Has the Autonomous Trucking Revolution Arrived?
The trucking trade is at present in a bind. On the one hand, it has a depressing security file, with over 5 thousand trucking-related fatalities yearly. As 60 Minutes highlighted to the world, many small carriers are shady, fly-by-night operations that repeatedly break even essentially the most fundamental guidelines of the street. And now that the Supreme Court has determined that freight brokers may be held chargeable for accidents brought on by carriers they e book—the most important of these brokers, C.H. Robinson, just lately misplaced a whopping $604 million suit—it appears inevitable that the trucking trade goes to have to scrub up its act.
On the opposite hand, as a result of a crackdown on “non-domiciled” business drivers licenses, i.e., these issued to international nationals, and a brand new emphasis on English language proficiency for drivers, trucking additionally has a brand new labor scarcity. This is mirrored in current wage will increase for drivers but additionally in excessive spot charges and tender rejection charges, pushed primarily by a lack of service capability reasonably than heightening demand.
With security issues changing into extra urgent and growing labor market tightness, the attraction of autonomous driving within the trucking trade is rising. As Waymos enter extra cities, it not looks as if a distant dream: if autonomous automobiles can navigate round San Francisco, how onerous can or not it’s to drive them straight alongside a freeway? The startups main the brand new know-how, such because the enterprise capital–backed Aurora, Kodiak, and Waabi, are all assured that their merchandise are prepared for business deployment and scaling, in addition to that the economics of autonomous trucking will quickly be of their favor. They are additionally adamant that autonomous know-how will clear up the protection points within the trucking trade.
Just how far alongside is autonomous trucking? While autonomous drivers are certainly transferring to business deployment, there are nonetheless many unanswered questions round their adoption, the solutions to which can significantly gradual their progress. In any occasion, autonomous deployment will not be taking place so shortly that it might spoil what I’ve beforehand argued in these pages is a massive organizing opportunity for truck drivers. But the larger query issues security: the autonomous car (AV) corporations would really like us to imagine that security is a technical downside, and that their programs’ technical capabilities merely outmatch these of their human equivalents. But in narrowing their focus to technical questions, the AV proponents danger lacking what really creates protected outcomes on the street: a regulatory surroundings supported by social investments.
The state of autonomous trucking
With Waymo working in ten US metropolitan areas, why have we seen a lot slower progress in autonomous trucking than in light-duty automobiles? The first barrier has been regulatory: Waymo piloted its robotaxis after which finally launched its business taxi service in Phoenix as a result of the town and state of Arizona supplied a lax regulatory surroundings, amongst different causes. Trucks, in contrast, essentially transfer between cities and states, and with competing regulatory authorities between completely different locales and no nationwide framework for autonomous deployment, getting permission has been a trouble.
That is all shortly altering, nevertheless: in April, a serious roadblock was lifted when the state of California allowed for the testing of autonomous trucks in preparation for eventual business deployment. Many states, together with Arizona and Texas, have already created pleasant environments for autonomous trucking startups, however California is a key domino to fall, as corporations can now think about automated journeys from the Inland Empire to Fort Worth, a key trucking hall. The Department of Transportation additionally just lately allowed autonomous trucking corporations to make use of cab-mounted warning beacons when their vehicles are stopped on a freeway shoulder, rather than the standard warning triangles positioned by a driver at a distance behind a trailer. The autonomous tech firm IMAMS Technology described the rule about warning triangles because the “one factor standing in [autonomous trucks’] manner from being really driverless.” And California congressman Vince Fong has introduced legislation to override state legal guidelines and create a nationwide framework for autonomous truck deployment, which has been integrated into the Build America 250 Act, the floor transportation invoice at present making its manner by way of Congress.
