Investors in the industrials, energy, and tech sectors have been buzzing about driverless cars' potential for several years now. It's no secret that a lot of money and effort -- a whole lot -- are being put into developing systems capable of safely operating vehicles on public roads, with no human intervention required.

Self-driving vehicles have the potential to transform how humans and goods get moved from place to place. And they could transform much more: If autonomous vehicles become the norm, our thinking around everything from private car ownership to truck transportation and even urban design and planning will be challenged and could become obsolete.

Like any revolutionary technological advance, the advent of driverless-vehicle technology could be a bonanza for investors. Here's an overview of what you need to know about this fast-approaching technology, as well as our best estimates about when autonomous vehicles will become commonplace on the world's roads.

In this detailed run down of everything investors need to know about investing in driverless cars we will cover:

What is driverless technology?

Defined formally, driverless technology is the combination of hardware and software that can automate the task of safely driving a vehicle to its destination, with no human intervention required. The key word here is "automate": If a human has to be on watch, it's not fully driverless technology.

Driverless vehicles are sometimes referred to as "autonomous" or "self-driving"; the terms are interchangeable.

A white Chrysler Pacifica minivan with Waymo markings and visible self-driving sensor hardware is shown on a public road in Chandler, Arizona.

The Google Self-Driving Car Project has matured into a full-fledged subsidiary company called Waymo. Waymo expects to launch an automated taxi service in at least one U.S. city by the end of 2018. Image source: Waymo LLC.

What are the components of driverless technology?

Most of the driverless-vehicle software that's in development is based on cutting-edge artificial-intelligence programming. Among other things, the systems incorporate "machine learning": algorithms that can adjust themselves and improve their effectiveness as more data is acquired.

As you can imagine, driverless vehicles acquire vast amounts of data as they travel and encounter new situations in traffic. Generally, these vehicles share the data they acquire and the lessons they learn with a remote data center, which in turn updates all of the vehicles using the system -- a process called "fleet learning."

On the hardware side, driverless technology consists of sensors, processors, communications systems, and other gear that controls the vehicle.

The sensor suites typically include cameras, radar, and lidar, which uses invisible lasers to measure the vehicle's position relative to nearby objects. In many of the systems currently under development, lidar images are compared with a highly detailed 3D map in order to determine the car's precise location.

Intel (INTC -0.58%) CEO Brian Krzanich has said that he thinks of autonomous vehicles as "data centers on wheels" -- a good way of conceptualizing the amount of computing power required. Intel rival NVIDIA (NVDA 1.38%) has already staked out a sizable position in the early market for driverless-vehicle processing with its Drive PX, which is essentially a mobile, liquid-cooled supercomputer. (We'll discuss Drive PX in more detail below.)

Many driverless systems also incorporate specialized computing hardware, such as the image-processing system-on-a-chip offered by Intel subsidiary Mobileye.

Communications hardware allows vehicles to connect with remote data centers via cellular broadband. In addition, the vehicles need throttle, braking, and steering systems that can be controlled by the computer, as well as redundant backups for safety.

Some current (human-driven) vehicles incorporate hardware that can easily be converted to an autonomous-driving system; this is why models like Ford's Fusion and Lincoln MKZ sedans are popular choices for autonomous-vehicle development efforts.

What are the different levels of driverless-vehicle technology?

The only systems that can truly be said to be "autonomous" or "driverless" are the ones that don't require any human intervention in the task of driving. But the industry recognizes a few different levels within that definition.

Some systems will be fully self-driving, meaning they can handle anything that a skilled human driver could handle. Others, including just about every system currently under development, are fully autonomous -- but only within certain limits.

Those limits can be geographic, or they can be related to driving conditions. For instance, most of the systems currently under development require highly detailed 3D maps, and can only operate in areas that have been mapped. (Such systems are often said to be "geofenced".) Others may be limited by weather conditions: Snow and ice present big challenges to a driverless vehicle, for instance.

When most experts talk about the "levels" of driverless-vehicle technology, they're referring to definitions set several years ago by SAE International, the professional association of automotive engineers that sets many technical standards for the auto industry.

The SAE's levels include categories that fall well short of fully autonomous driving. They're levels of "vehicle automation," ranging from Level 0 (no automation at all) to Level 5, a self-driving system that can match or beat a skilled human driver under all circumstances. A key thing to note: Under the SAE's framework, the level of a given vehicle system is defined by its manufacturer; there's no protocol for third-party testing and certification.

