vehicle, or the expense of manufacturing it. Add to that costs such as vigilant cybersecurity to shield the cars and their passengers from being hacked or hijacked, and the duty of meticulous maintenance by highly trained technicians to avoid malfunctions and accidents that could discredit the whole idea of vehicle autonomy. (With supercomputers, sensor arrays, and LIDAR range finders on board, ordinary garage mechanics wouldn’t do, Ramsey says.) Not to mention more mundane costs such as fuel, tires, repair garages, parts, parking, and property taxes on the fleet, equipment, and buildings. The dollar amounts of some of the expenses, such as insurance bills and state vehicle licensing fees, would likely climb upwards based on the high asset values of the cars alone.
“There may not be much financial upside,” Lux Capital partner Shahin Farshchi says. Self-driving taxi services would be operating on extremely slim margins, he says. “I’m talking grocery store margins.” (Lux Capital has a stake in the autonomous car future through its investment in Zoox, a startup developing self-driving vehicles.)
On top of that, Uber and Lyft would be negotiating for a share of any profits with powerful auto industry partners—not with individual drivers who have much less leverage against a big tech company.
“It’s all going to come down to bargaining power,” Farshchi says.
The ride-hailing companies might have a choice among many carmakers to partner with, judging from the global ranks of manufacturers eagerly working on autonomous vehicles, he says.
On the other hand, new startups can, and have, created their own apps to compete wth Uber and Lyft in certain regions. For example, Via competes with them in New York, and both Fare and the non-profit RideAustin are among their competitors in Austin, TX. Carmakers might also work with many ride-hailing partners—and possibly create or acquire such apps of their own.
“Anybody can make an app for hailing cars,” Farshchi says.
Ramsey carried out a thought experiment to illustrate the financial challenges of driverless taxi fleets, compared with conventional cab companies. Such legacy taxi outfits typically buy used rental cars a year or two old, costing about $10,000 or $11,000 each, he says. They run these cars for perhaps 300,000 miles over four or five years’ time, while the cars depreciate by about $2,500 a year.
But a self-driving car might cost as much as ten times the price of a conventional taxicab—perhaps $100,000, Ramsey says. “Your depreciation is enormous,” Ramsey says. “Now it’s $25,000 a year.”
It’s not clear what would happen to the earliest self-driving cars once they’d racked up four or five years of constant commercial use as taxis. Conceivably, a Web-connected autonomous car might reach the end of its useful life faster than a conventional car, because the swift pace of innovation among competitors might make it outdated. Would those used cars be scrapped entirely, or could they be sold to other fleets, or to individuals?
Farshchi raises another key question about the car markets of the future. Will self-driving cars end up being marketed to individual consumers? Or will fleets of autonomous cars lead to the dominance of “mobility as a service” —the sale of on-demand trips that make car ownership unnecessary and obsolete?
“The idea is, people will give up their cars to have car usage, like seeing a video on Netflix,” Ramsey says.
To swing the future in that direction, ride-hailing might have to remain about as affordable for most people as owning a car, or using other options such as traditional taxicabs and public transit. A hybrid system that included both car ownership and mobility services—as well as transit—-could continue to put downward pressure on the price of a ride in one of those costly autonomous cars—-squeezing the margins for high-tech fleet operators. At the same time, carmakers might be under constant pressure to spend on R&D and keep improving their vehicles to maintain a competitive edge.
“The only players who will capture margin are the innovators,” Farshchi says.
The hidden costs of autonomy
Uber and Lyft are now virtual transportation companies that have measured their successes by achieving scale—a classic pattern among Silicon Valley startups. As they attempt to take ride-hailing into a higher realm of technology with self-driving cars, they’re also trying to follow another Silicon Valley business model—software as a service—that creates an alternative to individual ownership.
While software as a service is easy to scale, things can get a bit messier when companies take that model into the realm of physical things. For one thing, each additional driverless car may cost a lot more to provide than access to software for one more customer’s staff, say.
Unexpected things also happen in the physical world. Ramsey gives an example: What if a passenger throws up en route in a driverless taxi?
Or, say the autonomous car gets plastered with mud by a passing construction truck? Who’s going to stand by to clean the car—-and where will they stand with mop and bucket—-so the car can get back out on the road and keep making money?
These are the kinds of new costs that some observers don’t take into account when they enthuse about eliminating drivers, Ramsey says. “There are just a host of things nobody’s talked about,” he says.
The tasks that ride-hailing drivers now carry out, in addition to transporting passengers across town, include keeping their vehicles clean inside and out to