A second key barrier has been some trepidation across the penalties of failure: if one thing goes mistaken with a truck on the freeway, the outcomes may be disastrous, given the load and velocity of the car. This barrier, too, has largely been overcome, thanks most significantly to the event of sensor stacks—for many corporations, a “sensor fusion” of radar, LiDar (Light Detection and Ranging, a distant sensing gadget that measures options within the surroundings utilizing laser pulses), and digicam programs—which have grown by leaps and bounds in sophistication. Now they’ll additionally profit from the event of imaginative and prescient language fashions (VLMs), which assist each with additional sensory precision but additionally with coping with the persistent downside of “edge circumstances,” i.e., conditions nicely exterior the driving norm. Inference seems at present to be too gradual to actually be useful in real time, however it’s solely a matter of time earlier than these fashions are delivered to bear within the AV house.
The objective with all of those developments is Level 4 autonomy (as outlined by the Society of Automotive Engineers), during which no human driver is current in situ and the autonomous system is absolutely chargeable for driving duties more often than not (however will name for distant help when wanted) and in restricted service zones. No one actually thinks Level 5 autonomy (full autonomy in all conditions with no distant help) will occur any time quickly, however the trade would love to keep away from selecting Level 3, “conditional” autonomy, the place a human driver is there to deal with any conditions that the autonomous system can’t. Level 3 is much less of a technological and security downside for the businesses—certainly, there’s a very good argument to be made that in situ help reasonably than distant help is safer—than an financial one. AV trucking adoption pencils out if labor is displaced; commercially it’s a nonstarter if human beings stay within the cab (certainly, if there even is a cab).
The present leaders in AV long-haul trucking are Aurora and Kodiak, each of which’s CEOs as soon as labored on Google’s self-driving automotive undertaking. Aurora, which was began by Chris Urmson in 2017, just lately started its commercial scaling phase, promising roughly 2 hundred vehicles by the top of the yr and signing an settlement with the auto engineering firm Roush to “upfit” (i.e., add autonomous programs to already constructed tractors) about one thousand vehicles per yr after that. Their vehicles have been doing common runs between Dallas and Houston on I-45 however at the moment are increasing to New Mexico and Arizona, having logged 500,000 driverless miles to date. The “Aurora driver” (their autonomous driving system) is constructed instantly into Volvo’s autonomous vehicles throughout the latter’s personal manufacturing course of, in an uncommon industrial partnership between the 2 corporations. Volvo is projecting $3 billion of annual revenue in simply 5 years from its autonomous phase. Aurora additionally has a dedication from Hirschbach Motor Lines so as to add 5 hundred tractors geared up with the Aurora driver by way of 2027 and 2028. Urmson confidently said on Aurora’s most up-to-date quarterly earnings name: “If you’re not utilizing our stuff within the subsequent 5 years, you’re not going to be aggressive in long-haul.”
Kodiak is a tad behind Aurora by the numbers: they have deployed thirty-five customer-owned automobiles and logged forty thousand hours of paid driverless operations (extra on the particularities of those claims in a second). To my data, the one buyer with deployed automobiles up to now is their “launch associate,” Atlas Energy Solutions, which makes use of Kodiak-driven vehicles to maneuver fracking sand within the Permian Basin. Kodiak additionally has many connections within the protection trade—for instance, they’re working with General Dynamics Land Systems “to create autonomous floor automobiles (AGVs) for protection functions.” Kodiak’s CEO, Don Burnette, touts the extra rugged industrial and protection functions of their know-how as hard testing their automobiles for broader business deployment: the corporate plans for driverless long-haul to launch by the top of the yr.