  • Level 0 is no automation, meaning that the vehicle's human driver is responsible for all aspects of what the SAE calls "the dynamic driving task." There may be systems that help the driver in very specific situations, like antilock brakes or even an automated emergency-braking system, but they don't count as "automation" because they don't replace the human for any part of the "dynamic driving task" on an ongoing basis.
  • Level 1 is defined as a driver-assistance system that can provide either steering or acceleration-and-braking control that is ongoing, but only under specific, limited circumstances. An adaptive cruise control would fall into this category: The system controls acceleration and braking to keep the vehicle at a set distance behind another vehicle in highway driving, but -- this is important -- the human is still responsible for all other aspects of driving, and is alert and ready to take over if the system shuts down.
  • Level 2 provides both steering and acceleration-and-braking control, but again only under limited, specific circumstances -- meaning that the human driver has to be paying attention and ready to take over quickly if needed. Tesla (TSLA -3.49%) has said that its current Autopilot system falls into Level 2, as does General Motors' (GM -0.66%) Super Cruise system. (Despite the claims that are occasionally made for both systems, neither is a true "self-driving" or "driverless" system -- yet.)
  • Level 3 is defined by the SAE as "conditional automation." In theory, a Level 3 system can be fully self-driving, but only under very limited conditions, and a human driver needs to be present and ready to take control of the vehicle. In practice, the line between Level 2 and Level 3 is a blurry one. Some automakers have chosen to steer clear of Level 3 because of the challenge of ensuring that the human remains alert (and isn't, for instance, napping while the car drives). Others, no doubt at the urging of their legal departments, have categorized systems that could fit into Level 3 as "Level 2," to emphasize the need for an alert human driver.
  • Level 4 is the beginning of full self-driving. The SAE defines it as "high automation"; in practice, the self-driving systems that are geofenced or weather-limited (or otherwise limited) are referred to as Level 4. Nearly all of the systems currently under development fall into this category.
  • Level 5 is "full automation": no limits and no human driver required. Most experts think that Level 5 systems are probably a decade or more away; they'll most likely evolve from Level 4 systems that have been widely deployed and learning for several years.

Only Level 4 and Level 5 systems are considered fully "driverless." Level 4 systems operate within geographical or other limits; they're the ones that are likely to come to market in the near future. Level 5 systems, which have no limits other than those that would apply to a skilled human driver, are further off.

How big is the potential market for driverless technology?

The market for driverless tech could be huge, in time -- but just how huge is a matter of conjecture right now. A study commissioned by Intel in 2017 predicted that the total economic opportunity created by autonomous vehicles will total $800 billion by 2035, and $7 trillion by 2050. That includes the value of the products and services enabled by fully autonomous vehicles, as well as the time and money saved by the increased freedom of movement.

Those are huge numbers. Here's some near-term context, again from Intel: Krzanich said last year that he expects the addressable market for self-driving vehicle systems -- the pieces that Intel hopes to sell -- to be over $20 billion a year by 2020, and around $70 billion a year by 2030.

A little more detail: In a presentation in November of 2017, General Motors' president Dan Ammann expressed the total addressable market for automated ride-hailing (think Uber or Lyft with self-driving cars) in the U.S. in terms of cost per mile. In GM's view, autonomous-vehicle technology will quickly push a ride-hailing customer's cost down to about $1.50 per mile, and then lower as the technology advances and becomes more widely distributed.

The total addressable market for (human-driven) ride-hailing in the U.S. is about $20 billion per year today, Ammann said. That could rise to $750 billion a year once autonomous-vehicle technology pushes the cost of delivering rides down to $1.00 per mile. At that point, automated ride-hailing could cover about 20% of total passenger-miles traveled. Once the cost falls below $1.00 per mile, those numbers could grow to a $1.6 trillion total market and 75% of passenger-miles traveled.

Naturally, Krzanich and Ammann are focused on the addressable markets applicable to their companies in the near term -- computing hardware and ride-hailing, respectively. But there's so much more: Automated vehicles are expected to substantially lower the costs of shipping and freight deliveries, for instance. Daimler, Tesla, and a slew of start-ups are already working to develop automated heavy trucks, while Ford is planning automated commercial vans for deliveries and related services in urban areas.

There are lots of niches for this technology that are just now beginning to be explored. Domino's Pizza and Ford are doing research that could lead to an automated pizza-delivery vehicle, just one of many possibilities for self-driving vehicles in commercial service. There are also potential markets for autonomous vehicles that won't be on public roads -- in mines or warehouses, for instance.

The takeaway: The potential markets for autonomous-vehicle technology, and businesses built around the technology, are together so vast that they're hard to quantify in any detail right now. But the numbers from Intel's study are probably a solid guide as to the magnitude of the overall opportunity.

