Investors have been buzzing about the tremendous potential of self-driving vehicles for several years now. There are good reasons for the buzz: In theory, at least, vehicles that can safely move people and goods from place to place without human drivers could lower costs, increase flexibility, and dramatically reduce the risk of passenger injuries -- all at once.
Looking a bit deeper, we start to see how self-driving technology could be transformative beyond lowering costs and improving safety. If self-driving vehicles become ubiquitous, will people still own cars? Will cities still need parking spaces?
Simply put, the potential for change -- and disruption -- is huge.
As with any potentially transformative technology, the advent of self-driving cars could be immensely profitable for investors who have the knowledge to position themselves in advance. That's what this article is about. We'll start by defining terms and answering some basic questions. Then we'll take a closer look at the size of the potential market, the types of companies that could profit, and the possible risks and pitfalls that these companies (and their investors) might face over time.
What is self-driving technology?
Self-driving technology is a combination of hardware and software that can fully automate the task of safely driving a vehicle from destination to destination. Take note of that word "automate": If a human has to be paying attention, ready to take over, then it's not "full self-driving," no matter what some people might claim.
Is self-driving different from autonomous driving?
No. The terms "self-driving vehicle," "driverless vehicle," and "autonomous vehicle" all mean the same thing: a vehicle that drives itself with no human intervention required.
What are the components of self-driving cars?
At the simplest level, self-driving systems have three types of components: sensors, software, and computing hardware.
- Sensors. Sensor systems include the cameras and radar units that help the system understand its surroundings. Most self-driving systems under development also incorporate lidar, a technology that uses invisible lasers to make precise images of the vehicle's surroundings as it moves. Those images are then compared with a highly detailed 3D map in order to determine the car's exact location.
- Software. Most of the self-driving software under development is based on artificial-intelligence programming, incorporating machine learning algorithms that can adjust themselves and improve the system's effectiveness as more data is acquired.
- Hardware. Self-driving systems typically require a lot of computing power. (Former Intel CEO Brian Krzanich once said he thinks of self-driving vehicles as "data centers on wheels," which gives a good idea of the amount of power required.) Most systems are based on graphical processing units (GPUs) similar to those developed for personal computers. Often, specialized processors, such as the image-processing system-on-a-chip developed by Intel subsidiary Mobileye, are incorporated as well.
All of these categories will offer opportunities for investors as the technology matures.
How many self-driving cars exist now?
This is a critically important point for investors considering this space to understand: The technology has a lot of promise, but it has yet to be perfected. Aside from a small number of vehicles operating within strictly limited areas that have been carefully mapped, fully self-driving cars don't yet exist.
There are many companies working on self-driving cars, there are prototypes undergoing testing (almost always with human monitors ready to take over) on public roads in various parts of the world, and there has been a lot of hype -- but there are no fully self-driving vehicles operating now.
Doesn't Tesla have self-driving cars?
No -- at least not yet. Tesla's Autopilot system is what we call an "advanced driver-assist system" (ADAS), not a self-driving system. Strictly speaking, it's "partial automation," because it requires a human driver to be present, alert, and ready to take over on short notice.
In the jargon of self-driving vehicles, Tesla Autopilot is still an advanced Level 2 system.
What are the different levels of self-driving technology?
As we've seen, the only systems that can be said to be fully "self-driving" or "autonomous" are the ones that fully automate the "driving task," the job of safely driving a vehicle from destination to destination. But there are systems like Tesla Autopilot that automate part of the driving task -- and there are systems in advanced stages of development that aim to provide full automation under certain conditions but not all the time.
To make sense of these distinctions, the industry has adopted a set of standards proposed several years ago by SAE International, the professional association of automotive engineers that has set technical standards for the industry for decades.
The SAE's standards define six levels of vehicle automation, ranging from Level 0 (no automation at all) to Level 5, a true self-driving system that can match or beat a skilled human driver under any circumstances.
|Level 0||Level 1||Level 2||Level 3||Level 4||Level 5|
Level 0: No automation
Level 0 simply means that there's no automation, so the human driver is responsible for everything involved in what the SAE calls "the dynamic driving task." Note that there's a clear distinction between systems that help the human driver for a few moments under very specific circumstances (like an automated emergency-braking system) and systems that replace the human on an ongoing basis. A Level 0 vehicle might have several of the former, but it will have none of the latter.
