The Road to Autonomous Trucking: Scale, Feasibility, and Economic Impact
This Week’s Focus: Autonomous Freight Hits the Road
While robotaxis dominate headlines, driverless trucks are quietly making history. In April 2025, Aurora completed the first driverless commercial freight run in Texas, from Dallas to Houston. After 3 million test miles and 10,000 pilot loads, autonomous 18-wheelers are beginning to roll—smoothly and safely—onto U.S. highways. Despite their promise, trucking deployments are lagging behind Waymo’s robotaxis due to slower iteration, tougher regulatory hurdles, and lower public visibility. The journey to full autonomy is still unfolding—and with it, a number of legitimate concerns: Will truckers become obsolete? Can the economics work? Nevertheless, one thing is clear: the era of driverless freight has officially begun.
Recently, there’s been significant attention on robotaxis and self-driving cars, yet we often overlook that, just a few years ago, driverless trucks were expected to arrive before urban autonomous vehicles.
And indeed, they are starting to hit public highways in pilot programs.
In late April 2025, Aurora became the first company to run a commercial freight route in Texas with no human in the cab, hauling frozen goods 200 miles from Dallas to Houston. This milestone comes after years of testing (over 3 million autonomous miles and 10,000 trial loads) to validate safety.
The prospect of 18-wheelers cruising down interstates on autopilot may sound unsettling, but proponents say it could make roads safer and supply chains more efficient. Indeed, early rides in Aurora’s driverless trucks have been described as smooth, courteous and predictable—in other words, “pretty boring,” which is exactly what you want from a highway semi.
Such breakthroughs, however, prompt several questions: How large is the trucking industry? What makes self driving trucks appealing and will truck drivers become obsolete? What are the challenges, and do the economics add up? Finally, why are we seeing self-driving cars before trucks?
So, let’s shift gears and hit the highway.
The Trucking Industry’s Massive Scale
Claiming that road transportation (i.e., trucking) is one of the backbones of any economy is not a cliche, considering it’s responsible for moving the majority of goods on land.
In the U.S., trucks carry about 72.6% of all domestic freight tonnage and account for 80.7% of the nation’s freight bill. Each year, U.S. trucks haul an astonishing 11.46 billion tons of freight—everything from food and pharmaceuticals to lumber and machinery. Economically, the global freight trucking market is valued at around $2.2 trillion, with the U.S. alone contributing roughly $532 billion in revenues. Nearly every sector relies on trucking; for example, over 80–90% of agricultural and wood products are transported by trucks.
This industry also provides millions of jobs. As of 2022, the U.S. trucking industry employed 3.54 million truck drivers and over 8.4 million people in trucking-related jobs (including mechanics, logistics, etc.). Truck driving is one of the most common occupations, and the workforce is sizable, albeit aging (average driver age ~47).
Despite employing millions, trucking companies chronically face driver shortages—an issue that has grown over decades. The American Trucking Associations (ATA) estimated a shortage of about 80,000 drivers in 2021, projected to reach 115,000 by 2025 and potentially 170,000+ by 2030 if trends continue. The shortfall is driven by an exodus of retiring drivers, insufficient new entrants (especially for long-haul jobs), and high turnover. This labor gap means some trucks sit idle, and freight goes undelivered, adding strain on supply chains.
Autonomous trucking technology is often touted as a solution to this problem.
Why Did We Expect to See Self-Driving Trucks before Cars?
The core reason is operational domain and complexity. Long-haul trucks primarily drive on highways, which have no pedestrians, no bicycles, no complex intersections or traffic lights, and generally have more predictable traffic flow. As a Reuters analysis succinctly put it: “long-haul trucks are easier to automate than robotaxis because major highways are simpler environments than bustling city roads.”
People expected fewer variables to make the technical challenge more tractable. Highway driving is largely staying in lane, following traffic, and dealing with occasional merging or lane changes—tasks that advanced driver-assist systems already handle fairly well.