I interpret Kodiak’s emphasis on “customer-owned” automobiles to be a slight dig at Aurora. While Aurora claims it’s going to scale with a “Driver-as-a-Service” (DaaS) mannequin, during which trucking corporations personal and insure the vehicles and the Aurora driver is offered as a service by way of a subscription price, it up to now has solely deployed beneath a “Transportation-as-a-Service” (TaaS) mannequin, during which Aurora owns the vehicles concerned itself. While TaaS brings in additional earnings, it has a decrease margin than DaaS given the prices of working an precise trucking firm, which in fact lies exterior the core competencies of any autonomous car participant. Aurora’s CFO David Maday revealed that TaaS earnings is within the “$2 plus” per mile vary, whereas its DaaS goal is “$0.85 plus” per mile. The trade common to function a truck in 2025 was $2.34 per mile, and driver compensation accounted for $1.03 of that price. The DaaS proposition to trucking corporations is thus primarily: as an alternative of paying a human driver $1.03 per mile, pay us $0.85 per mile (17 p.c much less) for a safer driver with better fuel efficiency and no hours of service limitations—and who won’t ever go on strike.
Other key gamers right here embody Waabi, which is aiming for one thing approximating Level 5 autonomy, with end-to-end driving capabilities; Bot Auto, which is constructing a bottom-up TaaS service, not as a placeholder for DaaS like Aurora however as an autonomous trucking firm; Gatik, which is uniquely targeted on middle-mile transportation in field vehicles; and Torc, the autonomous program of the truck producer Daimler. Many analysts suppose Torc might in a short time turn into the trade chief, as Daimler has 40 p.c market share amongst Class 8 (i.e., heavy-duty) tractor producers. Grayson Brulte, founding father of The Road to Autonomy, instructed me that he believes that “Daimler’s redundant chassis . . . might turn into an important a part of autonomous trucking. If Daimler opens that as much as everyone, they may have a 90 to 95 p.c market share.”
How shortly will this occur?
Perhaps the 2 most typical refrains that one hears from the autonomous car corporations and their boosters are, first, that “the autonomous revolution is here,” which means that autonomous car adoption is imminent and will probably be fast over the subsequent decade, and second, that we must always welcome this revolution as a result of autonomous driving programs are a lot safer than their human counterparts. Both claims deserve some scrutiny right here, as the fact is messier than the trade portrays.
McKinsey has estimated that 13 percent of trucks will probably be autonomous inside a decade, pushed primarily by labor market tightness. With roughly three million tractor-trailers on the street immediately, that will imply an astonishing 390,000 autonomous vehicles working within the United States in 2035, if the semi fleet measurement remained the identical. With the main autonomous firm ramping as much as one thousand vehicles per yr beginning in 2027, nevertheless, that looks as if a distant chance. Goldman Sachs extra conservatively estimates that AV truck earnings within the US will jump from $16 billion in 2030 to $105 billion in 2035, with about 25,000 autonomous vehicles on American highways in 2030. If AV truck numbers tracked the earnings progress they undertaking, there can be extra like 164,000 autonomous vehicles in 2035—nearer to five.5 p.c of vehicles. This extra modest projection corresponds roughly to Aurora’s own estimate of 170,000 autonomous vehicles by 2035.
Brulte predicts a key inflection level will are available in 2028:
An overwhelming majority of truck OEMs [Original Equipment Manufacturers] are closely investing in it. Carriers and shippers are experimenting with it, as are Walmart, Amazon, FedEx, UPS. I’d say it’s 18 to 24 months till you begin to see this unbelievable enhance within the adoption of autonomous trucking.
Goldman too sees 2028 because the yr when AV trucking will probably be aggressive with conventional trucking: whereas the cost-per-mile for AV trucking is projected to drop from $8.56 in 2025 to $2.03 in 2030, the cost-per-mile for conventional trucking will bounce from $2.55 to $2.84 in that very same interval. (Note that Aurora’s estimate of “$2 plus” per mile earnings for his or her TaaS service places their operation considerably within the purple if Goldman’s price numbers are appropriate.) From one angle, these easy economics matter extra for a know-how like autonomous trucking than for different improvements that require huge organizational and infrastructural modifications to be viable (resembling containerization for ocean freight) or whose final worth lies in revolutionizing demand (such because the jet engine enabling passenger air journey). In these different circumstances, many different issues went into technological adoption than, “is it cheaper?” By distinction, Aurora and Kodiak’s proposition to trucking corporations is just to pay a bit extra up entrance for an upfitted tractor, after which after that roughly 17 p.c much less for a robotic driver who’s by no means going to complain. “Keep every thing else the identical, simply use our driver as an alternative of a human one,” is the message.
