Logistics · Deep dive

Waabi

Toronto- and Texas-based autonomous trucking company betting that a neural simulator — Waabi World — can train and prove a driverless truck safe with a fraction of the road miles rivals burn, now expanding the same 'Waabi Driver' into Uber robotaxis after a $750M January 2026 Series C.

emerging

The question that decides it: Waabi's entire thesis is that Waabi World's closed-loop neural simulation can constitute the safety case — that a claimed 99.7% simulation-realism score (August 2025) plus mixed-reality testing can convince regulators, insurers, and Volvo's lawyers to pull the human out of the cab with a fraction of the road miles rivals have banked. Does that case actually close — a driver-out commercial launch on the Volvo redundant platform in the US Southwest by end-2026 as now promised — before Aurora's 18-month head start of real driverless revenue miles and Kodiak's public-market deployment data harden into the industry's de facto validation standard, and before the 25,000-robotaxi Uber expansion splits Waabi's 300-person team across two unproven markets at once?

Snapshot

Waabi is a Toronto-founded autonomous trucking company running commercial freight pilots (with safety drivers) on Texas lanes for Uber Freight and Samsung, built on a contrarian premise: a high-fidelity neural simulator, not millions of road miles, is the fastest way to train and certify a driverless truck. Founded in 2021 by Raquel Urtasun, former chief scientist of Uber’s collapsed self-driving unit, it closed a $750 million Series C in January 2026 co-led by Khosla Ventures and G2 Venture Partners — the largest venture round in Canadian history — plus roughly $250 million in milestone-based Uber capital tied to deploying 25,000-plus robotaxis exclusively on Uber’s network. It matters because it is the best-funded test of whether simulation can substitute for fleet miles in a safety case — and because its driverless launch, promised for end-2025, has slipped while Aurora already runs driver-out trucks for money.

Founding story

Waabi is the second act of one of the most public failures in autonomous driving. Urtasun, a Spanish-born computer-vision researcher, University of Toronto professor, and Vector Institute co-founder, was hired by Uber in May 2017 to build ATG’s Toronto research lab and serve as the unit’s chief scientist. She had a front-row seat to everything that went wrong: the March 2018 Tempe, Arizona pedestrian fatality, the billions burned on a hand-engineered stack, and the December 2020 fire sale of ATG to Aurora at a ~$4 billion valuation — roughly half its 2019 SoftBank/Toyota mark. Uber paid Aurora $400 million to take the unit; 1,200 ATG employees transferred. Urtasun did not. She left in early 2021 and launched Waabi that June with an $83.5 million Series A led by Khosla Ventures — with, tellingly, both Uber and Aurora on the cap table, plus personal checks from Geoffrey Hinton, Fei-Fei Li, Sanja Fidler, and Pieter Abbeel (Forbes, June 2021). Her founding argument: modular hand-coded stacks validated by brute-force road miles were why $100 billion had produced so little, and an end-to-end, simulation-trained system could commercialize faster with a fraction of the people and capital. The honest read: she is arguably the most credentialed AI researcher ever to found an AV company, and she watched the fleet-miles playbook fail from the inside — but she has never shipped a driver-out commercial product, and the thesis remains unproven where it counts.

How it works

Two pieces: the Waabi Driver and Waabi World. The Driver is a single end-to-end AI system — sensors in, driving decisions out — that Waabi describes as interpretable and verifiable rather than a black box, running on Nvidia compute with lidar, radar, and cameras. What differentiates Waabi is where the Driver learns. Waabi World is a closed-loop neural simulator: technology like UniSim (published research, 2023) takes a single real-world sensor log and reconstructs it into an editable, reactive simulation with what Waabi claims is near-zero domain gap — the system under test behaves the same in simulation as it would on the road. Engineers can then perturb the scene: spawn a cut-in, drop debris, move the sun, or re-render car-collected data as if a truck’s sensor rack had captured it. Because a truck logging highway miles mostly sees empty lanes, Waabi’s pitch is that rivals need hundreds of millions of road miles to encounter enough rare events to validate safety, while Waabi manufactures those events at will and reserves real roads for confirmation. In August 2025 the company claimed a 99.7% simulation-realism score and detailed mixed-reality testing — real trucks on closed courses reacting to virtual traffic injected into their sensor streams (Forbes, August 2025). The physical platform is the Volvo VNL Autonomous, built with redundant steering, braking, and power at Volvo’s New River Valley, Virginia plant, unveiled with the Waabi Driver integrated in October 2025. In June 2026 Waabi demonstrated transferring the Driver across truck platforms without retraining — the basis for its claim that one AI brain can also power robotaxis and, eventually, other robots.

