On 5 October, Volvo Autonomous Solutions and Waabi said customer operations have begun in Texas for the first time under their partnership. The concrete version: Volvo VNL Autonomous tractors fitted with the Waabi Driver are hauling freight along the Dallas–Houston corridor for Warp, an open freight network that sells less-than-truckload (LTL) and full-truckload capacity to transport companies. The freight moves directly to third-party facilities, not merely between the partners' own yards.
That last detail carries more weight than the announcement's smooth cadence suggests. A truck running a loop between two company terminals is a test. A truck delivering someone else's goods into someone else's dock is a service, and it has to survive the other party's schedule.
Why LTL is the unforgiving test
Full truckload is the easy case for automation: one shipper, one trailer, one destination, few handoffs. LTL is the opposite — freight from several customers has to be consolidated, moved and distributed on tightly coordinated timetables, and every missed connection ripples through the network.
Volvo's framing is that carrying freight directly between facilities, without transshipment, gives LTL operators a more predictable way to serve their networks as planned. Read that as an argument about handoffs: fewer docks and fewer transfers mean fewer places for a schedule to break. It is a reasonable claim, and it is still a claim. The release does not say how many loads are running, how often they leave on time, or how the trucks behave when a shipper's dock is blocked.
Warp's chief executive, Daniel Sokolovsky, made the network case rather than the truck case: autonomy is worth more when freight flows can be coordinated around it. Warp aggregates shipments from multiple customers and synchronises routes, capacity and schedules, so an improvement on a single lane can create value across the network. Volvo's release lists Walmart, DoorDash and Gopuff among Warp's customers.
The observer is still in the seat
As of the announcement, the autonomous runs are carried out with an observer in the driver's seat. That is standard for a first customer lane and it is also the honest headline: this is supervised freight, not driverless freight.
The vehicle underneath is purpose-built rather than converted. Volvo says the VNL Autonomous was developed from the ground up for autonomous driving, with redundancy in six safety-critical systems, and is designed to host driving systems from partners such as Waabi. Redundancy in braking, steering and the rest is the engineering answer to the obvious question — what happens when a sensor or an actuator fails at 65 mph.
What is not published is the more useful set of numbers: cost per mile, freight volumes, intervention rates, safety records. Neither company has offered them, and without them the milestone is a statement about capability, not about economics. That gap is not a scandal; first customer lanes rarely come with a data appendix. But it does mean the difference between "autonomous freight works" and "autonomous freight works commercially" is still unmeasured in public.
What is being sold, not just demonstrated
The offering is packaged as Autona/freight: the autonomous truck, Waabi's driving technology, vehicle and transport management, terminals and customer facilities, operations, and uptime support, bundled into a single transport solution built to be scaled up.
That packaging is the part worth watching. Selling a truck is a manufacturing business; selling a transport solution that includes terminals and uptime support is an operations business, with different margins and different failure modes. Nils Jaeger, president of Volvo Autonomous Solutions, cast the start as bringing autonomous freight into customer operations in one of logistics' most demanding segments, and said the corridor experience will be used to strengthen the partnership and improve the ecosystem as it scales.
Waabi's Lior Ron called the Warp work the first milestone on the path to trucks being integrated into everyday freight networks, and argued the Volvo–Waabi combination can handle everything from LTL network operations to direct delivery at third-party facilities.
Waabi's own figures come from the same release and should be read as company statements: founded by AI researcher Raquel Urtasun; already hauling freight in Texas for Fortune 50 customers with its own technology; expanding into robotaxi with at least 25,000 vehicles planned on Uber's platform in the coming years; more than $1.3 billion raised, with Khosla Ventures, NVIDIA, Volvo Group Venture Capital and Uber among its backers. None of it is independently verified here.
Next step, per the release: use the Dallas–Houston experience to prepare additional routes and support broader scaling in the coming years. No route list, volumes or timetable.
Japan is testing the other answer
A different response to the same pressures is taking shape in Japan. The Ministry of Land, Infrastructure, Transport and Tourism said in a 20 October release that demonstration experiments for its AutoFlow Road concept will begin on 28 October 2026 at the National Institute for Land and Infrastructure Management in Tsukuba, then move to Narita International Airport and company sites.
The concept is dedicated freight space on roadways, carrying goods with unmanned, automated transport powered by clean energy — motivated by the logistics labour crunch and the push for carbon neutrality. Where Volvo and Waabi put autonomous trucks on ordinary roads and let the software absorb the complexity, Japan is asking whether the road itself should be rebuilt for freight.
The schedule names consortia and windows. At NILIM, 28 October to 2 November, led by Taisei with Tier IV and Taisei Rotec. At Narita: Shimizu from 9 to 18 November with IHI and Nomura Real Estate; Chiba Prefecture from 19 to 25 November with Narita International Airport Corp., Obayashi, Cuebus and NTT Data; Kajima from 7 to 18 December with Mitsubishi Electric, Seino Transportation and Dynamic Map Platform. Company-site trials run in November and December. Press access is limited to designated viewing days, which tells you how early this still is.
Two models, one problem. The Texas lane is manufacturer-led and commercial, on existing asphalt, supervised, single corridor, no operating data. The Japanese programme is government-funded and infrastructural, pre-commercial, and also without results. If the dedicated-lane bet works, it sidesteps the hardest parts of driving; if it doesn't, the truck-side approach gets years of head start. For now the useful things to watch are dull ones: interventions per thousand miles, whether a second lane opens, whether the observer leaves the seat, and whether anyone publishes a cost per mile.