Waabi
Last reviewed
Jun 9, 2026
Sources
18 citations
Review status
Source-backed
Revision
v2 · 2,109 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 9, 2026
Sources
18 citations
Review status
Source-backed
Revision
v2 · 2,109 words
Add missing citations, update stale details, or suggest a clearer explanation.
Waabi is a Canadian autonomous-driving company based in Toronto that develops self-driving software for long-haul trucks and, more recently, robotaxis. It was founded in 2021 by the AI researcher Raquel Urtasun, a former chief scientist of Uber's self-driving division and a professor at the University of Toronto. Waabi is built around an "AI-first" design: rather than the modular, hand-engineered stacks used by most autonomous vehicle programs, the company trains a single end-to-end model called the Waabi Driver almost entirely inside a high-fidelity neural simulator called Waabi World. The company describes its work as "physical AI," and by early 2026 it had raised roughly US$1.28 billion across three financing rounds and partnered with Uber, Volvo, and NVIDIA. In January 2026 it announced plans to extend its system from self-driving trucks into ride-hailing.
| Key facts | |
|---|---|
| Founded | January 2021 |
| Founder and CEO | Raquel Urtasun |
| Headquarters | Toronto, Canada |
| Focus | Autonomous trucking; robotaxis (from 2026) |
| Core products | Waabi Driver (AI model), Waabi World (simulator) |
| Total funding | About US$1.28 billion (through Series C, January 2026) |
| Key partners | Uber, Volvo Autonomous Solutions, NVIDIA |
Urtasun incorporated Waabi Innovation Inc. in early 2021 after leaving Uber. Born in Spain and based in Canada, she earned a doctorate in computer science from the Swiss Federal Institute of Technology in Lausanne (EPFL) in 2006, did postdoctoral work at MIT and the University of California, Berkeley, and taught at the Toyota Technological Institute at Chicago before joining the University of Toronto as a professor in 2014. In 2017 she co-founded the Vector Institute, a Toronto AI research hub, alongside Geoffrey Hinton. That same year she joined Uber to lead the Toronto arm of its Advanced Technologies Group (ATG), the ride-hailing company's self-driving unit, serving as chief scientist.
When Urtasun left to start Waabi, she argued that the dominant approach to self-driving had stalled. Conventional stacks, which she calls "AV 1.0," string together separate hand-tuned modules for perception, prediction, and planning, and require enormous fleets, capital, and road miles to improve. Waabi's thesis is that modern deep learning and generative AI make it possible to replace that pipeline with a single learned model that generalizes from far less real-world driving. The name "Waabi" is drawn from an Ojibwe word meaning "she has vision" or the act of seeing.
Waabi's system pairs two generative AI components that the company frames as a "teacher" and a "student."
The Waabi Driver is the student: a single, end-to-end model that takes in sensor data and outputs driving decisions. Unlike a modular stack, it is one network rather than a chain of separate components. Unlike the "black box" end-to-end systems that some rivals pursue, Waabi says the Driver is interpretable and verifiable, meaning its internal reasoning can be traced and checked. Urtasun describes this as giving the model a form of cause-and-effect, deliberative reasoning rather than pure pattern matching, and the company contrasts it with camera-only approaches by noting that the Driver fuses lidar, camera, and radar for redundancy. In effect the Driver behaves as a single embodied AI agent controlling the vehicle.
Waabi World is the teacher: a closed-loop neural simulator that the company calls the most scalable and highest-fidelity such system built for driving. It reconstructs real environments and performs near real-time sensor simulation, recreating what the Driver would "see" through its sensors. Because the Driver is trained and validated mainly in simulation, Waabi says it can expose the model to billions of scenarios in the cloud, including rare and dangerous situations, without the cost and risk of logging those miles on public roads. The company has gone further than most by claiming it can mathematically demonstrate that performance in Waabi World corresponds to performance in the real world, a claim central to its safety and validation argument.
This simulation-first method underpins how Waabi addresses safety. Rather than retrofitting ordinary vehicles, Waabi partners with truck makers to build platforms with redundant steering, braking, and power so that the system can always bring a vehicle to a safe stop if a component fails. Urtasun frames the broader case in terms of road deaths, arguing that human error is responsible for the large majority of crashes and that capable, well-validated machines can reduce that toll. The company runs its software on NVIDIA hardware: at NVIDIA's GTC conference in March 2024 Waabi said it would use the NVIDIA DRIVE Thor in-vehicle computer and build on NVIDIA's DRIVE operating system.
Waabi launched its first commercial autonomous trucking operations in the fall of 2023, hauling freight on the corridor between Dallas and Houston using retrofitted Peterbilt trucks with safety drivers aboard. Its central goal, repeated when it raised its Series B in 2024, was to remove the human driver, a step known in the industry as "driver-out," and begin fully driverless commercial trucking in Texas in 2025.