In contrast, urban driving requires AI to interpret a chaos of cyclists cutting across lanes, jaywalkers, unregulated intersections, and so on—an open-ended problem that has proven harder than tech companies anticipated. The relative simplicity of highway driving means it’s easier for developers to geofence the operating area and optimize the system for known routes (for example, I-45 in Texas for Aurora’s pilot).
Indeed, companies are starting with “transfer hub” models: the autonomous system drives the middle, highway portion, and human drivers handle the local streets at each end. By handing off the load at highway-adjacent depots, the AI is exposed only to the easier scenarios. This significantly limits the variables and edge cases the self-driving software must handle, accelerating safe deployment.
There are additional factors that made people believe trucking is a more feasible early use-case for autonomy:
Controlled routes and environment: Trucking often follows fixed routes between warehouses or hubs. This repetition allows HD maps and route-specific tuning to greatly assist the self-driving system. Also, many long-haul routes are in regions with relatively favorable weather (e.g., the U.S. Sun Belt), avoiding rain, snow, and dense urban clutter. Early deployments deliberately stick to fair-weather, low-complexity corridors. For instance, Aurora’s trucks don’t operate in heavy rain or fog; if such conditions arise, the truck is programmed to safely pull over. By picking the low-hanging fruit—dry, flat highway routes in business-friendly states like Texas and Arizona—autonomous trucking firms can roll out faster than those trying to navigate Manhattan or San Francisco.
Regulatory environment: At least initially, regulators have been more permissive with testing autonomous trucks on highways than with robo-taxis in overpopulated cities. States like Texas, Arizona, and Florida actively welcome AV trucking tests on interstates, whereas other cities have imposed extra hurdles or bans on unmanned vehicles on their streets. Notably, until recently, California—home to many AV companies—banned autonomous trucks over 10,000 lbs on public roads (though that is now changing). Highways fall under state and federal jurisdictions that have been steadily updating rules to accommodate driverless freight, whereas city deployment often requires navigating local politics and public opinion. Social acceptance may also be easier to manage on highways: everyday city dwellers aren’t interacting with the vehicle.
Truck-friendly tech integration: Heavy trucks have ample space and power to accommodate the sensors and computing rigs needed for autonomy. A Class 8 tractor can easily bear the weight and electrical load of lidar units, radar, high-performance GPUs, cooling for electronics, etc. In contrast, a consumer car is more constrained. Operators are more willing to pay for expensive hardware if it delivers safety and cost benefits (in fact, most self-driving truck prototypes are recognizable by their sensor “turrets” and arrays mounted on the cab and trailer). The cost of an AV sensor suite may range from around $10,000 to $100,000 per vehicle (for the full lidar-radar-camera setup), but when spread over a commercial truck’s high mileage, it’s more financially sound.
Why are Self-Driving Cars More Common than Self-Driving Trucks in 2025?
Despite the seemingly simpler operational environment for autonomous trucks, self-driving passenger cars—particularly Waymo’s robotaxis in San Francisco and Phoenix—have become more visible and widespread in 2025.
Waymo alone has accumulated over 30 million autonomous miles by early 2025, with hundreds of robotaxis deployed daily across multiple cities. In contrast, autonomous trucking leader Aurora Innovation has logged about 3 million autonomous miles total, and only a small fraction of those were fully driverless.
Several key factors explain this apparent paradox:
Urban Density and Fleet Utilization: Self-driving cars operate predominantly in dense urban environments, enabling high vehicle utilization through short, frequent rides. Services like Waymo can rapidly gather data, refine algorithms, and demonstrate safety across millions of passenger trips, accelerating both technical maturity and regulatory approval. Trucks must accumulate long-haul highway miles, a slower process due to fewer repetitions of complex scenarios, and thus a longer path toward validation.
Regulatory Environment and Public Acceptance: Cities and states have actively collaborated with robotaxi companies to facilitate controlled, limited deployments. Waymo’s extensive testing in Phoenix benefited from proactive local regulation, making robotaxi services relatively commonplace. Conversely, autonomous trucking faces tougher scrutiny due to the potential catastrophic outcomes of high-speed truck accidents, which means regulators and companies proceed more cautiously.