Level 1: A little automation
Level 1 is defined as a driver-assistance system that can provide either steering control or acceleration-and-braking control that is ongoing but only under specific, limited circumstances. Adaptive cruise-control systems fall into this category: The system controls acceleration and braking on an ongoing basis to keep the vehicle at a set distance behind another vehicle in highway driving. But even when the system is on, the human driver is still responsible for everything else and is alert and ready to take over the full driving task if the system shuts down.
Level 2: A little more automation
Level 2 is similar to Level 1, but the system is able to provide both steering and acceleration-and-braking control under limited circumstances -- in highway driving, for instance. General Motors' Super Cruise highway-driving system is a Level 2 system, as is the current iteration of Tesla's Autopilot. The key thing to remember about Level 2 is that the human driver must be ready to take over on short notice. Generally, Level 2 systems will incorporate some way of making sure that the driver is alert and paying attention.
Level 3: It's complicated
Level 3 is defined by the SAE as "conditional automation." It's best thought of as an incremental step beyond Level 2. With Level 2, a driver must be ready to take over right away if the system encounters a situation that it can't handle. With Level 3, the expectation is that the system can handle all of the driving task as long as it's within its "operational design domain." The human's role is to be a "fallback."
In practice, Level 3 has turned out to be unwieldy, occupying a vague space between ADAS and systems that are fully self-driving under some circumstances. For reasons having as much to do with liability as with technology, many automakers including Ford Motor Company have chosen to steer clear of Level 3.
Level 4: Real self-driving but not full self-driving
Level 4 is the beginning of genuine self-driving. Per the SAE, a Level 4 system provides "high driving automation," with no human intervention required as long as the system is working within its "operational design domain." In other words, it's real honest-to-goodness self-driving but only within limits.
Most of the self-driving systems under development today depend on highly detailed 3D maps to help the vehicle know exactly where it's located, plus or minus a few centimeters. These systems generally use lidar to "map" the vehicle's surroundings from moment to moment, comparing the images generated to a stored 3D map.
Some systems use the lidar-and-map method as a primary method of locating the vehicle, while others use it as a backup. Either way, the vehicle can only operate as self-driving in areas that have been mapped. If it goes off the map, a human must take over.
That's Level 4: It only works under certain circumstances. But if it's within its operational limits, it's full self-driving, with no human intervention required at any time.
Level 5: The real deal
Level 5 is the dream: full-blown, unconditional (that is, no-limits) self-driving. If it can safely drive a vehicle under any conditions that a skilled human driver could handle, under any circumstances, with absolutely no human intervention required (and even with no human in the vehicle), then it's a Level 5 self-driving system. And importantly: If it falls short on any of those fronts, at any time, then it's not a Level 5 system. (And if it's not Level 5, it's not "full self-driving," period.)
How big is the potential market for self-driving cars?
It seems quite possible that the market for self-driving vehicles will be huge in time -- but how huge and in how much time are both matters of conjecture right now. In 2017, a study commissioned by chip giant Intel predicted that the total opportunity created by self-driving vehicles will reach $800 billion by 2035 and could reach $7 trillion by 2050. That includes all the value of the products and services enabled by self-driving vehicles, as well as the time and money saved.
Here's a somewhat more concrete way to think about it. General Motors estimates that right now, ride-hailing (with human drivers) costs passengers about $2.50 per mile. With self-driving taxis, GM has said, the cost would very quickly drop to about $1.50 per mile, and it would drop further over time as the technology scales and matures.
At $1.00 a mile, GM thinks the U.S. market for self-driving taxi services would be about $750 billion and would account for about 20% of total passenger-miles traveled. Once the cost falls below $1.00 per mile, the total market could grow to $1.6 trillion a year in the United States alone, GM believes.
What kinds of companies will profit from self-driving cars?
