Latest in AI
Kodiak and DTL Agree on 75 Driverless Trucks With Per-Mile AI Licensing
The agreement moves autonomous trucking toward customer-owned fleets, but a signed deployment plan is not a live highway service.
Kodiak AI announced an agreement on October 28 to equip 75 DTL Transport trucks with its autonomous driving system. DTL will own and operate the vehicles and pay per mile for Kodiak Driver. The announcement is a commercial commitment, not confirmation that those trucks are already operating without drivers.
What the Kodiak and DTL agreement covers
The official announcement identifies Dallas to El Paso and Dallas to Atlanta as the initial planned routes. Kodiak describes the deal as the first publicly announced definitive agreement for an interstate carrier to run driverless trucks in its own fleet through this business model.
That industry-first description is Kodiak's claim. The concrete facts are the fleet commitment, customer ownership and per-mile charging arrangement.
TechCrunch reports that deliveries start in 2028. The agreement therefore should not be confused with an immediate fleet rollout.
Driver-as-a-Service changes who owns the operation
Kodiak also says it plans to launch a separate driverless highway service carrying IKEA goods between Dallas and Houston by year-end. That planned service is distinct from the 75-truck DTL agreement.
The DTL structure matters because the customer owns the trucks while licensing the driving system. For an AI supplier, that separates the software relationship from the responsibility of financing and running every vehicle itself.
It does not eliminate operational dependence. If the driving system is unavailable, a carrier still has a truck that needs to move freight. The commercial value depends on the supported routes, operating conditions, intervention procedures and downtime arrangements, not just the nominal price per mile.
A pilot and a fleet commitment answer different questions
Kodiak and DTL previously ran a California pilot. The September announcement describes a 225-mile freight journey between Fresno and Los Angeles under a California testing permit, with a human safety driver behind the wheel.
A pilot can show that a particular route and load fit a system's capabilities. A fleet commitment asks a harder question: can the technology fit the carrier's dispatch, maintenance and customer-service operation repeatedly?
That distinction matters for evaluating physical AI businesses. A signed contract is stronger commercial evidence than a demonstration, but it is not the same as sustained revenue-producing operation. Weather, route changes and vehicle availability can all affect how much of a fleet's capacity becomes billable activity.
What to watch next
The useful milestones are deployment timing, actual operating lanes and the proportion of planned work completed under the intended commercial model. A count of committed trucks alone cannot answer those questions.
For AI companies selling into physical operations, this is a concrete example of a product boundary: provide the capability, let the customer retain the assets, and make the recurring charge track use. The harder work begins when that clean contract structure meets a daily operating schedule.
I'm Paras. I work at the intersection of AI, engineering and distribution.
I help technical companies explain complex AI products and business models through clear launch stories and founder-led content.