Autonomous trucks seen cutting US driver gap as deployments rise 30%

Start with the gap itself, because the size of it is larger than people picture. The American Trucking Associations still puts the US driver deficit near 80,000, and anyone who has tried to book a dry van out of the Midwest in peak season knows what that means on the ground: trucks you booked at nine in the morning get pushed, rates creep, and your delivery appointment slips by a day you cannot afford. Autonomous trucks do not close the whole gap, but a 30% expansion of Level 4 deployments across North America by late 2027 is not a rounding error either. The report is clear that these are no longer demo runs; they are carrying revenue loads on defined corridors, which changes the conversation from maybe to when.

What I find more interesting than the trucks is what sits behind them. The shift the report describes is whole fleets run by cloud software, paired with electric powertrains. That is a different cost structure from a driver in a cab billing per mile plus detention. When a fleet is scheduled by software, the empty backhaul, the silent killer of trucking margins, gets hunted down and filled. The report projects AI platforms will trim empty backhauls by 15%. Let me put a number on that with the assumption stated plainly. Suppose a carrier runs ten thousand loads a month and historically twenty percent come back empty, that is two thousand dead miles a month. Cutting fifteen percent of the empty share recovers three hundred loads of revenue the truck would otherwise have eaten for nothing. Spread across a national fleet and it is real money that eventually shows up as a lower rate on your invoice.

The labour side is where the importer feels it first. Distribution-centre labour costs are projected to drop 20% by 2028 under these platforms. I am sceptical of clean numbers from any forecast, but even half of that is meaningful when you are paying warehouse labour that has climbed every year. A distribution centre that costs you a million dollars a month in labour and trims even ten percent is a hundred thousand back, and that is before you count the accidents a tired driver causes and the insurance that follows them. For a mid-size importer running two or three DCs, that is a six-figure swing on a line you usually cannot move.

The safety point matters more than the spreadsheet. The report notes autonomy cuts tired-driver accidents. Anyone who has run a cross-country lane knows the 3 a.m. reload where the driver has been up too long, and the near-miss nobody logs. Fewer of those is not a headline, it is cargo that arrives undamaged and a claim you do not file. I have paid for a load a drowsy driver put into a guardrail, and the number on that invoice was not small, plus the customer relationship took longer to repair than the trailer did.

The honest caution is that this is corridor by corridor, not everywhere at once. The Level 4 deployments run on defined highways with mapped exits and support, not on the last mile into your customer's loading dock. So the benefit lands on the long-haul trunk, the leg between the port and the regional hub, not on the final delivery a consumer actually sees. If you import through Los Angeles and your goods must reach a Dallas distribution centre, the autonomous lane may carry it most of the way; the last fifty miles is still a human. Build your cost model around the trunk, not around the fantasy of a driverless box at the door, or you will over-promise and under-deliver.

For the importer and exporter, the move is to watch which carriers are actually deploying and to build the relationship now. The rates from software-scheduled, electric, autonomous fleets will undercut the traditional guys, and the early adopters who sign on get the stable capacity during the next squeeze. I tell clients to ask their trucking providers one question: what is your autonomy plan, and on which corridors. The ones with a real answer are the ones you lock multi-year, because their cost curve bends down while the others bend up, and in freight the curve you are on decides the quarter.

The procurement angle is the quiet win. When empty backhauls fall, the carrier can quote you a round-trip rate that used to be impossible, and that rounds your landed cost down in both directions if you ship both ways. Most importers only think about the inbound, but if you also export, the falling backhaul cost is your export freight dropping too. I have clients who pay more for the return than the headhaul simply because the truck would otherwise deadhead, and that distortion is exactly what this technology removes. Ask for the round-trip quote and watch the return lane quietly pay for part of the inbound.

The timing question is the one people get wrong. They hear 2028 and think it is decades off. It is not. Late 2027 is when the deployment expansion lands, and the rate impact shows up in contracts signed in 2026 and 2027. If you are negotiating a three-year trucking agreement this quarter and you ignore the autonomy curve, you will be paying peak human-driven rates while your competitor signed the software fleet at a discount. The window to capture the cheaper lane closes before the trucks are everywhere, because capacity gets committed early to the buyers who showed up first.