Who will be the key players in the driverless-vehicle market?

As I write this in mid-2018, there appear to be two leaders on the verge of breaking away from the pack. But there are several other serious and well-funded driverless-vehicle development efforts underway, and any (or all) of them could become big businesses. Given that the total potential market for this technology is so vast, it seems likely that there will be far more than two "winners" over the next decade or so.

Here are the major players in driverless technology as of now.

The leader: Waymo LLC

Any discussion of the commercial potential of autonomous-vehicle technology has to start with Waymo, the Alphabet (GOOG -0.39%) (GOOGL -0.50%) subsidiary that began almost a decade ago as the Google Self-Driving Car Project. Most observers agree that Waymo probably has the most advanced on-road driverless-vehicle technology.

As of right now, it looks likely that Waymo will be first to market: CEO John Krafcik has said that Waymo plans to launch an automated ride-hailing service in at least one U.S. city before the end of 2018, using vehicles built especially for Waymo by major automakers.

That's not just talk. Waymo has already ordered "up to" 82,000 vehicles from two automakers:

  • Fiat Chrysler Automobiles has agreed to build "up to 62,000" minivans for Waymo, with deliveries expected to begin late in 2018. The vehicles are a version of the Chrysler Pacifica Hybrid minivan that was jointly developed by FCA and Waymo engineers in 2016. FCA has already built about 600 of them for Waymo's test fleet. It's believed that these will be the default option in Waymo's new ride-hailing service.
  • Waymo riders will also be able to select a version of the new all-electric Jaguar I-Pace. Jaguar Land Rover and Waymo announced in March 2018 that Waymo has ordered "up to 20,000" I-Paces, to be delivered over the next two years.

At the March press conference with Jaguar, Krafcik said that Waymo may add vehicles from other automakers to its fleet in the future.

The contender: GM Cruise

GM surprised the world when it purchased a tiny autonomous-vehicle start-up, Cruise Automation, in early 2016. More surprises followed, as GM charged Cruise with building an automated urban ride-hailing service -- and gave it the money, resources, and independence to get the job done properly.

As of mid-2018, GM's driverless-vehicle software is thought to be somewhat behind Waymo's, but probably ahead of most or all other rivals'. But GM has other important advantages. For starters, it has the hardware ready to go -- in the form of the battery-electric Chevrolet Bolt EV, which was designed from the start with urban-taxi duties in mind. And Cruise has largely completed work on a version of the Bolt (called the "GM Cruise") adapted to its autonomous-vehicle technology.

A view of the GM Cruise's front seats and dashboard. The vehicle has no steering wheel or pedals.

The interior of General Motors' self-driving taxi, called the GM Cruise. GM expects to begin mass-producing the Cruise in 2019. Image source: General Motors.

That vehicle will be mass-produced at a GM factory in Michigan, on the Bolt's existing production line. It's ready to go: GM will begin manufacturing "thousands" of self-driving GM Cruises as soon as the software is declared ready to release. GM expects that to happen in 2019. As of right now, it looks like GM will be second to market, after Waymo.

How will GM Cruise go to market? Consider that it recently received a big endorsement when SoftBank Group's high-profile Vision Fund agreed to invest $2.25 billion, taking a 19.6% stake. SoftBank brings more than money to the table, as its portfolio includes significant stakes in several ride-hailing companies, including Uber Technologies and Didi Chuxing.

Why does that matter? Those companies will need self-driving taxis to be sustainably profitable, and GM will soon be ready to produce self-driving taxis by the thousands, so it's pretty clear why SoftBank wanted to bring them together.

Other auto-industry players

Most of the major global automakers and auto-industry suppliers have active driverless-vehicle development programs underway. Aside from GM, a few stand out:

Aptiv (NYSE: APTV), a company formed when giant auto-industry supplier Delphi Automotive split itself in two last year, is a key player in a couple of different alliances working on autonomous-vehicle technology -- and has considerable expertise in-house.

Aptiv includes the former nuTonomy, acquired by Delphi shortly before the split. NuTonomy was a self-driving start-up spun out of robotics research at the Massachusetts Institute of Technology (MIT); its software was thought to be fairly advanced at the time of the acquisition.

Aptiv is working with Mobileye and Intel to develop a self-driving system that will be made available to automakers by the end of 2019. The three are also working with BMW AG on a parallel effort; BMW is aiming to launch a self-driving vehicle by 2021.

