Kodiak’s 75-Truck Deal Puts Driverless Service Inside DTL’s Fleet

A truck can belong to a freight carrier while its driving system arrives as a service billed by the mile. That is the arrangement behind Kodiak AI’s new agreement with Fresno-based DTL Transport: 75 planned driverless trucks that DTL would own and operate, using the Kodiak Driver. The interesting number is not only the fleet size. It is who takes responsibility for keeping each paid mile possible.

Kodiak announced the binding agreement on October 28. Its release identifies two initial Sun Belt routes and a commercial model built around licensing autonomy to an existing carrier. It does not establish that all 75 trucks are already on the road. For shippers and fleet operators, the next chapter will be measured in deliveries, service interruptions and the cost of making the system work.

What the carrier is buying

Under the announced Driver-as-a-Service arrangement, DTL pays a per-mile fee to use Kodiak’s autonomous driving system. The company describes this as its first publicly announced definitive agreement for an interstate carrier to operate driverless trucks in its own fleet. That qualification matters: Kodiak already has experience with customer-owned driverless vehicles in industrial operations.

In the October 28 public announcement, Kodiak founder and CEO Don Burnette explained why he favors the licensing model:

The appeal is a division of work. A carrier brings freight customers, dispatch experience and a fleet operation; a technology provider supplies the driving system. But a per-mile charge does not make the rest of the truck disappear from the balance sheet. Operators still need to understand the vehicle, maintenance, support and route conditions that make an autonomous mile possible.

The announcement does not disclose a per-mile price or a complete delivery schedule for the 75 trucks. That leaves the commercial economics open. A signed commitment shows customer interest; it does not yet reveal the cost per completed load.

A California pilot and a different launch map

The relationship follows a September pilot carrying perishable goods between Fresno and Los Angeles. Kodiak says those trips covered 225 miles along Interstate 5 and State Route 99 under a California heavy-duty autonomous-vehicle testing permit. A testing program and a future driverless freight service are different operating stages, and should be assessed separately.

DTL’s announced initial driverless lanes are Dallas–El Paso on Interstate 10 and Dallas–Atlanta along Interstate 20. Kodiak also says it plans a separate driverless highway launch carrying IKEA goods between Dallas and Houston by the end of 2026. Those are forward-looking plans, with different customers and route boundaries.

A route is more than a line on a map. A shipper would want to know where the automated segment starts and ends, how a disrupted delivery is recovered and who can authorize a change. Expanding a fleet should mean repeating a dependable service, rather than assuming that a successful trip settles every later route.

Uptime needs a careful definition

Kodiak’s October 20 reliability update reports 64 consecutive days without powering off or restarting an unsupervised industrial truck in the Permian Basin. The company defines the streak as the truck and autonomy system remaining running and available for service. That is not a claim that the vehicle drove continuously for 64 days.

In that public reliability update, Burnette connected the measure to what customers need from the service:

Availability is valuable, but it is only one part of the operating picture. A fleet buyer would also ask how much freight moved, how often support was needed, how long repairs took and whether the vehicle was available during the hours customers actually needed it. Industrial results offer evidence to examine, while highway operations must establish performance in their own conditions.

The hardware still has a job

Another October 28 announcement, from Stoneridge and Kodiak, puts the service model beside a very physical component: MirrorEye camera systems incorporated into newly deployed Kodiak-equipped trucks. The companies say the compact camera arrangement gives Kodiak more flexibility in designing its SensorPods. They also plan to explore deeper integration; that future work should not be described as completed.

Hardware that is easier to replace can matter to a business charging for usable miles. A damaged component means something different when the question is how quickly an operator can restore service. Sparknify’s recent reporting on GXO’s planned Digit warehouse expansion raised a related point: a working machine depends on the handoffs, support and recovery procedures around it. A truck and a warehouse humanoid do different jobs, but both need a clear answer when ordinary operations stop following the expected script.

Follow the completed load

The most useful next updates would connect fleet deliveries with actual service: which lanes are running, how performance is defined, what support costs and what changes after an interruption. They would also distinguish the announced DTL fleet from Kodiak’s other customers and deployments.

For Silicon Valley’s physical-AI companies, this agreement makes the product question concrete. Selling autonomy by the mile means explaining what the customer receives when hardware, software and freight operations meet. If the 75-truck plan progresses, will the clearest proof of autonomy’s value be a completed load arriving as reliably as the carrier promised?

Subscribe to Sparknify for more tech news.