Complex Logistics and Infrastructure Requirements: Autonomous trucks must integrate seamlessly with intricate logistics chains involving loading docks, depots, maintenance hubs, and transfer stations. This complexity requires significant infrastructure adjustments and coordination across multiple stakeholders. Passenger cars face fewer integration hurdles.
Visible Consumer Presence: Finally, passenger vehicles have immediate consumer-facing appeal and visibility. Companies like Waymo gain widespread public exposure, making the technology seem ubiquitous even if geographic coverage remains limited. Trucks, operating largely unseen on long-distance routes and warehouse hubs, lack similar public visibility despite significant progress.
Thus, while autonomous trucks promise substantial economic benefits, robotaxi deployments appear more widespread in 2025 because they offer quicker iteration, clearer regulatory pathways, and higher public visibility—even though trucking autonomy might ultimately offer a simpler technical challenge in the long run.
Crunching the Numbers: Economics and Viability of Driverless Trucks
This brings me to the question of whether autonomous trucking is economically viable?
To approach this, we’ll perform a back-of-the-envelope costs and savings analysis by breaking down the cost structure of operating a truck to see how autonomy changes the equation.
A conventional long-haul truck operating in the U.S. has an average operational cost of around $2.20–$2.30 per mile for for-hire fleets, including fuel, driver wages, benefits, truck payments, maintenance, insurance, etc. For a truck that runs ~100,000 miles a year, that’s roughly $220,000/year in expenses. The major components of cost per mile are typically:
Fuel: (~25–30% of total) e.g. $0.55–$0.65 per mile, though fuel price swings affect this.
Driver compensation: This includes wages and benefits. Driver wages average about $0.72 per mile, plus benefits perhaps ~$0.15, totaling roughly $0.80–$0.85 per mile. That means driver pay accounts for roughly 40–45% of the total operating cost—by far the single largest cost category.
Equipment (lease or depreciation): Truck and trailer payments might be ~$0.33 per mile.
Maintenance & tires: around $0.20 per mile.
Insurance, permits, admin, etc.: Perhaps accumulate to $0.15–$0.20 per mile.
Now, what changes with a self-driving truck? The most obvious is labor—if the truck truly operates without a human on board, the direct driver wage cost (about 40% of total) could be greatly reduced or eliminated. However, there will be new costs: The capital cost of the autonomous system (sensors, computing, and R&D recoupment)—perhaps financed or amortized over miles. Remote operations staff or safety oversight: There may be a remote operator for many trucks, but it still falls under labor costs (albeit much smaller per truck). Increased maintenance or repair costs for the complex sensors and computing equipment. Also, those components have their own lifecycle.
Let’s attempt a numeric scenario. Suppose an autonomous hardware/software package adds an upfront $150,000 to the truck cost. If the truck is used for 5 years covering 120,000 miles per year (which is feasible if it runs more hours per day than a human-driven truck), that’s 600,000 miles life. The amortized cost of the AV system is $150k/600k = $0.25 per mile. Now, perhaps there’s a remote monitoring center where one employee oversees 10 trucks and earns $100,000/year (loaded cost). Per truck that’s $10k/year, and if that truck runs 120k miles, that’s $0.083 per mile for remote supervision. Maintenance for sensors/software might be, say, another $0.05 per mile (just an estimate). So new costs add maybe ~$0.25 + $0.08 + $0.05 = $0.38 per mile.
On the savings side, we remove driver wages $0.72 and benefits $0.15, so ~$0.87 per mile. Fuel might also improve: autonomous driving can be more fuel-efficient (no hard accelerations, optimal cruising). There’s also the possibility of platooning or just smoother driving. Let’s be conservative and assume autonomy improves fuel economy by 5%, saving $0.03 per mile. So total savings of $0.90 per mile. Net savings after adding new costs ($0.38) would be $0.52 per mile.