Daimler AG is the parent company of luxury brand Mercedes-Benz -- and also the parent company of several heavy-truck and bus manufacturers, including Freightliner, Western Star, and school-bus manufacturer Thomas Built. Daimler was one of the first automakers to invest in self-driving research and development, seeing it as an essential feature on luxury cars, an important technology for long-haul trucking, and possibly applicable to urban bus routes as well.

Mercedes-Benz was the first to bring a hands-off highway traffic-jam driving system to market, back in 2014, beating Tesla's Autopilot by several months. More recently, Daimler has demonstrated its technology in heavy trucks, successfully running an automated convoy of three tractor-trailers in a competition on Germany's Autobahn in 2016.

Ford accelerated its driverless-vehicle effort in early 2017 when it spent $1 billion to take a majority stake in Argo AI, a software start-up founded by two artificial-intelligence experts who were veterans of self-driving programs at Waymo and Uber. Ford has since set up Pittsburgh-based Argo as its center of autonomous-vehicle software expertise, and charged it with delivering a road-ready system by 2021.

Ford hasn't given a great deal of detail about its plans, though it has run some public experiments that offer hints as to its thinking. Right now, it appears that the Blue Oval's first driverless vehicle will be a van intended for commercial fleets, and that Ford expects to begin mass-producing that vehicle in 2021.

Ford is a dominant player in the commercial-fleet markets in the U.S. and Europe, and a significant player in China's. It's a segment where Ford could have a unique set of advantages as the world begins to transition to driverless technology.

The chip giants

NVIDIA has long specialized in processors known as "graphics processing units" (GPUs). It turns out that GPUs are especially well-suited for the intense calculation needs of self-driving vehicle systems. As noted above, NVIDIA has already made significant inroads into this market, supplying automotive-grade processors to several major self-driving development efforts. NVIDIA is nearly certain to be a major player in the first wave of autonomous vehicles, and could well secure an enduring advantage.

Intel, long the dominant player in processing chips, was seen as falling behind its old rival -- until it acquired Mobileye for $15.3 billion last year. Mobileye, an Israeli company specializing in machine-vision processors for automotive applications, had developed relationships with nearly all of the world's automakers and had become the supplier of processors and technology for nearly all advanced driver-assist systems that used cameras.

Wisely, Intel didn't try to integrate Mobileye into its U.S.-based business; instead, it's working to add Intel processors and products to Mobileye's suite of offerings.

Startups to watch

There are several start-ups that could emerge as major driverless-vehicle players in time. They're not public yet, but they could go public at some point. Among them:

  • Zoox, a secretive California start-up founded in 2014, has received at least $250 million in funding and is thought to be working on a self-driving system for ride-hailing. It has recently given a few journalists a look at its technology; it's thought to be quite advanced, possibly in the same ballpark with Waymo and GM Cruise.
  • Velodyne LiDAR is the maker of the lidar units used by nearly all of the autonomous-vehicle development efforts to date. The company's units are very expensive, but it has worked with several automakers to bring the costs down significantly over the last few years. Velodyne could emerge as the preferred supplier of lidar units once the technology goes to market. But there are several efforts to develop (much) lower-cost lidar units, including internal efforts at both Waymo and GM; those efforts could eventually render Velodyne's technology moot.

What about Tesla?

Tesla appeared to be an early leader in self-driving technology when it launched its first advanced driver-assist system, audaciously named "Autopilot," in October of 2014. But things haven't unfolded as expected. Mobileye was a key supplier to Tesla's Autopilot program, but the companies parted ways after a 2016 accident in which a Tesla running on Autopilot hit a tractor-trailer at high speed, killing the car's driver.

Since then, it has appeared that Tesla has had some trouble replacing Mobileye's technology. Recent versions of Autopilot appear to be buggy; there have been several more well-publicized accidents involving Teslas running on Autopilot, and Tesla has recently backed away from its more grandiose claims for the technology.

As of mid-2018, it appears that Tesla's technology is significantly behind both Waymo's and GM Cruise's. All of that could change. But right now, investors looking for exposure to the driverless-vehicle trend are probably better off looking at companies other than Tesla.

What about Uber?

Uber founder Travis Kalanick saw driverless-vehicle technology as essential to reaching sustainable profitability, and made substantial investments in an internal program to develop the technology. That program appeared to make good early progress, but it suffered a significant setback when one of the company's test vehicles struck and killed a pedestrian in March of 2018.

Kalanick's successor as CEO, Dara Khosrowshahi, had already begun backing away from the need to develop the technology in-house when the accident happened, raising the possibility that Uber would work with Toyota or other automakers to develop self-driving taxis. Since then, it appears that he has largely shut down Uber's internal effort.

