At first blush, the red Budweiser truck rolling down the Colorado interstate bound for Colorado Springs looked like any other big rig making a delivery. It was October 2016, and the red-and-white tractor trailer, loaded with beer, had traveled 120 miles.
Welcome to our field guide on trucking. Check out other parts of our deep dive here.
But this delivery was unlike any of the other millions of deliveries that day, or anything the world had seen until then. It was missing something: a driver to command the vehicle.
This semi, engineered by Uber-owned self-driving truck maker Otto, became the first self-driving vehicle to make a commercial shipment. Immediately, the internet began to speculate whether this was a public relations stunt or the dawn of the robo-driving era.
Perhaps the scene came as an inevitable outcome in a battle that has dominated truck stops and robotic engineering labs in recent years. Our world is increasingly dependent on e-commerce, and e-commerce is entirely dependent on trucks. Yet these two things couldn’t be more incompatible: One is a fast-paced, increasingly automated, tech-forward and the other is, well, not that.
These two worlds now are on a sort of collision course, with one side requiring everything to be delivered RIGHT NOW, and the other limited by available manpower and regulations governing how long a person can drive. And while automation may be the future, it’s unclear when and to what scale, in spite of what all of the hype may suggest.
These different forces—the immediate demands of e-commerce and the rise of automation—are perhaps at their most advanced stages in the US, but the implications are global. According to a report on the future of trucking by Deloitte, the US and other developed nations are already on track to realize greater efficiencies in trucking. This could lead to a stagnation of trucking jobs, as shippers will be able to wring more pounds-per-mile from their existing fleets.
At the same time, BRIC countries are rapidly moving ever-larger numbers of people from poverty to prosperity (and also from rural to urban areas), and will require more trucking than before. And given those countries don’t yet have the robust technological infrastructure that exists in places like the US, efficiencies will be harder to come by at first. Until automation catches up and reorders the entire shipping labor market, a trucker in Billings may want to consider moving to Beijing.
There’s an adage “If you bought it, a truck brought it.” But beyond your Zappos order, this war that is quietly taking place in the lane next to you on the highway could have tremendous implications for the labor market overall. By some counts, trucking is the number one job in 28 states in the US, employing a staggering 3.2 million people as delivery and haul drivers. By some tallies it is also among the most deadly of professions. It has also long been riddled with driver health concerns and sexual harassment claims, decades before we had a hashtag for them.
What happens under the hood of the trucking industry could offer lessons to other worlds that could be further impacted by AI, which is, to say, all of them.
So, put on a trucker hat and your favorite Willie Nelson playlist and let’s get to it: your complete guide to the future of trucking.
If there’s any one group that’s going to have to change the most, it’s the American trucking companies. Their basic way of doing business hasn’t changed radically in decades. But, one could argue they also have a huge advantage: They know the world of trucking, have a long-standing infrastructure, and deep client relationships.
It’s worth noting there’s no one single trucking industry. There are four, actually. Here’s how it breaks down:
Parcel Small items, all going to different destinations. Three companies are responsible for more than 50% of this segment’s revenues.
Less than truckload (LTL) Bigger than parcel, but still not big enough to fill an entire truck. About 25 companies make up the lion’s share of the revenue generated in this segment, including parcel carriers that also operate in this market.
Full truckload (FTL) Single containers, usually filled with homogenous items. This $600-billion market breaks into two camps: trucks for hire and private fleets.
Trucks for hire contract with manufacturers and distributors to provide transport services. This segment is still highly fragmented, as it’s not a huge investment to buy a few trucks and start a company. That said, there are some larger trucking concerns (and they’re only getting bigger, as you’ll read below).
Private fleets carry goods made or distributed by their parent company. The segment represents the smaller portion of the FTL market, responsible for around $250 billion in value. Among private carriers, these are the three biggest:
Truck owners are facing two trends which will radically change the industry: consolidation and sharing. Already, trucking companies are combining to offer greater scale and flexibility to their customers (Knight and Swift merged in late 2017 to create a new behemoth, and more deals are expected). “The trend towards larger fleets of more than 100 vehicles will continue to increase,” says the Deloitte report.