Product and business overview

Three named components. The Waabi Driver, the AI driver sold as the product itself. Waabi World, the simulator — internal tooling, not sold separately, but the core asset investors are underwriting. And the physical network: the Lancaster, Texas terminal opened April 2024 — eight-plus acres south of Dallas with a 24,000-square-foot shop, fueling, scales, and high-speed data offload — anchoring routes between Dallas, Houston, and an expanding Southwest footprint (2026 company statements). Distribution runs through a 10-year Uber Freight deal (September 2023) feeding Waabi trucks marketplace demand, and the Volvo Autonomous Solutions agreement (February 2025) covering joint development and factory production of autonomy-ready trucks. The January 2026 Uber robotaxi agreement adds a second product line: 25,000-plus robotaxis, exclusive to Uber, vehicle partner unnamed.

Business model and pricing

Waabi is converging on “driver as a service”: selling the Waabi Driver — per-mile or per-truck subscription — to carriers and fleets who buy Volvo VNL Autonomous trucks, with Uber Freight supplying demand (TechCrunch, January 2026). No pricing has ever been published. Today’s actual revenue mechanics are more modest: supervised trucks hauling loads for Uber Freight customers and Samsung on Texas lanes, booking freight revenue like a small carrier. The economic logic: driver labor is roughly 43% of per-mile truck operating cost (industry analyses, 2025-26), so a working software driver prices against a $70,000-plus annual salary per seat. The robotaxi deal borrows Uber’s network economics — Uber funds deployment in milestone tranches and takes distribution, Waabi supplies the driver. Until a driver-out launch this is all prospective; there is no disclosed revenue, and the one circulating third-party estimate (~$39M in 2024, Latka) is unaudited and looks like pilot freight billings at best.

Traction over time

The pattern to notice: capital and partnerships compound — Uber, Volvo, Nvidia, Porsche SE are all repeat investors — but the operational milestone that matters has moved. In June 2024 Waabi raised its Series B explicitly “to launch fully driverless trucks in 2025.” By October 2025 that became driverless software readiness by end-2025, an observer remaining until the redundant Volvo hardware finishes validation (Forbes, October 2025). By early 2026 the target was fully driverless operations across the US Southwest by end-2026. Zero driverless commercial miles as of July 2026, against Aurora’s 14 months of them.

Market analysis

US trucking generates roughly $900 billion in annual revenue (ATA, 2023-24) and is structurally short of drivers: an estimated 82,000-driver shortage in 2024, projected toward 160,000-plus by 2031 (ATA). The autonomous-truck segment is pegged at ~$42.6 billion in 2026, growing to ~$74.2 billion by 2031 (11.7% CAGR, Mordor Intelligence, 2026). The structural forces are real: hub-to-hub highway freight is the most tractable AV problem, Texas regulation is permissive, and labor is the largest variable cost. The robotaxi expansion adds a second, hotter TAM — Uber assembles AV supply partners (Waymo, Nuro, WeRide, now Waabi) rather than building its own — but it drops Waabi into urban driving, where Waymo has a decade and millions of driverless miles of lead. The graveyard tempers everything: TuSimple (delisted January 2024 after a 2022 highway crash and a governance scandal), Embark ($5B SPAC to $71M asset sale, 2023), Locomation, Starsky — all dead before meaningful revenue.