Lots of companies stand to profit from self-driving vehicles, especially once we start to consider (for instance) automated urban delivery vans or long-haul trucks. In the intermediate term, as self-driving vehicles start to become commonplace, investors will do well to focus on companies that fall into the categories listed below. But note that more categories (and many more companies) will emerge over time.
|Focus of Company||Description||Examples|
|Components||Makers of the sensors, software, and computing hardware required to make a vehicle self-driving||
|Vehicle manufacturers||Automakers that will have the ability to build and deliver complete self-driving cars, trucks, or vans||None yet, but as of mid-2019, General Motors is probably closest.|
|Self-driving fleet operators||Companies that will profit by operating large fleets of self-driving vehicles||
|Maintaining fleets of self-driving vehicles||Self-driving fleet maintenance will likely emerge as a significant business opportunity once the technology is deployed at scale.||
AutoNation and Ford Motor Company have indicated interest in this market.
Key self-driving car stocks to watch
As of mid-2019, investors eyeing the self-driving space are facing one big obstacle: There are no publicly traded pure plays, or companies focused entirely on the self-driving opportunity. But there are a lot of companies involved in developing self-driving vehicles and businesses built around the technology, all of which have the potential to profit in significant ways as the technology comes to market.
|Alphabet (NASDAQ:GOOG) (NASDAQ:GOOGL)||
Alphabet's subsidiary Waymo, the company formed from the Google Self-Driving Car Project, is widely considered to be the leader in the race to deploy self-driving taxis.
|Aptiv (NYSE:APTV)||Once the high-tech arm of legacy auto-industry supplier Delphi Automotive, Aptiv makes sensors and components used in today's ADAS systems and owns Massachusetts-based self-driving start-up nuTonomy.|
|Domino's Pizza (NYSE:DPZ)||The pizza-delivery giant, which has partnered with Ford to test self-driving pizza-delivery vehicles, is one of many already-established companies that could profit as this technology becomes commonplace.|
|Ford Motor Company (NYSE:F)||Ford is a major investor in Pittsburgh-based self-driving start-up Argo AI, one of the leaders in the race to launch the technology, and is aiming to take a significant share of the future market for self-driving commercial vehicles.|
|General Motors (NYSE:GM)||San-Francisco-based GM subsidiary Cruise is among the self-driving leaders, with a large fleet of GM-built self-driving test vehicles deployed in San Francisco and elsewhere.|
|Intel (NASDAQ:INTC)||The chip giant owns Mobileye, the leader in image-processing components for ADAS and a key player in several self-driving efforts.|
|Lyft (NASDAQ:LYFT)||The second-largest U.S. ride-sharing company has allowed several autonomous-vehicle development efforts to use Lyft's network for testing while it works on its own self-driving system.|
|NVIDIA (NASDAQ:NVDA)||The chip giant has staked out a major presence in the race to supply automotive-grade processors for self-driving cars.|
|NXP Semiconductors (NASDAQ:NXPI)||The leading supplier of automotive semiconductors, NXP already has an important presence in the market for ADAS components, including automotive radar systems.|
|Tesla (NASDAQ:TSLA)||The electric-vehicle leader has won significant mindshare with its Autopilot ADAS system, but it has yet to show that it can safely deploy truly self-driving vehicles.|
|Uber Technologies (NYSE:UBER)||Self-driving technology is thought to be key to Uber's eventual profitability, and the company has a well-funded in-house development effort that counts Toyota among its partners.|
|Volkswagen AG (OTC:VWAGY)||VW recently joined Ford as an investor in Argo AI, contributing both cash and engineering resources to the company's development effort.|
A closer look at three potential self-driving leaders
As I write this in mid-2019, analysts are resetting their expectations for the timing of the arrival of self-driving cars on public roads. It has become clear over the last year or so that there is still much work to be done, even by the technological leaders, before self-driving vehicles will be able to operate at scale on public roads.
How much work? It isn't clear at the moment. For understandable reasons, these companies are careful to limit the data they share with the public, as well as the standards they are applying to decide whether their technology is ready for deployment.
But with all that said, a rough pecking order among the companies that appear to have advanced self-driving efforts has started to emerge.
Any discussion of the commercial potential of self-driving cars has to begin with the company that started out in 2009 as the Google Self-Driving Car Project: Waymo LLC. Now structured as a subsidiary of Google parent Alphabet, Waymo appears to be closest to deploying on-road self-driving technology at scale, and it has announced plans to gradually roll out a self-driving taxi service in the United States. A pilot version of the service, called Waymo One, began operation in Phoenix, Arizona, in late 2018.