But if consolidation threatens the smaller truck-fleet owner or individual owner-operator, the prospect of truck sharing may be a lifeline. Like Uber, but for freight, there are companies (including, of course, Uber) looking to leverage communications technology to match drivers to customers with shipping needs. Earlier this year, trucking company Ryder launched a pilot program of truck-sharing in Atlanta. While the program is designed to make better use of excess capacity, it’s expected that truck sharing will grow—Deloitte estimates 15% of the truck market will be using sharing platforms by 2026.
71% of all freight in the US moved by trucks.
Truckers are staring down the barrel of increased automation but, right now, somewhat paradoxically, they’re in high demand. Right now, in fact, is a great time to get a trucking job: According to the American Trucking Associations, an industry group, the US is currently 50,000 drivers short of what it needs. The rise of e-commerce and a heated economy has led to increased demand for goods, but demography and technology are placing limits on how many trucks are currently on the road.
The larger, more difficult question, is whether it will be good to be a trucker in, say, 10 years. Automation is only progressing and while truck sharing may connect lots of drivers with customers, it’s unclear how lucrative that relationship will be. If Uber’s ride-sharing platform is an example, the savings realized by a sharing economy do not necessarily convert to more money for drivers.
To be a trucker in the US, one needs a commercial drivers license, or CDL. Typically that requires specialized training, like at a trucking school. No college degree is required. Controversially, many truck companies promise drivers big bonuses to remain longer with them, but sometimes that carrot can be used to keep drivers in unsafe jobs (paywall). With demand and turnover so high, companies are desperate to retain drivers and train new people to get behind the wheel.
Different truckers work different ways.
Owner operators There are only a handful of these around today. The driver owns and operates the truck.
Lease drivers The driver leases a truck and pays a monthly fee.
By-the-mile drivers This is the bulk of freight drivers out there. The driver is hired by a trucking company and paid a per-mile wage, but he or she does not own the truck. In the fine print, truckers are often considered subcontractors, which some truck companies have later argued in court, exempts them from many labor and workplace protection laws. Critics of the pay-per-mile model argue that it incentivizes truckers (and companies) to keep rigs in motion and skip that nap or bathroom break.
Hourly drivers These are drivers who companies pay per hour, often with benefits. This is the Walmart or FedEx model. These jobs are considered the top of the trucking totem pole and are very competitive.
Here’s one big problem for the trucking industry: drivers are getting older. According to the Bureau of Labor Statistics, the average US truck driver is 55 years old, which is 10 years older than the average in comparable industries like manufacturing or construction. And as those truckers retire, the younger generation isn’t replacing their numbers. What used to be a romantic, open-road job in the 1970s and ’80s is now more likely to be perceived as lonely and onerous. The truck-driving community has also remained rigidly homogenous, according to the American Trucking Associations. It is overwhelmingly male—only 6% of drivers are female—and it’s two-thirds white.
At the end of 2016, there was one truck for every four loads that needed shipping. In about a year, that ratio has soared to 12 loads per truck.
If current demands on trucking continue, the ATA estimates that the current 50,000-driver shortage will swell to 174,000 by 2026.
And that doesn’t account for replacing drivers who, because of those older demographics, will be retiring in the coming years as well. In the next decade, the ATA estimates, roughly 900,000 new drivers will need to be hired to replace retirees and meet future demand. Automation is looming out there, but the kind of automation that will obviate drivers entirely is still likely to be at least a decade away (see below for the different levels of automation that are in the works). Until technology—and regulations, and legal and liability frameworks—catch up, the increased demand that comes from clicking “add this to my cart” will require a new generation of drivers.
Truckers would argue that another thing is putting the squeeze on the industry: the ELD, or electronic logging device. Mandated in December, 2017, an ELD monitors how often a truck is in use. The purpose of this monitoring is to enforce road-safety rules governing how long a driver can be behind the wheel. (The rules allow 11 hours of driving a day in a 14-hour period, followed by a 10-hour rest.) For decades, drivers would record their activity in paper logs, but those were easily doctored. (The logs have been called “comic books” by drivers.)
