The history of autonomous vehicles is littered with the corpses of companies that promised too much and delivered too little. Waymo has spent over $5 billion since 2009 and operates in a handful of geofenced cities. Cruise imploded spectacularly after a pedestrian was dragged under one of its vehicles in San Francisco. Uber itself abandoned its self-driving program in 2020 after years of setbacks and a fatal accident in Arizona. The industry became synonymous with overpromise and underdelivery, a graveyard of investor capital and engineering hubris.
Against this backdrop, the announcement last week that Waabi, a Toronto-based autonomous vehicle company, had raised $1 billion in new funding and struck a partnership with Uber to deploy at least 25,000 robotaxis demands scrutiny. This is the largest technology fundraising in Canadian history. The investors include Khosla Ventures, G2 Venture Partners, Nvidia, Volvo, Porsche, BlackRock, and a subsidiary of the Abu Dhabi Investment Authority. Uber itself contributed $250 million in milestone-based funding tied to the robotaxi deployment.
What makes Waabi different from the failures that preceded it is not merely ambition or capital. It is architecture.
The Simulation-First Thesis
Raquel Urtasun founded Waabi in 2021 after leaving her position as head of Uber’s autonomous vehicle research lab. She is one of the most cited researchers in machine learning and computer vision, and she built Waabi around an insight that diverges from the conventional wisdom of the industry.
The traditional approach to autonomous driving, exemplified by Waymo and Cruise, relies on accumulating millions of real-world miles. Vehicles equipped with sensors drive through cities, recording data, encountering edge cases, and gradually building a corpus of experience from which the AI learns. This approach is extraordinarily expensive. It requires fleets of vehicles, teams of safety drivers, and years of operation in each new geographic area.
Urtasun’s thesis is that simulation can compress this timeline by orders of magnitude. Rather than logging millions of real-world miles, Waabi uses what the company calls the “world’s most advanced neural simulator” to generate synthetic training data. The AI encounters edge cases, rare events, dangerous scenarios, in simulation rather than on public roads. When it does encounter real-world situations, it has already trained on millions of variations.
This is not the crude simulation of earlier generations, where a human designed scenarios and the AI learned to respond to them. Waabi’s simulator generates novel scenarios from learned distributions. The AI does not memorize responses to predetermined situations. It learns to generalize from an effectively unlimited corpus of synthetic experience.
The approach has a second advantage: it is dramatically cheaper. Waabi can explore the long tail of dangerous scenarios, the pedestrian stepping out from behind a bus, the construction zone with confusing signage, the vehicle swerving to avoid debris, without putting anyone at risk and without the expense of operating physical vehicles.
One Brain, Many Bodies
The most consequential aspect of Waabi’s architecture is what the company calls “Physical AI”: a single AI model that can drive different vehicle types across different geographies and driving conditions. The exact same neural network that pilots an autonomous semi-truck on a Texas highway can pilot a robotaxi through downtown Toronto.
This is architecturally novel. Most autonomous vehicle companies build separate systems for different vehicle types. Waymo’s trucking unit operates on a different stack than its passenger vehicles. The sensors differ. The planning algorithms differ. The training data is siloed.
Waabi’s unified approach means that learning transfers between domains. Urban driving capabilities developed for robotaxis directly improve the trucking system’s ability to handle complex highway scenarios, construction zones, and last-mile delivery. Highway expertise from trucking operations feeds back into the robotaxi system. The company accumulates experience in a single model rather than fragmenting it across separate products.
This is what makes the Uber partnership structurally interesting rather than merely financially significant.
The Partnership Mechanics
Under the arrangement, Uber and Waabi have divided responsibilities along clear lines. Waabi focuses exclusively on building the AI driver. It does not own vehicles. It does not manage fleets. It does not handle operations. Uber handles everything else: vehicle procurement, maintenance, cleaning, charging, customer acquisition, and the platform that connects riders to vehicles.
This division mirrors the structure of asset-light technology businesses in other domains. Airbnb does not own hotels. Uber does not own vehicles in its traditional ridesharing business. Waabi does not need to become an automotive company to scale its AI.
The $250 million milestone-based investment from Uber is tied to deployment targets. Waabi must demonstrate that its AI can safely operate robotaxis before Uber releases capital. This aligns incentives in ways that pure venture funding does not. Uber is not simply betting on Waabi’s potential. It is paying for delivered capability.
