The 85-Cent Mile: Who’s Really Paying for Your Cheap Driverless Freight?

Somebody is losing close to $1 billion a year to haul your freight for less than half what it costs you today. Not by accident, by design. And it’s the best deal you’ll see in this business for years, right up until the day it isn’t.

You’ve seen this movie before. Remember when a cross-town ride cost less than the gas to drive yourself there? That price was never real. Venture capital money covered the gap, one underpriced trip at a time, until it ran out and the fare you’d built a habit around doubled overnight. Autonomous freight is in that exact window right now. If you run a fleet, you’re the one being handed the cheap ride. The only question is what happens when the meter resets.

The Number That Doesn’t Add Up

Start with the pitch, and it’s a good one. The benchmark cost to put a truck on the road ran about $2.26 a mile last year, close to a dollar of that in driver wages and benefits. Then this spring a war in the Persian Gulf sent diesel to $5.64 a gallon, up more than 60% in a year, and that benchmark started climbing toward $2.50 and past it. Against a cost that’s rising, Aurora, the one driverless operator far enough along to talk price in public, is quoting its service at a flat $0.85 a mile.

So, read the next line. The same company quoting $0.85 told its investors it expects just $14 to $16 million in revenue for all of 2026, while burning $190 to $220 million in cash every quarter to get there. Its loss in the first quarter alone ran $244 million. The $0.85 isn’t what the service costs. It’s what it costs to win your business while the technology scales. And Aurora is the only player whose books you can read. Kodiak, public now too, raised another $100 million this spring at a discount steep enough to knock its stock down a third in a single day. Every serious name in this race is spending far more than it earns, on purpose, to be standing there first when the math turns.

None of that means the trucks are fake. They’re real, the freight is moving, the safety record clean. You’re being handed freight below cost on an investor’s dime, and for a fleet willing to pilot now, that’s an opportunity with a clock attached. The question isn’t whether the deal is good today, because it plainly is, but rather what happens when the subsidy ends.

Three Ways to Buy a Truck that Drives Itself

Autonomous trucking isn’t one business model. It’s three, and the difference decides who carries the risk. You can buy the move and not the truck, paying per mile while the technology company owns the equipment and eats the loss; that’s the way Aurora and Waabi sell it. You can buy the truck and rent the driver, the way Kodiak’s first customers do, which puts the asset on your books but keeps you off the hook for someone else’s pricing. Or you can wait for it to come built-in at the factory, the way Plus is building it with the truck makers. Which one you sign up with matters less than knowing which one the industry is driving toward. And it isn’t the first one.

The Cheap Service is a Bridge, Not a Destination

Listen to these companies talk to investors instead of customers and the long game is consistent. Nobody’s endgame is renting out company-owned robot trucks by the mile forever. The destination is autonomy built into the truck on the assembly line, and the leading developers are already integrating their systems into production trucks with Volvo, PACCAR, and International. Goldman Sachs expects a driverless truck to cost less per mile than a human-driven one somewhere in the 2028 to 2033 window, subsidy or no subsidy. That’s when the training wheels come off, and everything between now and then is a preview rate.

And if you own the equipment, the bill isn’t subtle. On top of a roughly $200,000 tractor, the autonomy package adds $125,000 to $150,000 today, a premium Goldman expects to fall toward $35,000 to $40,000 by 2035. Which is exactly why a developer would rather sell you cheap miles now and the truck later.

Where the Money Actually Comes From

Here’s where the diesel spike actually matters. The robot truck burns the same fuel you do, so cheaper diesel was never the autonomy pitch. The case doesn’t even rest on firing drivers. It rests on the truck never sleeping. A human is capped at 11 hours, so an expensive asset sits parked half its day, and when every mile costs more to run, idle hours are the most expensive thing on your lot. Take the clock off and you double the hours that truck earns. Aurora’s trucks for Werner already average more than 4,000 miles a week, which no human can match. That utilization gain, not the wage line, is the engine, and the labor savings everyone fixates on are the smaller half of the story.

The longer-range case follows from there. McKinsey puts the cost-per-mile cut on long hauls at 42%, and underneath it sits a driver shortage the American Trucking Associations, the industry’s trade group, pegs near 80,000 today and expects to reach about 160,000 by 2030. That’s the demand pull that keeps capital flowing while the unit economics stay underwater.

It’s worth being clear about what these trucks still can’t do. Weather is still a wall; it constrained Aurora’s operations roughly 40% of the time through 2025, and the map is still mostly Sun Belt, on validated lanes. This isn’t a go-anywhere product yet, and to their credit, the developers are the first to say so.

When Does It Make Sense for an LTL Carrier?

Now the part almost nobody frames correctly. On its face, less-than-truckload (LTL) looks like the worst fit in freight. It’s hub-and-spoke, lots of touches, messy city pickup-and-delivery, shorter regional legs. Everything running today is tuned for the opposite, the long straight point-to-point haul.

But LTL isn’t one job. It’s three, local pickup-and-delivery, sortation at the terminal, and linehaul between terminals. That middle leg, dock-to-dock, is the same highway problem these trucks already handle well. And the handoff at the terminal, the same one I argued is quietly bleeding out hub-to-hub truckload economics, is a cost LTL already pays by design. The terminals are there regardless, the handoff already sunk. LTL may be the one corner of freight where an autonomous linehaul leg drops in without adding a single new touch.

So, the question for an LTL carrier isn’t “should we automate the network?” It’s narrower, and it’s the same test any operator should run on any lane. Take one route and ask three things. Does it sit on a corridor the trucks are validated to run? Does it move at high frequency, especially overnight, where running around the clock is worth the most? And is it where your driver shortage bites hardest? When those three line up, it’s worth a pilot. The pickup-and-delivery ends stay human for a long time yet.

And the pressure isn’t hypothetical. LTL is the one freight sector still making real money, around an 11.6% operating margin last year while truckload ran in the red, and it’s shedding linehaul drivers faster than anyone. Long-haul LTL employment fell by 5,000 in a single month heading into 2026, near levels last seen when a major carrier collapsed. The sector with the most room to invest is the one feeling the labor squeeze hardest. That’s the ground this technology takes root in.

The Takeaway

Driverless freight at $0.85 a mile is the real thing and the temporary thing at the same time. Take the cheap miles while they’re on the table, there’s no medal for paying more. But plan like you know what you know. The price is subsidized, the service model is a bridge, and the truck you eventually buy will be priced by a market, not by an investor underwriting your early adoption. The fleets that come out ahead won’t be the ones that landed the lowest rate in 2026. They’ll be the ones that understood, while the sale was on, exactly who was paying for it.