Electronic devices don’t allow for any creative accounting, and if a driver runs afoul of the limits, automatic fines are imposed. Drivers say that the regulation works against them, as they are being regulated by the time they spend working, but they’re usually paid by the number of miles driven. A delay at a pickup or dropoff location can eat into their available working time, as once the clock has started, it can’t be paused. Put this in today’s context of a “ship this to me yesterday” expectation that companies like Amazon have acculturated consumers to, and it’s not unreasonable to think that drivers are getting squeezed at both ends.
ELDs are also not cheap, costing between $100 and $600 per vehicle. While larger trucking companies have had some form of electronic monitoring for years for internal purposes, most of the trucking fleet is small-operator owned (and not sitting on big piles of unused cash), so adding new equipment to a fleet can be costly.
The value of the truck-shipping market is $700 billion.
When it comes to self-driving vehicles, most of the world’s attention is on the self-driving car, but trucking will likely get there first. There’s an obvious profit motive to move toward greater autonomy—according to the ATA, 43% of a trucking company’s operational cost is compensation—and a truly autonomous truck would be able to drive nearly 24 hours a day.
But autonomy is not an on/off switch to flip. It’s more like a spectrum, one defined by the Society of Automotive Engineers (SAE) with six increasing levels. Below is the SAE’s language defining the levels, along with our take of what that means in the real world.
Level 0 The human driver does all the driving.
If your vehicle has nothing more than an old-fashioned, speed-maintaining cruise control, your vehicle is in this category.
Level 1 An advanced driver assistance system (ADAS) on the vehicle can sometimes assist the human driver with either steering or braking and accelerating, but not both simultaneously.
Some newer vehicles have a feature like automatic braking or lane-departure assist. If your ride has one of these, it’s at level 1.
Level 2 An advanced driver assistance system (ADAS) on the vehicle can itself actually control both steering plus braking and accelerating simultaneously under some circumstances. The human driver must continue to pay full attention (“monitor the driving environment”) at all times.
More vehicles are now sold with a suite of safety systems which meet this description. This usually includes things like automatic braking, lane guidance, blind-spot collision avoidance, and radar-based cruise control, all working in concert. You could almost take your hands off the wheel in certain circumstances, but it would still be a bad idea (and no car will allow you to, anyway).
Level 3 An automated driving system (ADS) on the vehicle can itself perform all aspects of the driving task under some circumstances. In those circumstances, the human driver must be ready to take back control at any time when the ADS requests the human driver to do so.
This is a little like what commercial aircraft are like today: Pilots take off and land, but the cruising is mostly automated. There are no vehicles currently for sale that are level 3. Truckmakers Freightliner and Peterbilt have demonstrated trucks with level 3 autonomy, but those projects are still in development.
Level 4 An automated driving system (ADS) on the vehicle can itself perform all driving tasks and monitor the driving environment—essentially, do all the driving—in certain circumstances. The human need not pay attention in those circumstances.
Vehicles at level 4 would still have to have things like steering wheels and brake pedals, but it’s likely they’d rarely be used. You know when the police wave you to an auxiliary parking lot that’s just a grass field? That’s the kind of situation where you’d have to step in.
Level 5 An automated driving system (ADS) on the vehicle can do all the driving in all circumstances. The human occupants are just passengers and need never be involved in driving.
This is full-on future territory. Cars and trucks have no steering wheels, nor pedals or any other vehicle controls. Imagine a shipping container with wheels (or a rolling office for passengers), and you’ve got the idea.
There are 3.6 million heavy-duty trucks on US roads.
Automated trucking could mean a mountain of riches for whoever cracks it first. Already, billions have been invested by some of tech’s biggest actors in the efforts. But the players in this space currently have little, if any, exposure to the trucking industry. Can they do a better job than the companies with decades of experience? Or is that new-kid-on-the-block approach precisely what trucking needs? Here are the groups vying for self-driving supremacy.