The initial deployment of 25,000 robotaxis will operate on the Uber platform, competing for rides alongside human drivers. If the technology works, Uber gains access to vehicles that operate 24 hours per day, require no wages, and improve their driving through accumulated experience. If the technology fails, Uber has limited its exposure to the milestone-based investment rather than the billions it spent on its own ill-fated autonomous program.
The Trucking Timeline
Waabi’s autonomous trucking operations are further advanced than its robotaxi program. In October 2025, the company announced integration of its AI software into Volvo’s fleet of autonomous trucks, which provide freight delivery services on highways in Texas and at mining and quarrying sites in Norway and Sweden. These operations currently use safety drivers.
Urtasun has stated that Waabi will not launch fully driverless trucking operations until the Volvo platform is “fully validated,” a decision she frames as prioritizing safety over speed. Volvo has indicated that full validation is “just quarters away.” The commercialization target for trucking remains 2027.
This measured approach contrasts with the industry’s history of premature deployment and subsequent disaster. Cruise launched robotaxis before its technology was ready and paid the price when a pedestrian incident forced it to halt operations entirely. Waabi appears to be learning from these failures, though only time will reveal whether its caution reflects genuine technological limitation or prudent restraint.
The Trillion Dollar Market
The stakes are substantial. Urtasun has described the addressable market for autonomous vehicles as exceeding $1 trillion. This figure encompasses trucking, ridesharing, delivery, and private vehicle applications. The company that solves autonomous driving at scale will not merely disrupt transportation. It will become one of the most valuable enterprises on Earth.
Waabi’s current valuation, while undisclosed, is almost certainly in the multi-billion dollar range given the size of the funding round and the caliber of investors. Whether that valuation proves justified depends on execution across multiple dimensions: regulatory approval, safety performance, geographic expansion, and competitive response from Waymo, Tesla, and emerging Chinese players like Pony.ai and WeRide.
What is evident today is that Waabi represents a genuinely different approach to the autonomous vehicle problem. The simulation-first methodology, the unified Physical AI architecture, and the asset-light partnership model with Uber collectively constitute an alternative thesis to the approach that has dominated and largely failed for the past decade.
What This Means for Everyday People
For truck drivers, the implications are existential. There are approximately 3.5 million trucking jobs in the United States alone. If autonomous trucking reaches commercial viability, these jobs will not disappear overnight, but they will transform. Long-haul routes will be automated first. Drivers may shift to first-mile and last-mile operations, supervision of automated convoys, or exit the industry entirely.
For urban residents, the prospect of tens of thousands of robotaxis competing with human drivers raises questions about congestion, employment, and the character of cities. Robotaxis do not need to park. They can circle endlessly, waiting for the next ride. Whether this reduces traffic or increases it depends on regulatory frameworks that do not yet exist.
For the broader economy, successful autonomous driving represents a productivity gain comparable to containerized shipping or commercial aviation. Goods move faster and cheaper. People reach destinations without owning vehicles. The value creation is immense. So is the disruption.
Waabi is not the only company pursuing this future. It may not be the company that ultimately wins. But its architecture, its approach, and its partnership with Uber make it one of the most credible contenders to emerge in years. The simulation-first thesis may prove correct. The Physical AI architecture may enable scaling that competitors cannot match. Or it may join the long list of autonomous vehicle companies that promised transformation and delivered disappointment.
The billion-dollar bet has been placed. The results will unfold over the next several years.
For inquiries and analysis contact laterstack@proton.me
Frequently Asked Questions
What is Waabi’s Physical AI platform?
Waabi’s Physical AI platform is a unified artificial intelligence system that can drive different vehicle types across various geographies and conditions. Unlike competitors that build separate AI systems for trucks and passenger vehicles, Waabi uses a single neural network that transfers learning between domains, allowing urban driving skills to improve highway performance and vice versa.
How does the Waabi Uber partnership work?
Under the partnership, Waabi focuses exclusively on building the AI driver while Uber handles all operational aspects including vehicle procurement, maintenance, cleaning, charging, and customer platform integration. Uber invested $250 million in milestone-based funding tied to deployment targets, meaning capital is released as Waabi demonstrates safe robotaxi operation.
When will Waabi’s autonomous vehicles be available?
Waabi’s autonomous trucking operations using Volvo trucks are currently operating with safety drivers in Texas and Scandinavia, with fully driverless operations expected once the platform is “fully validated,” likely in 2027. Specific robotaxi deployment timelines for the Uber partnership have not been announced but are expected in the coming months.