Autonomous Truck Cost Parity Could Arrive in the US by 2028, Kearney and MIT Estimate
Electric Vehicles Thomas Wasson 420 reads
Kearney and MIT project US driverless truck costs could meet conventional rates in 2028, but the forecast depends on fleets scaling to 40,000 trucks by 2035.

The cost case for autonomous long-haul trucking has a potential US tipping point: Kearney and the MIT Mobility Initiative estimate driverless trucks could match conventional trucks at about $2.60 per mile in 2028. Their model starts from a wide gap, however. In 2026, it puts autonomous operation at $4 per mile, compared with $2.50 for a truck with drivers.
That projected crossover matters to fleets weighing when driverless technology might fit their operations. But the date is conditional, not a near-term cost advantage. Commercial deployment remains limited and concentrated in a small number of places, even as Aurora has logged more than 500,000 driverless miles on public roads. For buyers, the important question is whether the model’s assumptions about fleet growth and operating costs materialize.
Labor costs shape the regional timeline
Kearney and MIT forecast US cost parity first, ahead of Europe by two years and the Middle East by four. The difference in their model comes down largely to driver expense: the report estimates a fully loaded US long-haul driver costs about $10,000 a month, including roughly 30% for benefits. That figure is about twice the European cost and four times the Middle East cost.
The report says drivers account for about half the per-mile cost of a US truck. Replacing that expense creates a large potential saving, but a virtual driver does not make the rest of the operation free. The cost estimate also has to account for the vehicle, the autonomous system and the infrastructure needed to support driverless service. The model’s $4-per-mile estimate in 2026 shows how far those costs would still have to fall before the labor savings translate into parity.
Conventional trucking costs provide a useful reference point. The American Transportation Research Institute’s 2026 update to its operational cost analysis put the industry average at $2.336 per mile in 2025, a record and a 3.4% increase over 2024. Kearney’s model begins slightly higher, at $2.50 per mile in 2026, then raises the conventional truck figure with inflation to $3.50 by 2035.
Assumptions explain the gap between estimates
The Kearney model uses a $200,000 Class 8 truck with a five-year useful life. It assumes the truck runs 185,000 miles annually, with two drivers each covering 92,500 miles. These details matter because miles driven and the period over which a vehicle is used affect how its costs are spread across each mile.
Other estimates show how much the starting point can vary. Aurora’s Fort Worth-to-Phoenix case study put its virtual driver cost at $1.84 per mile, compared with $1.99 for a human driver. Goldman Sachs Research also projected US cost competitiveness in 2028, but from a much higher starting figure: it estimated autonomous trucking at about $8.56 per mile in 2025, excluding corporate operating expenses, declining to about $2.03 by 2035.
Those figures shouldn’t be read as interchangeable quotes for the same service. They come from different models and assumptions, and the Aurora figure describes a particular route case. Kearney and MIT say their forecast draws on more than 25 interviews with autonomous vehicle developers, truck makers, carriers, brokers and investors. The shared 2028 milestone in the Kearney and Goldman estimates is notable; the different cost paths show that the precise economics remain sensitive to what each analysis includes.
Scale is central to the 2028 forecast
Kearney expects driverless costs to fall about 49%, reaching roughly $2 per mile by 2035. Much of that reduction comes from spreading the cost of developing the virtual driver across a larger fleet. The model assumes $750 million in global research and development spending each year, amortized over five years, and a worldwide fleet of 40,000 autonomous trucks by 2035. It also assumes remote operators will move from supporting one truck in every 10 to one in every 20.
The fleet projection is a key hinge in the forecast. If autonomous fleets reach just 10,000 trucks by 2035, the report says US cost parity could be delayed until as late as 2032. A smaller fleet means development costs are shared across fewer vehicles, weakening one of the main sources of the projected per-mile decline. That makes adoption pace a direct part of the cost calculation, not just a measure of how quickly the technology spreads.
Deployment conditions also differ by route. The report identifies Texas, Arizona, Florida and Georgia as places where fully driverless operation is permitted, while many interstate and cross-border routes still require a human. It also points to unanswered questions around how a virtual driver is certified and how responsibility is divided among a carrier, truck maker and virtual-driver provider after an incident. Those open issues affect where companies can operate and how they structure a service, so fleet growth depends on more than vehicle economics.
Four business models compete for driver value
Kearney groups the market into four approaches: an automaker-led platform, represented by Volvo’s VNL Autonomous; a purpose-built truck, such as Humble’s; an autonomous carrier, such as Bot Auto; and a virtual-driver provider, such as Waabi and Aurora. Each approach places the vehicle, carrier operation and autonomous software in a different part of the business. The report values the driver cost that these models seek to replace at $1.26 per mile.
Private truckload fleets are one of three priority customer segments identified in the report. That segment is valued at about $450 billion, with an estimated $180 billion in driver costs. For-hire truckload and less-than-truckload operations are the other two segments.
Each model will need to show how its share of the savings compares with the cost and complexity it takes on. Kearney’s estimate makes 2028 a useful marker for fleet planning, but its own sensitivity to vehicle counts argues for watching deployment scale and per-mile operating data closely. The business model that turns driver savings into repeatable service economics may matter as much as the year the forecast’s lines cross.