Where I still have doubt is worth saying out loud. Electric powertrains on a long-haul truck depend on charging infrastructure that is not built out evenly, and autonomy depends on state laws that differ corridor to corridor. A truck that is legal in Arizona may not be in Illinois. So the 30% deployment number is real but lumpy, and the benefit shows up on the sunbelt corridors first. If your freight lives in the northeast or the Pacific Northwest, you wait longer. I price that lag into the plan rather than assuming the whole country moves at once, because assuming the whole country moves at once is how you miss a delivery date.

The carrier selection criterion should change this year, not in 2028. A traditional carrier with no autonomy plan will keep billing you the human premium and the detention that comes with it, and when the driver gap bites in peak they will push your load like everyone else's. A carrier with live Level 4 lanes can promise a trunk rate that holds because the software does not call in sick. I have moved two clients onto such carriers this year and the stable trunk rate alone covered the switching headache within a single peak season.

Think about what the 20% distribution-centre labour cut does to your network design. If labour is getting cheaper to manage through automation, the old logic of putting a DC near the cheapest labour market weakens, and the logic of putting it near the fastest autonomous trunk strengthens. That is a multi-year site decision, but the importers who start modelling it now will own the better footprints when the rates diverge. I have seen network decisions made on last decade's cost map cost a client seven figures over the lease, and the map is moving again.

Insurance is a sleeper benefit nobody quotes. Fewer tired-driver crashes means lower cargo and casualty claims, and over a few years the loss ratio on an autonomous-heavy fleet should look different from a human-heavy one. If your freight is high-value, ask the carrier for the loss-ratio trend and factor it into the rate, because a cheaper rate with the same claims is not cheaper. I have had a forwarder win on price and then cost me more in claims than the saving was worth, and the pattern repeats with autonomous fleets if you only watch the line rate.

The empty-backhaul math deserves one more pass because it compounds. Suppose the three hundred recovered loads a month from the earlier example sit at an average rate of, say, $1,800, that is $540,000 a month of revenue the carrier previously threw away, and the carrier who recovers it can shave your rate to win your business. On a ten-thousand-load book that recovered capacity is a structural cost advantage, not a promo. I tell clients to ask the carrier how much of their board is autonomous-scheduled, because that number predicts your rate trajectory better than any sales pitch.

The integration with your own systems is the part that decides whether you feel the saving. A carrier running on cloud software can hand you real-time position and a reliable appointment, which lets you tighten your own receiving dock and cut the labour you pay for waiting. The importer who keeps a manual dock and a paper appointment wastes the very efficiency the autonomous trunk just created. I have watched a client's receiving crew shrink by a quarter once the carrier started giving honest arrival windows instead of guesses.

The risk nobody prices is the transition gap. Between now and 2028 the fleet is mixed, part autonomous, part human, and a mixed fleet has mixed failure modes. A software truck and a human truck on the same corridor behave differently in weather and in congestion, and your plan has to cover both. I model the human backup as still necessary through 2028, then revisit, because treating the autonomous lane as fully proven too early is how a weather event becomes a missed delivery.

I will close where the risk is sharpest and the opportunity is clearest. The driver gap of 80,000 is not closing because we trained 80,000 people; it is closing because the truck started driving itself on the lanes that matter most. The importer who treats this as science fiction will keep paying the human premium and the detention that rides with it. The one who builds the carrier relationship and the corridor knowledge now will be on the cheaper, more reliable lane when the next capacity crunch hits, and in freight that is the whole game.

The wage story underneath the gap is worth naming. Driver pay has climbed because the 80,000 shortfall bids up the rate for the humans who remain, and that wage inflation flows straight into your trucking invoice. Autonomy does not replace those drivers overnight, but it takes the top off the bidding war by covering the lanes where nobody wants to drive at 3 a.m., which is exactly where the wage premium hurts most. I have watched a carrier's rate hold flat for a client only after it put autonomous trunks on the worst lanes, because the human drivers could be paid normally on the lanes people actually want.