However, there are lots of questions but to be resolved that complicate this proposition. For one, there’ll nonetheless be many human beings concerned in ferrying autonomous vehicles on to their preset locations: folks to connect and detach trailers, drive the vehicles round yards as soon as they attain depots, and refuel them alongside the way in which. The corporations themselves declare that their merchandise will create at the very least as many roles as they displace (a curious factor to say whereas promising labor financial savings). Who will probably be chargeable for these prices? The DaaS or the service? And if the latter, how a lot is handed on to the shipper? What about when the truck inevitably breaks down? As one nameless trucking trade skilled put it, “We wish to suppose that [a breakdown is] an rare prevalence, however it isn’t. Currently, should you break down you may sometimes limp to the subsequent exit or at the very least to the aspect of the street. Would an AT have the ability to try this?” And once more, the all necessary query: who’s paying for it whether it is?
Then there are the inevitable lawsuits that can include better autonomous adoption. As Freightwaves analyst Thomas Wasson instructed me, “[the companies haven’t really] been sued but. That’s what they inform me. How do we all know how protected one thing is? We get sued a bunch of instances to get the case legislation. So there’s a bizarre situation right here the place all of the suppliers are ready for different suppliers to get sued, to allow them to take notes and never mess this up.” Insurance prices will probably be adjusted when these fits are determined, and as one other nameless trade skilled claimed, “the insurance coverage corporations run our trade. I imply, they really are extra highly effective than our clients.”
Add up all the prices concerned in these questions and processes sorting themselves out and the AV firm’s financial argument seems to be a lot much less enticing. Trucking is a low-margin trade, the place everybody goes to need quite a lot of certainty that DaaS is certainly cheaper on the entire and in the long term earlier than committing to the autonomous future. The AV corporations would possibly get there quickly, however for the second, what they’re touting as a easy one-to-one swap for a less expensive various is something however. Given all of this, even 5 p.c adoption of Level 4 autonomous vehicles by 2035 seems to be aspirational for the second. As Wasson jogged my memory,
When we see Morgan Stanley and other people speaking about these percentages, these are aspirations. If you’re good like an autonomous firm, you’re going to pay somebody to [produce projections], so there’s at all times a tie-in. These corporations are burning tens to a whole bunch of tens of millions of {dollars} a yr to get this factor off the bottom, so it behooves them to additionally construct consensus.
The de-socialization of security
Again, the enterprise case for autonomous vehicles comes all the way down to bringing the price of working an AV down under that of a standard, human-driven one. With fast enhancements in autonomous know-how, that case might turn into very convincing within the subsequent few years, however it’s at present removed from the elegant proposition that aspiring DaaS corporations would really like it to be.
The security case for AVs can also be introduced in easy phrases however proves far more difficult upon nearer inspection. That case goes one thing like this: human beings are flawed creatures. We get distracted. We want sleep. Sometimes we’ll take a last-second detour on a selected journey to cease by a restaurant whose burgers we like, and typically we’ll drive just a little sooner than we’d must to get house for the night time after an extended day. When autonomous trucking corporations discuss security, it’s this stuff they bear in mind. Replace that imperfect creature with a classy mechanical system, their argument runs, and our highways will probably be a lot safer than they at present are.