- Uber Uber Advanced Technology Group had claimed that it had plans to reinvent the future of trucking, but in July, the company announced it was shuttering its division devoted to self-driving trucks. That followed news that a self-driving Uber SUV killed a pedestrian in Arizona, a tragic lesson in the limitations of the technology. The company has launched Uber Freight, an app that connects drivers with deliveries, akin to its longstanding counterpart with cars. Fascinatingly, the app has put a new face on an old issue— driver shortages.
- Tesla Last year, Elon Musk unrolled Semi, a semi-electric semi-truck (semi-semi?) that can go 500 miles between charges with semi-autonomous driving. The big investment in self-driving trucks is part of his broader vision of the future, but there’s a risk that Musk, who’s had a rough year, could be his own worst enemy. (The “Mad Max” memes are already in full throttle.)
- Alphabet The company’s self-driving Waymo cars have received a fair amount of digital ink, but this spring came the news Alphabet is also making a play with self-driving trucks, initially schlepping Google’s cargo. Nearly a decade into their self-driving journey, Waymo announced in May that its self-driving cars will start moving freight for Google’s data centers in Atlanta.
- Peloton Technology A startup that envisions “platooning” trucks that communicate with each other via radio waves and travel in packs. Place your whale metaphors here. This is more like a fancy version of cruise control, so drivers would still be behind the wheel, but serving a different function. By platooning, the company says it can help reduce aerodynamic drag and coordinate vehicle-to-vehicle braking and acceleration.
- Software Developers Someone has to build the bots and that could be a (very lucrative) boon to whoever does it first and best. The following have acknowledged tasking their in-house talent with the quest.
- Bosch, with the University of Amsterdam, has started funding research and development of AI transportation software.
- NVIDIA has announced a partnership with PACCAR, the company that makes Kenworth, Peterbilt and DAF truck lines.
- Intel has partnered with Waymo on its self-driving vehicles.
- Microsoft is in a partnership with the American Center for Mobility, a Michigan-based test center that was originally built by Henry Ford during World War II to manufacture aircraft. The union came with the blessing of lawmakers in the state, many of whom hope that self-driving could help breathe new life into the auto manufacturing infrastructure of yore.
39 billion gallons of diesel fuel are consumed by US trucks in a year.
There are many long-standing companies that may have to completely rethink the way they do business in an AI world:
- Daimler AG (maker of Freightliner) The world’s largest truckmaker has taken the lead in developing autonomous systems. Primarily operating in the US as Freightliner, Daimler has been testing its Level 3 autonomous truck, the Inspiration, on public roads in Nevada, making it the first manufacturer approved to do so.
- Kenworth and Peterbilt The brands, which operate under parent company PACCAR out of Bellevue, Wash., have been around in some form for more than a century. Peterbilt, in particular, has been a rolling test bed of autonomous technology, working with third parties like NVIDIA, TuSimple of China, and Embark to integrate self-driving systems into an 18-wheeler.
- International In addition to long-haul freight vehicles, the company makes everything from snow plows to fire trucks to sanitation rigs. The company’s CEO has said that it plans to have more autonomous trucks on the road than Tesla by 2025.
- Ford This year, the company unveiled a self-driving truck that looked like a cousin of the Rocketeer, the F-Vision Future truck. Perhaps it’s not surprising that engineers said that the design was inspired by Marvel characters. The company has also invested $1 billion in Argo AI, a robotics company with roots in Google and Uber.
- Volvo The Swedish truckmaker (no longer affiliated with its passenger-car namesake) has announced that a fleet of six trucks will begin autonomous operation at a Norwegian mining site by the end of 2019. The trucks will carry limestone across a three-mile route from the mine to a processing facility near a shipping port. In addition, the company has shown off Vera, an electric, Level 5 prototype that lacks even a cab for a driver to sit in.