That 2025 driver-out launch did not happen on the original timeline, and it is important to state the milestone accurately. Waabi reported that it was on track to reach "driverless software readiness" around the end of 2025, which it framed as evidence that its verifiable end-to-end software was ready to operate without a human. But the company chose to keep a safety operator in the cab until the fully redundant truck hardware it is building with Volvo was validated, a decision Urtasun framed as prioritizing safety over speed. As of its January 2026 funding announcement, Waabi's trucks were still running commercial pilots in Texas with a human in the front seat, and TechCrunch reported that the fully driverless highway launch had slipped to "sometime in the next few quarters."
The Volvo collaboration is the vehicle for that launch. Waabi and Volvo Autonomous Solutions announced a long-term partnership in February 2025, and on October 28, 2025, at TechCrunch Disrupt, they unveiled a purpose-built autonomous truck based on the Volvo VNL Autonomous, produced at Volvo's plant in Virginia with redundant backup systems designed specifically to allow the human driver to be removed. Urtasun said the truck was "built from the ground up for the redundancies so that you can remove the human driver." Rival Aurora Innovation began limited driverless freight runs in Texas in 2025 using the same Volvo-built platform but at times kept a human observer in the cab; Waabi has said it aims to launch with no safety driver or observer at all.
Waabi has raised money in three principal rounds, each among the larger venture financings in Canadian history.
| Round | Date | Amount (USD) | Lead investors | Selected participants |
|---|---|---|---|---|
| Series A | June 2021 | $83.5 million | Khosla Ventures | Uber, Radical Ventures, 8VC, OMERS Ventures, Aurora Innovation, Geoffrey Hinton, Fei-Fei Li, Pieter Abbeel, Sanja Fidler |
| Series B | June 18, 2024 | $200 million | Uber, Khosla Ventures | NVIDIA, Volvo Group Venture Capital, Porsche SE, Scania Invest, Ingka Investments, HarbourVest, G2 Venture Partners |
| Series C | January 28, 2026 | $750 million (plus up to $250 million from Uber) | Khosla Ventures, G2 Venture Partners | Uber, NVentures (NVIDIA), Volvo, Porsche SE, BlackRock, Radical Ventures, BDC Capital, Export Development Canada |
The 2021 Series A of US$83.5 million, led by Khosla Ventures, was one of the largest first rounds ever raised by a Canadian startup and drew backing from AI figures including Geoffrey Hinton and Fei-Fei Li. In June 2024 Waabi raised a US$200 million Series B co-led by Uber and Khosla Ventures, bringing total investment past US$280 million and adding strategic backers such as NVIDIA and Volvo.
The largest round came on January 28, 2026, when Waabi announced a US$750 million Series C co-led by Khosla Ventures and G2 Venture Partners, plus a separate, milestone-based commitment of up to US$250 million from Uber tied to robotaxi deployment. Waabi presented the combined package as US$1 billion in new funding and described it as the largest fundraise in Canadian history; the round brought cumulative investment to roughly US$1.28 billion. The company did not disclose a valuation, though The Globe and Mail reported that the financing valued Waabi at about US$3 billion, making it one of Canada's newest autonomous-vehicle "unicorns."
The January 2026 round marked Waabi's move beyond freight. Alongside the Series C, the company announced a partnership to deploy robotaxis powered by the Waabi Driver exclusively on the Uber platform, with Uber committing milestone-based capital to support development and a target of deploying at least 25,000 Waabi Driver-powered robotaxis over time. Waabi described it as likely the largest self-driving deal in industry history, though the companies did not give a firm timeline and said specific robotaxi schedules and additional truck and vehicle manufacturer partnerships would be announced later.
The expansion rests on the claim that a single model can serve multiple vehicle types. Because the Waabi Driver is one generalist system rather than a stack tuned for a specific platform, Urtasun has argued the same "brain" can drive both a Class 8 truck and a passenger car, letting the company enter robotaxis without rebuilding its technology. On the trucking side, Waabi continues to run commercial pilots in Texas through a multi-year capacity agreement with Uber Freight, the Dallas to Houston lane having served as its first route since 2023, with plans to expand across the U.S. Sun Belt and eventually North America.
Waabi is notable for three reasons. First, it is a leading test of the "AI-first," simulation-heavy approach to autonomy. Where Waymo and Aurora built their lead partly through very large real-world fleets and mileage, Waabi bets that an interpretable end-to-end model trained mostly in Waabi World can reach the same safety bar with far less capital and on-road testing. If that thesis holds, it would reshape how the economics of self-driving are understood; if it does not, the simulation-equivalence claims will face hard scrutiny.
Second, Waabi has become a flagship for Canadian deep tech. Its founder co-built the Vector Institute, its rounds rank among the country's largest, and the January 2026 financing was widely reported as the biggest in Canadian history, anchoring a high-profile AI company in Toronto rather than Silicon Valley.
Third, the Uber and NVIDIA relationships place Waabi inside the broader race to commercialize "physical AI." It is one of a small group of companies attempting to ship autonomous trucks and robotaxis from a common foundation model. Its near-term credibility will hinge on execution of the milestone it has so far been careful not to rush: a genuinely driver-out commercial launch, expected on Volvo's redundant trucks in the quarters following its 2026 raise.