These similar creatures, nevertheless, fly giant metallic tubes by way of the sky, and the United States repeatedly sees zero onboard fatalities in business aviation (2025 being a notable exception). How is it that business aviation, certainly a way more technically difficult endeavor than floor transport, is comparatively protected, whereas business trucking is chargeable for greater than 5 thousand deaths per yr? The easy cause is that business aviation is extremely regulated, and the overseeing authorities administration is comparatively well-funded: the Federal Aviation Administration has a funds of $25 billion per yr and employs greater than 4 thousand safety inspectors, whereas the Federal Motor Carrier Safety Administration is working with lower than $1 billion per yr and 350 safety investigators. After the Supreme Court’s Montgomery resolution, which made freight brokers chargeable for accidents involving carriers they work with, there has naturally been better scrutiny of carriers and a want to know who ought to rely as “unsafe.” It is difficult to take action provided that absolutely 94 percent of interstate motor carriers haven’t any security ranking from the FMCSA.
The autonomous car corporations would really like us to imagine that security is a query of particular person capability, when it’s actually one in all regulation and public asset placement. Under good situations, all different issues being equal, it appears simple that we’re at a degree technologically the place autonomous programs might carry out at the very least in addition to the most effective of human drivers. But all different issues are usually not equal within the low margin, cutthroat international stage of business trucking. As however one illustration: final May throughout International Roadcheck, a three-day interval when the Department of Transportation conducts random inspections of roughly fifty thousand business automobiles, about one in 5 vehicles had been positioned out of service for security violations. In different phrases, about twenty p.c of the semis on the street immediately are in such unhealthy form that the DoT thinks they shouldn’t be working. Without altering the incentives round car upkeep and maintenance, why ought to we count on tractors outfitted with autonomous programs to be any extra well-kept in 5 to 10 years, once they have a whole bunch of 1000’s of miles on them, as an alternative of rising glowing new from the OEMs?
Wasson and Brulte each imagine that the AV trucking corporations merely can’t afford to have automobiles out of service, and for the second, whereas all eyes are on them, that appears true sufficient. But with trucking’s extremely thin margins, and with no federal regulator able to validating the protection of American carriers, the trade is just not one that can rigorously preserve the brand new subtle programs that their vehicles will probably be outfitted with. And till it’s, the concept the tech alone will make trucking safer is simply AV advertising.
Progress will not be automated
“We are led to imagine that know-how mechanically confers advantages on society. Technology is represented because the unbiased variable, the causal issue,” wrote Jack Conway, administrative assistant to United Auto Workers president Walter Reuther, for Business Topics in June 1955. “The goal of attributing to inanimate equipment the flexibility to make progress mechanically is evident. If machines can mechanically give us greater residing requirements, extra leisure, and the higher life, then on what grounds do unions make calls for for greater wages and for higher working situations? Why battle to make progress when automation mechanically offers progress?”
Conway was intervening in a fevered discourse going down on the time in regards to the “Second Industrial Revolution” and the inevitable social dislocation that cybernation and automation would convey. Search and change just a few key phrases in his article and it reads like a recent commentary on AI. The similar is true of a lot different writing on automation from the Nineteen Fifties. Though the overlords of know-how are way more hubristic now than they had been then, our issues with technological displacement are nothing new, and neither is the ideological cowl with which the technologists try and blanket them.
American highways are unsafe as a result of trucking trade deregulation within the Nineteen Eighties created a state of affairs during which it’s tremendously simple for inexperienced, unvetted, and shady operators to maneuver freight. The autonomous corporations are proposing to exchange these operators with skilled, vetted, and clear robotic ones, however no technological repair can, by itself, change the broader incentives of the trucking trade. It was a social resolution that received us into this mess, and it should be a social resolution to re-regulate the trucking trade and to spend money on modernizing public transport systems and oversight that will get us out of it. While developments in autonomy unquestionably promise to lighten the burden of human toil and make industrial society safer, to ensure that that promise to be realized, there should be a corresponding social battle that each reverses the errors of the neoliberal interval and interprets productiveness beneficial properties into social profit. Without this battle, we’re merely introducing spinning plates right into a society unprepared to stability them.
