The Texas driverless truck has to finish the job

The freight job sorts the operators

Aurora reports public-road driverless freight; Gatik describes recurring driverless regional networks. Kodiak’s Permian work is ahead of its highway program, while Waabi and PlusAI disclose people onboard. I would compare those released services before comparing autonomy labels.

The purchase has to clear three tests: the technology can perform the lane in its released conditions, the provider can keep supplying it at an attractive cost, and the shipper saves enough at the quoted terms. A completed shipment opens the first inquiry. I would use the pilot and contract to resolve the other two.

Driverless does not mean laborless

Aurora’s July 22 explanation describes an empty driver’s seat in the second-generation International LT, with a rear-seat roadside-support person on some trips. The company says that person neither drives nor supervises driving. I would put the remaining support task, its supplier, and its price into the tender. Platform and support model.

Aurora’s Transportation-as-a-Service owns and operates trucks while customers buy dedicated capacity; customer ownership comes later in its plan. Kodiak’s Driver-as-a-Service puts technology on customer-owned and operated trucks. I would reconcile the asset, staffing, and operating work before comparing their per-mile prices. Aurora’s model and Kodiak’s arrangement sell different commitments.

FMCSA limits a property-carrying human driver to 11 driving hours after 10 consecutive hours off duty, within its 14-hour window and stated exceptions. I would compare the driverless offer with the teams, relays, or drop-and-hook service the buyer can already obtain, then follow the proposed extra turns through dock availability, freight demand, and released operating conditions before paying for a tractor’s ability to work across a longer schedule. Human-driver hours.

Aurora’s July letter reports driverless work at Midland–Monahans customer facilities while describing fueling and weigh-station navigation as supervised testing. I would require the released configuration and handoff plan for those activities before pricing a continuous long-haul itinerary. Operating account.

Permian work is ahead of the highway promise

Kodiak reported 35 Atlas-owned driverless trucks and more than 40,000 paid driverless hours at June 30. Atlas then announced a second Dune Express load-out point, with public-road work targeted for early 2027 subject to milestones. I would underwrite the existing industrial operation on its own routes and contract. The industrial measures and expansion plan describe successive stages.

The September highway release targeted unsupervised Dallas–Houston service by year-end 2026. The October 8 Charger operation on Dallas–Laredo still used a safety driver. Its 96% Autonomy Readiness Measure counted materially completed safety-case claims and evidence. I read it as progress through that safety case, with highway availability determined by the released service and Charger operating terms.

A store route and a highway load need different trucks

Gatik uses 26- and 30-foot trucks between facilities and stores in its disclosed DFW networks. PepsiCo’s June announcement identifies existing driverless work in Texas, Arizona, and Arkansas. I would specify store windows, temperature requirements, and the receiving dock before inviting a long-haul comparison. The regional vehicles serve the PepsiCo network’s job.

Bot Auto’s Houston–Hutchins shipment is a linehaul example. Its release reports no onboard safety staff or dependence on low-latency remote human feedback. I would ask for repeated-load costs and a current tariff to assess its profitability claim. Commercial shipment.

If a provider takes the easiest windows and best-balanced lanes, its accepted-load record can improve while the conventional remainder gets harder and more expensive, so I would give every candidate the same tender and carry rejected loads into the comparison at the price needed to move them. The shipper still has to finish the job.

Customer freight can still have a person onboard

Waabi and Volvo’s October 5 Warp release reaches third-party Dallas–Houston facilities with an observer in the driver’s seat. The International/Ryder PlusAI trial on Laredo–Temple reports 92% autonomous route coverage under a safety driver and inspections below 30 minutes. I would price those disclosed human roles into the current service. The Warp release and Ryder pilot describe operating shipments.

PlusAI targets factory-built commercial launch in 2027. Torc pairs DFW testing and its fleet hub with closed-track driver-out validation and a 2027 market-entry plan. Einride’s SH 130 memorandum and Austin demonstration cover a testbed and an R&D project. I would use those plans to prepare a supplier bid and require a released service before assigning next week’s freight. The PlusAI roadmap, Torc account, and Einride’s testbed and demonstration identify the stages.