At least as of right now, Waymo has no plans to build its own vehicles. It expects to work with established automakers to get vehicles that can incorporate its technology. It has already placed at least two orders for such vehicles:
- Fiat Chrysler Automobiles agreed in 2018 to build "up to 62,000" minivans for Waymo. The vehicles are Chrysler Pacifica Hybrid minivans that have been modified to accept Waymo's hardware. It's believed that these will be the default option in Waymo's robot-taxi fleet.
- Tata Motors' luxury subsidiary, Jaguar Land Rover, agreed in March 2018 to deliver "up to 20,000" units of the all-electric Jaguar I-Pace to Waymo over two years. As with the Chrysler minivans, the I-Paces delivered to Waymo will be modified to accept the necessary self-driving hardware.
Waymo has also inked deals with ridesharing company Lyft (to give Lyft riders in Phoenix the option to select a Waymo vehicle), used-car megadealer AutoNation (NYSE:AN) (to provide service and maintenance to Waymo's fleet), and automakers Nissan and Renault (to explore opportunities outside of the United States), among others.
Though it began life as a 40-person San-Francisco-based start-up, Cruise was acquired by General Motors in early 2016. Now structured as a GM subsidiary and closely integrated with General Motors, Cruise has expanded to more than 1,500 employees and made great progress in developing a complete self-driving taxi.
Cruise's vehicle is an extensively reworked version of GM's battery-electric Chevrolet Bolt hatchback. Of note, it's the first self-driving car that can be manufactured at scale on a high-volume assembly line, just like any other mass-produced vehicle. No post-production modification is required.
Cruise has been testing prototype versions of its vehicle on public roads in San Francisco since 2016. It's not the only self-driving company to have accumulated thousands of miles of testing on public roads, but it was the first to focus on testing in a challenging urban environment -- the likely turf of future self-driving taxi services.
In late 2017, GM said that it expected Cruise to begin deployment of its self-driving taxi service before the end of 2019. But in mid-2019, Cruise CEO Dan Ammann stepped back from that expectation:
When you're working on the large scale deployment of mission critical safety systems, the mindset of "move fast and break things" certainly doesn't cut it. With such high stakes, our first deployment needs to be done right and we will only deploy when we can demonstrate that we will have a net positive impact on safety on our roads.
While Cruise is a GM subsidiary and is still controlled by GM, GM no longer owns it outright. Cruise has accepted significant equity investments from SoftBank Group's Vision Fund and rival automaker Honda Motor (NYSE:HMC).
Pittsburgh-based Argo AI was founded in late 2016 by two veterans of self-driving programs at Waymo and Uber Technologies. Shortly after it was founded, Ford acquired a majority stake in the company, saying that it would invest $1 billion over five years -- and that Argo AI would take over Ford's self-driving-software development team. Argo has since acquired a second partner: Argo, Ford, and Volkswagen struck a deal in July 2019 under which VW will acquire a stake in Argo that is equal to Ford's, and VW's self-driving software team will join Argo. The three companies said at the time that Argo will not seek additional automaker partners.
Argo is believed to have made significant progress toward an urban-focused self-driving system suitable for a variety of commercial vehicles, including taxis and delivery vans. As of mid-2019, Argo employed about 500 people, and it was testing its prototype system on public roads in five U.S. cities: Pittsburgh, Miami, Detroit, Palo Alto, and Washington, D.C. It had not yet given a date for deployment, though Ford had said previously that it intends to begin production of a self-driving commercial vehicle in about 2021.
Start-ups to watch
A number of start-up companies could emerge as major players in this space in time. None are public yet, but that could change. Here are two that bear watching:
- Velodyne Lidar is the maker of the lidar units used by nearly all of the self-driving developments under way as of 2019. The company's lidar sensors are expensive, but it has worked with several automakers to lower costs considerably over the last few years. While some well-funded entrants (including Waymo and Cruise) have internal efforts under way to develop lower-cost lidar units of their own, Velodyne still has the potential to be an enduring leader in this space. Investors should be prepared to take a close look if and when the company announces plans for a public offering.