For a truck running 120,000 miles, that is $62,400 saved per year—a substantial sum. Even if our estimates are off, it aligns with other analyses that find on the order of 20–30% reduction in per-mile costs can be achieved. Indeed, a detailed model by Deloitte projected “30% or more per-mile cost reduction” with autonomous trucking, thanks to lower labor costs, better utilization, and fuel efficiency—more than enough to offset the cost of technology. Deloitte also highlighted that an automated truck can operate almost 24/7 and potentially double the daily miles driven (from ~600 miles with a human to ~1,200 miles). This doubling of productivity means fewer trucks (and fewer trailers and tractors) could carry the same freight, or the same number of trucks could carry more freight, increasing revenue. The capacity of the freight network would expand without a proportional increase in assets, which is economically very powerful.
To illustrate, consider a 1,200-mile trip (e.g., from Texas to California). A human driver legally would take about 3 days to cover that (with mandated rest). A driverless truck could do it in 24–30 hours non-stop (charging or fueling stops aside), delivering in barely a day. That cuts transit time by 2/3, which has value for shippers (especially for perishable goods, supply chain inventory reduction, etc.). It also means one autonomous truck can do the work of perhaps 2–3 separate driver-shifted trucks in terms of turnaround time. The economic value of faster, more predictable deliveries (no driver illness, no hours-of-service delays) could shift more freight to trucks (from rail or other slower modes) and command premium pricing for urgent goods. Shippers benefit from potentially lower costs per mile and faster service, while carriers benefit from needing fewer drivers and gaining more revenue per truck.
From a macro perspective, McKinsey estimates that, once deployed at scale, autonomous trucks could save the U.S. for-hire trucking industry $85–125 billion annually by 2030—mainly through labor savings, fuel efficiency, and greater utilization—translating into higher profits, lower freight rates, or both. They also forecast a large new market for autonomous truck technology, potentially worth ~$400–$600 billion globally by 2035, including the production of new trucks, software services, and supporting infrastructure.
That said, our simple calculation and these projections depend on certain conditions: The technology cost must decrease with volume; trucks must indeed be allowed to run driverless at length; and maintenance and insurance should not eat away all the savings. Insurance is interesting—currently, the human factor is a big part of accident risk. If autonomous trucks prove significantly safer (e.g., no drunk or tired driving, and faster driving reflex), insurance premiums could drop. The industry spends billions on insurance (about 6–7% of costs); a safer AV fleet might reduce accident frequency and severity, which in turn could save money. On the flip side, any early crash by a driverless truck could spike insurance until trust is built.
Another aspect is energy costs: As noted, if trucks become electric, the cost per mile for “fuel” (electricity) might be lower and less volatile than diesel. This could further reduce operating costs—some studies find an electric semi has a lower (~13%) total cost per mile than diesel, if battery costs drop to certain levels. However, electric trucks have a higher purchase price, which is partly offset by cheaper energy and maintenance (electric drivetrains are simpler and energy is cheaper per BTU). Autonomy combined with electrification offers compounded cost benefits but also requires upfront investment in both technologies.
Viability tipping point: The big question for a fleet owner is: When do the savings outweigh the costs and risks? This will likely happen first on longer routes with high volume (higher utilization), and where driver costs and shortages are most acute. For example, a dedicated lane between two logistics hubs 500 miles apart—you could imagine outfitting trucks for that lane and achieving immediate savings by removing the need for an overnight driver or team drivers. On such a lane, the payback period for the autonomous system might be just 1–2 years given tens of thousands saved annually per truck. Once that is demonstrated, scaling up becomes attractive.
In the near term (2025–2027), autonomous trucks will probably run in service with safety drivers or observers and charge a premium (as Aurora is doing under a partnership with Uber Freight and others). They’ll gather data and prove reliability. As confidence and volumes grow, the safety drivers will be removed, and the economics will markedly improve. Eventually, we may see a “transport-as-a-service” model: instead of selling a truck, companies like Aurora might license their self-driving “Driver” to carriers for a per-mile fee. In Aurora’s case, they envision a subscription where the carrier buys trucks from OEMs (like Paccar or Volvo) with the Aurora Driver installed, and then pays Aurora per mile for the software service. If that fee is, hypothetically, $0.30/mile, the carrier will compare that cost to the ~$0.85/mile they’d otherwise pay a human driver (wages/benefits). There’s room for both the AV provider and the carrier to come out ahead, splitting the savings.