The SoftBank Vision Fund's recent investment in GM Cruise could be a clue as to Uber's current plan: The fund is also one of Uber's largest shareholders.

What could block the widespread rollout of driverless vehicles?

At this point, it appears that regulation -- or in some cases, the lack thereof -- will be the biggest obstacle to the widespread adoption of autonomous-vehicle technology on public roads. But it's also possible that public qualms about the technology could slow or limit its eventual adoption.

At GM, CEO Mary Barra and other executives have recently begun appending the phrase "pending regulatory approval" to their timing estimates for various self-driving milestones. That's a lesson from experience. GM Cruise has been running its self-driving prototypes on city streets in San Francisco for about two years, mostly without incident. For several months now, GM has been trying to get permission to begin doing the same in New York City, but despite some supportive statements from the city and state governments, it still doesn't have an answer.

That's not a problem limited to New York. Regulators and lawmakers around the world are just beginning to confront the implications of driverless vehicles on public roads. There's no doubt that standards will be established -- in fact, some of that work has already begun in some places. But it will take time, possibly years.

And of course, if regulators or lawmakers in a major automotive market like the U.S., the European Union, or China decide to put obstacles in the way of autonomous vehicles, then all of the optimistic timelines we've seen will have to be reset.

There's another thing that could slow or halt the widespread adoption of autonomous vehicles: lack of public acceptance. If it turns out that most people don't want to ride in driverless vehicles, then this revolution might be over before it gets started.

What could turn the general public against a technology with so much promise? It may seem like nothing could, but a few terrible accidents early on, combined with grandstanding politicians, could do a lot of damage to the technology's image in the public mind.

It might not even take horrific accidents to limit the adoption of self-driving technology -- it may just never quite catch on. Consider the Segway: It's a niche product now, but when it was first introduced, some thought it had the potential to revolutionize urban transportation. Dean Kamen, the Segway's inventor, expected to be selling 10,000 a week within a year of its release.

Of course, that demand never materialized. Six years later, he had sold a grand total of 30,000. The Segway turned out to be well-suited for a few specific tasks, but it never developed widespread appeal.

It is not out of the question that driverless-vehicle technology could suffer a similar fate.

When will this driverless revolution begin to happen?

Here are the timelines that have been announced by the major players working to develop driverless-vehicle technology.

Level 2 or 3 systems for highway driving

  • Honda and Toyota both expect to have highway self-driving systems by 2020.
  • Volvo Cars and giant auto-industry supplier Autoliv expect to bring their jointly developed highway self-driving technology to market by 2021.
  • BMW expects to roll out its highway-driving system in 2021.

Level 4 systems for well-mapped areas

  • Waymo will launch its driverless ride-hailing service in at least one U.S. city by the end of 2018.
  • GM expects to begin deploying its driverless taxis at scale in 2019.
  • Ford expects to begin mass-producing a driverless vehicle, likely a commercial vehicle, in 2021.
  • Audi expects to launch a Level 4 system in 2020 or 2021. It's likely that other Volkswagen AG brands (including VW and Porsche) will bring the technology to market around the same time.

Several others, including Daimler and Aptiv, have backed away from specific deadlines, saying that potential regulation and public concerns about the technology make timing uncertain.

But when will this technology hit critical mass? Most experts think that self-driving taxis will become commonplace in developed-world cities within a decade or so. But full self-driving -- Level 5 -- may be further off.

How to invest in the driverless revolution?

It's a tough question to answer. In part, that's because of the uncertain timeline. But it's also tough to answer because right now, there are no "pure play" stocks, no companies that are primarily about self-driving.

Consider: An investment in Alphabet or General Motors will give some exposure to the trend, but both have major lines of business aside from driverless cars. The same is true of all the other automakers that look likely to thrive in a driverless world. That includes Tesla, whose future will likely be determined by the ongoing demand for electric vehicles and its own ability to master high-volume auto manufacturing, not driverless technology.

What about the suppliers? There are a lot of intriguing reasons to invest in NVIDIA. Most are related to various applications of chips for artificial intelligence, of which self-driving cars are one. Aptiv is another possibility: Like NVIDIA, it's well-positioned to supply key components of self-driving vehicle systems, but unlike NVIDIA, Aptiv also has an advanced software-development effort, thanks to its acquisition of nuTonomy.

Ultimately, you'll have to do some research of your own before you invest. But the information here should be more than enough to get you started -- and to guide you as you evaluate the growing number of companies that hope to be major players in this important emerging technological space.