A 1985 article in the New York Times wondered what had happened to the fad of laptop computers, which had (the article claimed) been relatively common in the early 1980s but had since dropped in popularity. "People don't want to lug a computer with them to the beach or on a train to while away hours they would rather spend reading the sports or business section of the newspaper," the Times wrote. A 1980 article about early online news services made a similar point: "you cannot tuck a computer under your arm as you head for the subway."
The recent swirl of speculation around Apple making a car has provided many opportunities for people to make a similar mistake — this time about autonomous cars. When they think about the future of self-driving vehicles a couple of decades from now, too many people envision a world that looks mostly like our world today except that drivers can activate a sort of super-cruise control that takes over the steering wheel. The Economist, for example, cites a research report predicting that the market for cars with self-driving capabilities will never exceed 25 percent of the market.
But the history of the PC revolution illustrates how wrong this way of thinking is. As PCs got cheaper and more powerful, the rest of the world didn't stand still. Ubiquitous PCs made it possible for ordinary consumers to get on the internet. A wide variety of industries — music, newspapers, books, retail stores, maps, cameras — got re-built around digital devices. New services like social media and streaming video were invented that made PCs and smartphones even more useful.
The same point applies to self-driving cars. Cars in 2035 could look as different from cars today as today's iPhone looks from a PC circa 1995. A number of industries — trucking, restaurants, retail stores, delivery services, public transit, real estate — will adapt to the new capabilities of self-driving vehicles, making those vehicles increasingly central to the economy. And people will develop totally new services using self-driving vehicles that are hard to imagine today.
Self-driving cars are likely to be rented, not owned
Probably the biggest difference between today's car market and the self-driving market of the future is that the dominant business model is likely to be Uber-style ride-sharing, not personal car ownership. Right now, most cars are sold to individual families because the need for human drivers makes it too labor-intensive to share vehicles between multiple families. But once drivers become unnecessary, it will seem (and be) wasteful for individuals to own cars that sit idle 90 percent of the time.
And this has big implications for the economics of self-driving cars. For example, people have argued that the low-margin auto industry is a bad place for Apple, a company that has traditionally made high-margin products. As my colleague Matt Yglesias points out, this argument doesn't make much sense because Apple has a long history of building high-margin products in low-margin industries like smartphones and PCs. But it's also true that margins will become a lot less important in a future where cars are mostly sold to ride-sharing services like Uber instead of directly to consumers. Ride-sharing companies may be willing to pay a hefty premium for a car that offers extra reliability, efficiency, or other features.
Self-driving cars will look a lot different from today's cars
Today's cars all look pretty similar: most have seating for four or five passengers, a trunk, an engine powerful enough for freeway driving, and a big enough fuel tank for hundreds of miles of driving. Yet for many people, the primary use of their car is for single-person commuting for no more than 10 or 20 miles.
Self-driving cars will allow greater specialization, and therefore greater efficiency. If you're commuting by yourself, you might be able to hail a 1- or 2-seat vehicle that costs less and gets better mileage. If your trip doesn't require freeway driving, you might get an super-efficient electric car that doesn't go faster than 30 miles per hour. When you do need to transport more people or stuff, you'll be able to get a gasoline-powered minivan or pickup truck at a higher price.
The biggest change, though, may be the rise of vehicles designed for zero people. For example, right now if you order a pizza, it's usually delivered by a human driver in a full-sized car — that's thousands of pounds of steel and glass to deliver a pizza that weighs a few pounds. But once you eliminate the need for a human driver, there's no reason for delivery vehicles to be so big, heavy, and expensive.
Navigating this kind of change will be difficult for conventional car companies, which are organized around the assumption that a car must cost tens of thousands of dollars and have room for several human passengers. Gadget makers like Apple, which are used to operating on a smaller scale and squeezing out every ounce of unnecessary weight and cost, might have an edge.
Self-driving won't just be a luxury feature
The Economist quotes a Boston Consulting group study predicting that self-driving cars will account for just 10 percent of the market in 2035, and that demand for automated driving capabilities in general will top out at 25 percent.
Obviously, it's hard to know when fully self-driving cars will first reach the market — technological and regulatory obstacles could delay its introduction beyond the 10 years Boston Consulting is assuming. But once the technology enters the market, it's likely to reach a lot more than 10 percent of the market in its first decade.
This is because the big technological barriers remaining to full self-driving capabilities are mostly software-related. The sensors needed for self-driving capabilities exist now, and while they're currently expensive, they should be a lot more affordable in a decade. The problem is that we don't yet know how to write software that can gracefully handle all of the emergency situations a car might encounter.
But once suitable software is developed, it should be possible to deploy it widely. It might even be possible to upgrade the software on many semi-autonomous cars to give them full self-driving capabilities. Of course, our roads won't become 100 percent self-driving overnight, because people won't want to immediately junk their conventional cars when self-driving alternatives come on the market. But the convenience and safety advantages should cause people — especially younger riders — to adopt those cars very quickly.
And once self-driving cars become safer than human-driven ones, there will be a serious debate about whether people should be allowed to drive their own cars at all. After all, human-driven cars killed 33,561 people in 2012, many of which were due to drivers who were too distracted, tired or drunk to drive safely. Once a safer alternative is available, people may view that as an unacceptable hazard.
Self-driving cars will reshape other industries
In a couple of decades, self-driving cars are likely to control a big chunk of the conventional car market. But self-driving technology is likely to create new product categories as well.
For example, self-driving vehicles will pose a huge competitive threat to package delivery services like UPS and FedEx and catalog and online retailers that rely on them. Because we send drivers out on daily delivery routes, package delivery times are usually measured in days. But once you dispense with the driver, routes and delivery times can be much shorter. Small, automated vehicles could allow people to order everything from diapers to books in the same way we order pizzas today — and at a fraction of the delivery cost.
It's hard to predict what this market will look like, but it's safe to say that a new generation of retailers will want to buy a lot of self-driving vehicles and that these vehicles will look different from any car on the road today.
Existing car companies will struggle to compete
The Economist warns that Apple will need to "catch up with the established carmakers, which are also busy hiring software talent and which have been introducing ever more sophisticated 'assisted driving' features in their models, such as the ability to park themselves, and to navigate stop-go traffic unaided." But thriving in a software-heavy industry — which is what cars will increasingly become — requires more than just hiring a bunch of programmers.
The experience of legacy media organizations is instructive. Over the last two decades, record labels, newspapers, Hollywood, and other traditional content producers have poured millions of dollars into trying to build internet-based platforms for their content. Yet the market is increasingly dominated by outsiders — YouTube, Netflix, and Amazon Prime for video, iTunes and Pandora for music, startups like Buzzfeed and the Huffington Post for news, and so forth.
Building great software requires a certain kind of culture that Silicon Valley startups have and most other types of companies do not. For example, a big obstacle to making car software more secure and reliable is the fact that today's cars are cobbled together from components supplied by dozens of contractors. These contractors treat their software's source code as a closely-guarded trade secret, which makes it almost impossible for the company manufacturing the car to audit the software looking for bugs or security flaws.
Car companies will need to rethink how they manage their supply chains if they want to avoid producing cars with huge security flaws. In contrast, companies like Apple and Google have long experience developing secure and reliable software and are less likely to make the same mistakes.