The customer ESG ask is the part that sneaks into the request for proposal. More buyers now ask how their freight was moved, and an electric autonomous trunk scores points a diesel human lane does not, even when the cost is close. I have had a retailer threaten to move volume over the carbon story alone, and the autonomous electric lane was the answer that kept the account. Treat the green angle as a commercial tool, not a slogan, because the buyer who cares will pay a little more for the cleaner lane and the buyer who does not will still take the lower rate.

Insurance deserves one more pass with numbers. Suppose the tired-driver crash rate on a trunk lane runs at, say, two incidents per million miles and the autonomous lane runs at a fraction of that. On a fleet doing fifty million miles a year, that difference is dozens of claims avoided, and each avoided claim is a deductible, a premium hit, and a customer you did not have to apologise to. The carrier may not pass all of it to you, but the stable one will, because its loss ratio lets it quote a rate that holds. I price the claim avoidance into the carrier choice, not just the line rate.

Your 3PL or broker has to keep up or get cut. A legacy trucking broker sells you capacity it does not control and quotes you the human rate because that is the only lane it knows. In the autonomous age the broker must sit on top of the software fleets and show you the trunk rate that holds, or it is just a more expensive phone call. I have moved two clients off brokers who could not show an autonomous lane, because the saving was real and the broker was the wall between the client and it. Vet your broker the same way you vet the carrier now.

The small importer is not shut out, only later. A shipper with a handful of trucks a month cannot contract a software fleet directly, but the consolidators and 3PLs that aggregate volume will, and the saving will trickle to the small buyer through the programme rate. The move for the small importer is to ask the consolidator which lanes are autonomous-scheduled and to shift the regular trunk volume there, because the aggregated rate will reflect the cheaper lane even if the small buyer never signs the carrier. I have seen a ten-truck-a-month client drop its trunk cost by asking exactly that question.

Measure this quarterly, not yearly. Set a target for the share of your trunk volume on autonomous-scheduled lanes, track the rate trend against a human-only benchmark, and review the carrier mix every ninety days. The lane is moving fast and a plan you set in January is stale by July. I report this to clients as a one-page scorecard with the autonomous share, the rate delta, and the next corridor to shift, because what gets measured is the only part of the freight plan that actually improves.

Port drayage is the next frontier and it touches your container the moment it lands. The same software that runs the long-haul trunk is moving into the dray segment, where the chassis shortage and the queue at the terminal are their own cost. An importer who watches only the ocean rate and the long-haul rate misses the dray, and the dray is where a lot of the recent pain has lived. I ask carriers whether the dray is on the same autonomous platform, because a cheap trunk that hands off to a chaotic dray still costs you the delay.

The truth I keep coming back to is that autonomy is not a magic trick, it is a cost curve that bends down on the lanes that matter, and the importer who reads that curve early rides it, while the one who waits pays the human premium until the market forces the change. Freight rewards the planner who believes the numbers in front of them, and the numbers this week say the trucks are driving themselves on the trunks that feed your DC.

• Ask every trucking provider this quarter for its autonomy plan and deployment corridors, and prioritise the ones with live Level 4 lanes for multi-year contracts.
• Negotiate any 2026-2027 trucking agreement with a rate-review clause so you capture the autonomy-driven discount instead of locking peak human-driven rates.
• Map your port-to-DC trunk lanes against the sunbelt corridors where deployment lands first, and shift volume there wherever it is legal.
• If you both import and export, renegotiate round-trip rates now to bank the 15% empty-backhaul saving on your return freight.
• Track state-by-state autonomy law changes every quarter and flag any corridor that goes live so you can move capacity early.
• Request the carrier's loss-ratio trend and factor it into the rate, because a cheaper line rate with the same claims is not actually cheaper.