Competitive intel

Aurora is the direct antagonist — publicly traded at ~$13.4 billion (July 2026), running commercial driverless freight on Dallas-Houston since May 2025 with Uber Freight and Hirschbach, expanding to El Paso and Phoenix. It absorbed Urtasun’s old ATG organization; Aurora attacks Waabi with incumbency of proof — actual driver-out miles, a published safety-case framework, and OEM deals with Paccar and Volvo (the latter shared with Waabi, awkwardly). Kodiak AI went public via SPAC (September 2025; ~$825M market cap and $1.8M Q1 2026 revenue on 28 driverless trucks in the Permian Basin) — small, but disclosing real numbers quarterly, which resets diligence expectations for everyone private. Torc/Daimler brings the deepest pockets and a 2027 launch on Freightliner hardware; if OEM integration is the moat, Daimler owns its own. Plus supplies software to TRATON and IVECO — the asset-light version of Waabi’s driver-as-a-service ambition, achieved by surrendering the fleet. Bot Auto, run by TuSimple’s founder on ~$45M, did a humanless Texas run in 2025 and frames the race as cost-per-mile discipline — a living rebuke to billion-dollar raises. Waabi’s genuine edges: the field’s strongest AI-research pedigree, a simulation asset that compounds across vehicle classes, Volvo factory production rather than retrofits, and Uber locked in as investor and demand channel on two products. None of that yet substitutes for one driver-out revenue mile.

History and evolution

• May 2017 — Urtasun joins Uber ATG as chief scientist; builds the Toronto lab.
• Mar 2018 — ATG truck-and-robotaxi program halted after the Tempe fatality; the unit never recovers.
• Dec 2020 — Uber sells ATG to Aurora (~$4B valuation); Urtasun declines to transfer.
• Jun 8, 2021 — Waabi launches with $83.5M Series A led by Khosla; Uber and Aurora both invest.
• Feb 2022 — Waabi World unveiled; UniSim research published 2023.
• Sep 2023 — 10-year Uber Freight partnership; supervised commercial loads begin on Dallas-Houston.
• Apr 2024 — Lancaster, TX terminal opens.
• Jun 18, 2024 — $200M Series B led by Uber and Khosla; Nvidia, Volvo, Porsche SE join; “fully driverless in 2025” promised.
• Feb 4, 2025 — Volvo Autonomous Solutions partnership; factory-built VNL Autonomous.
• May 2025 — Aurora beats everyone to commercial driverless trucking.
• Aug 2025 — Lior Ron, founder and decade-long CEO of Uber Freight (a $5B-revenue business), joins as COO; 99.7% simulation-realism claim published.
• Oct 28, 2025 — Production Volvo VNL Autonomous with integrated Waabi Driver unveiled; driverless launch reframed as software-readiness, observer retained.
• Jan 28, 2026 — $750M Series C co-led by Khosla and G2; Uber commits ~$250M milestone capital for 25,000+ robotaxis; ~$1B total. Valuation undisclosed.
• Jun 26, 2026 — Cross-platform AI-driver transfer demonstrated; driver-out target now end-2026, US Southwest.

What people say

The case for. MIT Technology Review (March 2025) gave Waabi’s simulation-based safety case a serious hearing, noting the genuine statistical problem it solves: highway trucks rarely encounter dangerous events, so road miles alone cannot surface enough edge cases in any reasonable time. Forbes’ AV analyst Richard Bishop (August and October 2025) treated the mixed-reality testing regime and Volvo integration as substantive engineering, not vaporware. CNBC named Waabi a 2025 Disruptor 50 company. The strongest signals are behavioral: Nvidia, Volvo, Porsche SE, and Uber all re-upped in 2026, Uber tied a quarter-billion dollars to Waabi milestones, and Lior Ron — who built Uber Freight — left to be COO. Glassdoor sits at 4.6/5 across 16 reviews (2025-26), with praise for the caliber of the team.

