intelligence Lab, and rose to vice president of AI research while maintaining her faculty position at the University of Toronto, where she has taught computer science since 2014. She was also a founding member of the Vector Institute, Toronto's prominent AI research and commercialization hub.
Her co-founders are Huan Ling, who also worked at the Spatial Intelligence Lab and completed his doctorate under Fidler's supervision, and Zan Gojcic, a Zurich-based former Nvidia research director. The three incorporated the company, legally named Veeda Innovation, in early June 2026, according to the filings reviewed by The Logic.
The paper trail of the round is unusually concrete for a seed deal. In late July, Veeda issued 60.6 million seed shares priced at one US dollar each, and on the same day added two new directors: Radical partner Tomi Poutanen and Khosla partner Sven Strohband. Board seats for both investors alongside a priced share issuance are the standard evidence of a closed financing. Khosla Ventures confirmed to The Logic that it has backed the company.
Why World Models, and Why Now
Veeda's pitch is a technical wager about how robots should learn. Training embodied agents through trial and error in the real world is unsafe, hardware does not scale the way compute does, and real-world experience cannot be parallelized across thousands of machines. The alternative is training inside high-fidelity simulated environments generated by world models — an approach the company frames, only half-jokingly, as building "the Matrix" for physical AI.
"World models is where the next breakthrough lies, and it is around the corner," Fidler wrote in the post announcing her departure from Nvidia.
The founders have run this playbook before. At Nvidia's Spatial Intelligence Lab, Fidler's team developed software simulating how autonomous vehicles and humanoid agents interact with the physical world, and both Fidler and Ling were part of the group behind Nvidia's flagship world model for developers building physical AI systems.
A Crowded, Expensive Race
The investor pairing is one of the sharpest signals in the round. Khosla Ventures and Radical Ventures have repeatedly co-backed research-heavy AI teams — earlier in August 2026, the two firms co-led the founding round for Discovery Loop, the AI-for-science startup built by former Google chief scientist Jeff Dean.
Both firms also carry direct exposure to the world model category. Radical is an investor in World Labs, the spatial intelligence company founded by its scientific partner Fei-Fei Li, and its portfolio includes Decart, which builds real-time world models. Khosla co-led autonomous trucking company Waabi's $750 million round in January 2026, with Radical participating; Waabi builds its own world models for self-driving.
The broader funding landscape is even more competitive. Yann LeCun's AMI Labs raised $1.03 billion in March 2026 to pursue world models, and World Labs raised $230 million in a round co-led by Radical Ventures. Against that cohort, Veeda's $90 million seed is modest in absolute terms, but it lands near the top of Canada's all-time seed rankings, according to PitchBook data cited by The Logic.
A Global Footprint from Day One
Veeda lists offices in Toronto, Mountain View, Singapore, and Zurich. A Swiss subsidiary registered in July names Gojcic as local managing director and Fidler as management chair. A job posting says the company is hiring to "push the frontier on generative world models," including staff to prepare training data and manage chip clusters — a reminder that data pipelines and compute, not just research talent, dominate a foundation lab's early budget.
For the robotics industry, the arrival of another well-funded world model startup deepens a trend that has been building all year: the major bottleneck for physical AI is no longer thought to be robot hardware, but the scarcity of realistic training environments. If Veeda and its rivals succeed, the next generation of robots could spend its formative "childhood" entirely inside simulation, only meeting the real world once it has already learned to act.
Whether simulation-first training can truly close the gap to real-world dexterity remains an open research question — one that Fidler, with a nine-figure war chest and two of Silicon Valley's most research-friendly investors behind her, is now positioned to answer.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