The $1,150 quote becomes a $1,310 route

The first-year model moves 1,000 one-way Dallas–Houston shipments, each covering 250 loaded road miles. The current carrier charges $1,400, and the shipper performs $100 of warehouse work. I compare every new offer with that $1,500 total.

The $1,150 direct quote becomes $1,310 after inspection, release, and warehouse work. The $950 transfer quote becomes $1,380 after two local moves and the added handoffs. I also charge each new service $40,000 of setup in the first year. The table shows the full price stack.

I count provider equipment, fuel, tolls, maintenance, normal support, and repositioning inside the service quote. Charging them again would duplicate the bill. I add the shipper’s warehouse and release work at its assigned economic value, because existing staff still have to do it.

The rejected loads stay on the shipper’s bill

I give the new service 800 completed shipments, including recovered loads, and leave 200 with the current carrier at $1,500. Direct then costs $1,388,000 including setup before extra exceptions, against the old arrangement’s $1,500,000. Transfers cost $1,444,000.

At a 2% incremental exception rate on new-service loads, sixteen events add $350 of recovery and $500 of lost contribution each. That $13,600 burden leaves $98,400 of first-year saving for direct and $42,400 for transfers. I count cash costs, allocated labor, and lost contribution on the same economic basis.

I would ask the backup carrier to quote the 200-load remainder. If it contains harder windows or worse return freight, the old average price may disappear. A cheap accepted subset can leave an expensive transport business behind it.

The model leaves $123 per load for surprises

Each accepted direct load contributes $173 toward $40,000 of setup after its expected exception cost. Break-even is about 231.2 shipments, making 232 the first whole load above it. At 800 accepted loads, I have $123 per load left for additional costs before the $98,400 saving disappears.

The road-service ceiling is $1,273. From the old $1,500 total I subtract $50 of setup per accepted load, $17 of expected incremental exceptions, and $160 of shipper work. The $1,150 quote sits $123 below that ceiling. I would negotiate the service and its recovery commitments against that headroom.

An extra $100,000 production interruption turns the base saving into a $1,600 loss. I would test replacement capacity against a disruption that leaves several loads needing help at once, because a fallback that handles the average small exception can still fail on the receiving day when the production line and the remaining savings depend on all those movements together.

The roadside task has a legal cost

FMCSA’s October 9 exemption, effective October 7, 2026, through October 7, 2031, permits specified cab-mounted beacons in place of conventional roadside warning devices. Aurora is covered; other qualifying Level 4 carriers must give prior written notice. The conditions restrict operating domains, cargo, equipment, and activation. Beacons must activate within five minutes of a qualifying stop. I would obtain the recovery response time as its own service commitment. Federal conditions.

For genuinely intrastate freight, the checked Texas incorporation and exception rules had not adopted the new relief. Two Texas endpoints alone leave commerce classification unresolved. I would require the applicable legal basis or a compliant roadside arrangement before taking that task out of the price. The incorporation rule and exceptions set that starting position.

FMCSA’s chart lists liability filings and cargo requirements by carrier type. A nonhazardous for-hire property carrier’s public liability filing leaves contractual cargo coverage to verify. The May 2026 FAQ also marks legacy Licensing and Insurance records as historical-only for newer Motus filings. I would obtain current evidence for the contracting carrier and the promised protection for the goods, using the filing requirements and record-system guidance.

A pilot needs a viable service behind it

I would tender the same replenishment job to every bidder: cargo, equipment fit, release time, receiving window, and return movement. Then I would reconcile each acceptance, rejection, invoice, and recovery. A truck waiting outside an incompatible dock has removed a person without completing the purchased task.

Aurora’s second-quarter 2026 letter reports about $2 million of revenue across driverless and operator-supervised loads against $7 million of cost of revenue excluding stock compensation. That is an early mixed operation, with scale and cost-reduction plans still to come. I would assess whether the provider can fund the service as well as whether I like the quote. CFO discussion.

Early losses can coexist with a valuable new route. I would require enough financial backing, notice of service changes, and a backup carrier able to take the difficult remainder. A utilization advantage can make those safeguards less costly over time; the pilot should show how the advantage reaches the shipper.

I would buy repeated, completed freight with accountable handoffs and room for recovery costs. The empty driver’s seat makes the offer possible. The receiving window, the remainder, and the provider’s ability to keep supplying it decide whether I would renew.