The complaints. Wayve chief scientist Jamie Shotton, in MIT Technology Review (March 2025), delivered the polite version of the field’s skepticism: Waabi’s realism claims lack disclosed technical detail, and real-world testing remains indispensable — simulation validated against the same data it was built from risks circularity. The delay is the sharper critique: “fully driverless in 2025” was the Series B’s headline promise, and it did not happen; the company’s explanation (software ready, hardware validation pending) is plausible and also exactly what a slipping program sounds like. The robotaxi expansion drew focus-risk criticism in early-2026 coverage — a ~300-person company that has not shipped driver-out trucks now also promising 25,000 urban robotaxis against Waymo, Tesla, and Zoox. Small-sample Glassdoor negatives are pointed: one review alleges retaliation for reporting issues and “toxic non-technical leaders.” And the disclosure gap persists — no revenue, no valuation, no third-party audit of the 99.7% realism figure — in an industry where TuSimple and Embark taught investors what undisclosed fragility costs.

Outlook: the open question

Waabi works if the simulation-first safety case closes in the real world on schedule — meaning driver-out commercial freight on the Volvo redundant platform in Texas or the broader Southwest by end-2026, accepted by regulators and insurers, with the safety methodology published in enough detail that neutral experts stop calling it unverifiable. If that happens, the thesis inverts the industry’s cost structure: driver-out reached on roughly $330 million of pre-Series-C capital where Aurora consumed billions, a simulator that ports the same driver to robotaxis at low marginal cost, and Uber’s dual demand channels solving distribution — the piece every dead AV trucking company lacked. Waabi fails if driver-out slips again into 2027 — at which point the pattern is the promise moving, not the product — or if the first driverless launch requires so much supplementary road validation that the simulation advantage was rhetorical, leaving Waabi as a subscale Aurora, two years behind, with a robotaxi program subsidized by Uber precisely because Uber pays only on milestones Waabi has not yet demonstrated it can hit. Watch three things through 2027: whether trucks go driver-out commercially by December 2026 and stay out; whether Waabi publishes its safety case (Urtasun has herself demanded industry-wide safety-data disclosure — FreightWaves, 2025); and whether a named robotaxi platform and launch city materialize or the 25,000-unit figure quietly recedes. The capital and pedigree are real. But in autonomous trucking only demonstrated driverless miles have separated the living from the dead — and as of July 2026, Waabi has none.

Sources and further reading

• AI pioneer Raquel Urtasun launches self-driving technology startup with backing from Khosla, Uber and Aurora (TechCrunch, June 8, 2021)
• Uber sells self-driving unit to Aurora, ending a tumultuous era (The Robot Report, December 2020)
• Uber, Khosla, Nvidia invest in $200 million funding round for autonomous trucking startup Waabi (CNBC, June 18, 2024)
• Waabi says its virtual robotrucks are realistic enough to prove the real ones are safe (MIT Technology Review, March 11, 2025 — includes the Wayve skeptic’s case)
• Volvo to deploy Waabi self-driving gen AI to get more autonomous freight trucks on the road (CNBC, February 4, 2025)
• Mixed Reality Testing Underpins Waabi’s Commercial Driverless Launch (Forbes, August 7, 2025)
• Waabi And Volvo Reach Key Goal In Integrating Driverless Tech (Forbes, October 31, 2025 — the driverless-launch reframing)
• Autonomous trucking startup Waabi raises $750 million to expand into robotaxis (CNBC, January 28, 2026)
• Waabi raises $1B and expands into robotaxis with Uber (TechCrunch, January 28, 2026)
• Aurora Begins Commercial Driverless Trucking in Texas (Aurora IR, May 2025)
• Kodiak AI reports 74% Q1 revenue growth, fleet reaches 28 driverless trucks (FreightWaves, 2026)
• Waabi demonstrates AI driver transfer between truck platforms (Trucking Dive, June 2026)
• Waabi CEO sees success in simulation, wants AV industry to disclose safety data (FreightWaves, 2025)