- Zoox is a secretive California-based start-up that is developing a self-driving system aimed at the ridesharing market. It's well funded, and its technology is thought to be quite advanced, possibly on par with Cruise's.
What could slow or stop the rollout of self-driving cars?
Right now, in mid-2019, it still looks likely that regulation (or the lack thereof) will be the main obstacle to the widespread rollout of self-driving technology on public roads. Regulators in the U.S., China, Europe, and elsewhere are just beginning to confront the implications of the technology; it's quite possible that the rush to deploy self-driving taxi services will be held up until regulatory frameworks can be put into place.
Another factor that could slow or halt the rollout of self-driving technology, at least for passenger vehicles (including taxis): public acceptance. If people don't want to ride in self-driving taxis, the race to win that market might be over before it gets underway in earnest.
That might sound unlikely, but consider: One or two grisly accidents involving self-driving taxis plus a few grandstanding politicians could do an awful lot of damage to the technology's image in the public mind.
It might not even take a horrible accident. Investors looking at this space should be mindful of the example of the Segway, which -- at first -- was widely touted as an invention with the potential to revolutionize urban transportation. Inventor Dean Kamen expected to be selling 10,000 Segways a week within a year of its launch.
But as we now know, that demand never materialized. While the Segway has proven to be a useful product in certain small market niches, it never generated the widespread demand that its inventor (and others) expected.
As of right now, it seems unlikely that self-driving technology will suffer a similar fate, but it isn't out of the question.
When will self-driving cars be available?
Tesla and GM already have Level 2 systems for hands-free highway driving available, and several other automakers are preparing to deploy similar systems, including Honda, Toyota, Volvo Cars, and BMW, among others.
But what about Level 4 and Level 5 systems? As recently as early 2019, many observers expected Level 4 self-driving taxis to begin to appear within months, with companies like Waymo and Cruise beginning to deploy at scale by 2020 or thereabouts. Other companies had announced similarly ambitious timelines, including:
- Ford has said it plans to begin mass production of a Level 4 self-driving commercial vehicle in 2021.
- VW subsidiary Audi has said it expects to launch a Level 4 system in 2020 or 2021.
But as of mid-2019, those expectations are being reset as new technological roadblocks have become clear. As noted above, in July of 2019, Cruise CEO Dan Ammann said that the company wouldn't meet its previously announced goal of launching a self-driving taxi by the end of 2019. Both Ammann and Waymo CEO John Krafcik, who has also been hedging expectations, have emphasized that their systems won't launch until rigorous safety standards are met. Other companies, including Aptiv and German auto giant Daimler AG, have also backed away from previously announced deadlines over the past year.
That's the right position for these companies to take, of course. But the secrecy that surrounds their (and other) self-driving development efforts makes it hard for investors to gauge when self-driving vehicles will be deployed at scale.
Add in the regulatory concerns, and it's hard to be more specific than "In the not-too-distant future but not soon."
Final thoughts about investing in self-driving cars
With all of this in mind, how should investors approach the potentially huge opportunity here? It's a tough question to answer, in part because of the uncertain timeline and in part because there are no public companies that are entirely focused on self-driving as of this moment.
An investment in (for instance) Alphabet or General Motors will give some exposure to the opportunity (and maybe the chance to get shares of Waymo or Cruise should either be spun off), but both have other lines of business that will deliver the vast majority of their revenues for at least the next several years. Even Tesla, which has long been near the center of self-driving investment hype, will likely rise or fall on future demand for electric vehicles (and its ability to profitably mass produce them), with its self-driving technology a secondary factor.
Some of the suppliers, notably Aptiv, could deliver more direct exposure to the self-driving market. But Aptiv isn't in a position to begin building self-driving vehicles on its own -- it's a company that supplies automakers. At least for the time being, its fortunes are likely to rise and fall with those of the overall auto industry, and autos are a cyclical business not a high-growth one.
Ultimately, you'll have to do more research on your own before making a significant investment in the self-driving trend, and that research should include keeping an eye on the smaller companies that aren't yet public. But the information here should be enough to get you started and to guide you as you evaluate the emerging players in this potentially transformative space.