Will Autonomous Trucks Replace Truckers? (Impact on Jobs and Operations)
One of the biggest public concerns is whether self-driving trucks will put millions of truckers out of work. It’s a nuanced issue. In the short-to-medium term, autonomous trucks are more likely to reshape truckers’ jobs than outright eliminate them. In fact, the current deployments treat autonomy as a way to augment a strained workforce rather than replace it.
First, recall the severe driver shortage discussed earlier (tens of thousands of open positions)—the industry cannot find enough people willing to do long-haul routes. This means any autonomous capacity can help fill unmet demand. Executives note that many truckers today prefer shorter routes that get them home nightly, rather than exhausting multi-week cross-country tours. Automation could handle the long interstate stretches and allow human drivers to focus on local pickups and deliveries, which improves driver quality of life. “Instead of long over-the-road trips where truckers are gone for weeks, they’ll be able to sleep in their own beds every night,” says Mike Roeth, head of the North American Council for Freight Efficiency. In other words, the job could shift toward more regional driving and supervising autonomous runs, attracting younger drivers who balk at the old lifestyle.
Second, early autonomous trucking models explicitly keep humans in the loop. The prevailing approach is the “transfer hub” model as mentioned above. This means for each autonomous freight run, two drivers (origin and destination) are still involved, but their duty is limited to the first/last 10–50 miles. Those drivers can be home after their short leg, while the driverless system covers the monotonous 500-mile middle stretch. The net effect is that truckers may transition into first-mile/last-mile specialists or oversee multiple short shuttles to feed the autonomous convoy. This reduces total driving hours without eliminating the need for human involvement—especially for tasks like coupling trailers, handling paperwork at pickup, inspecting the vehicle, and securing loads. In fact, even with transfer hubs, it’s estimated that perhaps only three-quarters of total truck-miles could be automated, covering roughly one-third of all loads (the long ones)—the rest (short-haul and multi-stop routes) would still use human drivers.
Third, as autonomy rolls out gradually, there will be new human roles created. Companies will need remote supervisors or “fleet shepherds” who monitor multiple autonomous trucks from a command center, ready to assist or take control remotely if a truck encounters an unusual situation. A single remote operator might oversee 5–10 trucks, stepping in only occasionally. This can be a less taxing job than driving and could even be done from an office. Additionally, demand will grow for technicians and maintenance personnel skilled in AV systems—calibrating sensors, updating software, and repairing autonomous platforms. These positions could be filled by retrained former drivers, offering a path for workers to transition into new careers. Some forecasts also imagine an escort role for drivers in early deployments: for instance, riding in the cab as a safety operator or in a lead vehicle in a convoy, until the technology is proven enough to run fully unmanned.
Initially, during Aurora’s April 2025 launch, there was no one in the cab; a few weeks later, at the truck manufacturer’s request, they temporarily put a human “observer” back in the driver seat as an extra precaution.
Long-term, fully driverless trucks that require zero human input on any leg could eventually handle many routes. If that comes to pass, there would indeed be a reduction in demand for truck drivers for highway runs. However, the mitigating factors mentioned above suggest that we won’t see 3.5 million jobs vanish overnight, if at all.
Conclusion
The road to autonomy will have bumps, but the direction is set.
As the CEO of an autonomous trucking startup noted, “Level 4 trucks on highways are not a question of if, but when.”
The coming years will likely answer that “when” as pilot programs expand.
For now, we can say that the era of driverless trucks has begun, with Texas freeways offering the first glimpses of a new epoch in transportation.
And who knows? Maybe someday we’ll nostalgically tell our grandchildren about the good old days when trucks still had angry drivers honking at us.