There have long been regulations for truck drivers and the companies that employ them, but historically they have been focused on labor concerns. Millions have funneled into lobbyist coffers from the trucking industry to try and push against more red tape. Autonomous trucks could either create an opportunity or a catastrophe for it—the legal framework today is still unclear. What’s more, there’s disagreement over what the safety guidelines for driverless vehicles should be. Critics of self-driving vehicles argue that having the cars on public roads is simply not worth the risk. Firms trying to develop driverless technology claim that there’s no replacement for being able to test products in real-life conditions and that the long-term safety gains could be world-altering.
Some states, like Arizona, California, Florida, and Nevada have taken the lead in allowing autonomous testing on private and public roads, but the national framework is patchy at best. While American regulators are stalling, it’s possible that governments in Europe and Asia, could embrace automated trucking. Here’s looking at you, China.
About 4,000 Americans a year die in trucking-related collisions, the vast majority of which involve human error. Of those deaths, 66% are people in passenger vehicles, 17% are truck occupants, and 16% are pedestrians, bicyclists, and motorcyclists.
So, the federal government, among others, has assumed that self-driving trucks are inevitable. But the big question is, what will automation of trucking mean for jobs?
There’s a scenario in which truckers could come out like bank tellers—relatively the same in number, but performing very different functions. There’s another scenario in which they come out like switchboard operators, which for anyone under the age of 35 who doesn’t watch old movies, were the people who physically connected people’s phone calls with wires.
On either end of that spectrum, the future will require tremendous retraining of a huge workforce. And right now, there isn’t much evidence that efforts are underway to take this seriously. This could be where community colleges come in, as a low-cost, accessible way for blue collar workers to retrain.
There could also be unintended winners like gas station attendants. Self-driving trucks may not be able to refill or plug into a charger themselves, let alone perform their own maintenance checks. A job that has been automated out of existence except in two states that mandate attendants (shoutout to Oregon and New Jersey!) could become indispensable.
In 1920, the senior senator from the great state of Washington, Wesley Jones, introduced the Merchant Marine Act (aka the Jones Act), a bill that laid down some rules about shipping within the US (ostensibly for national-security reasons). It said that ships carrying cargo from one American port to another must be US-flagged, built in the US, owned by US citizens, and crewed by US citizens or permanent residents. In other words, like very few ships that exist today. As a result of this legislation, freight that travels up and down the coasts (where a not insignificant number of people live), cannot be carried by ship—even though doing so would be cheaper and better for the environment. And what steps in the gap to move all that material that ships cannot? Trucks.
The combined weight of everything moved by a truck in the US in a year is 10.5 billion pounds.
The gorilla of a company has used several trucking companies as subcontractors, but there are whispers that it is working to create its own private fleet and has already started poaching good drivers. It’s unclear what role automation will play in Amazon’s own fleet, but given Jeff Bezos’s penchant for AI, it’s likely this will be more Space Age than Smokey and the Bandit. (This is the same company that has wanted to deliver goods by drone, after all.)
One could argue that Amazon is a friend of trucking in that more goods are being moved around than ever thanks to the part it’s playing in encouraging people to buy stuff online, but it also could be seen as a complete foe if it obliterates the trucking industry and replaces drivers with bots overnight. As trucking economists have noted, Amazon is also using data to make its supply chain more efficient in ways that many older companies could have implemented decades ago. Amazon is also reducing the emphasis on long haul trucking by building out a network of distribution centers closer to big, urban areas so medium- and short-haul drivers can make shorter (and speedier) deliveries. The Amazon effect on trucking could mean smaller trucks and a new or different role for 18-wheelers.
It’s one thing for an 18-wheel semi to cruise autonomously down the interstate; it’s another for that truck to navigate city streets or a construction site where the main entrance is blocked by a cement truck, and you have to look for the guy in the orange vest to wave you into the side lot.
You can see how automation may run up against certain limits. What this means is that the chain of vehicles used to deliver goods may well have to change. An autonomous semi may pull into a local distribution center, where parcels are loaded onto smaller vehicles still piloted by humans. Eventually, even those vehicles may become autonomous or, if Jeff Bezos has his way, airborne.