Since ride-hailing's inception, there has been a prevailing assumption about the business: Once ride-hailing networks replace expensive human drivers with competent self-driving systems, operating costs can be dramatically reduced.
Companies such as Uber and Lyft, which lose hundreds of millions today, could then find a faster path to profitability. For consumers, ride-hailing could become cheaper than conventional ownership.
That's been a key storyline, especially as Uber and Lyft made their initial public offerings this year. But a new study from the Massachusetts Institute of Technology suggests turbulent times remain for these companies in the automated future.
Even with robots behind the wheel, MIT researchers Ashley Nunes and Kristen D. Hernandez conclude that ride-hailing will still be more expensive than conventional ownership on a per-mile basis. Examining the San Francisco market, the researchers found the price for an automated taxi would range between $1.58 and $6.01 per mile vs. $0.72 for the per-mile cost of conventional vehicle ownership.
"When we started going into this work, we found there's a lot of hand-waving," Nunes told Automotive News. "There was a notion that 'All we have to do is remove the driver, assume a reduction in insurance, and there's our great number.' We said, 'Let's hold it up to scrutiny.' It didn't hold up."
It’d be easy to assume the newfound expenses of fleet ownership and maintenance, burdens that independent drivers currently bear, made the numbers worse for Uber and Lyft. Perhaps surprisingly, that's not the key contributor to the disparity.
Human-driven or autonomous, the taxi market will remain inefficient, Nunes says, because operators drive too many miles without a customer paying a fare. In San Francisco, the MIT researchers found a 52 percent utilization rate for ride-hailing and in Beijing, they say vehicles are utilized 56 percent of the time. Even if the vehicles had riders 100 percent of the time, Nunes says they'd still be unable to provide a fare that's comparable to car ownership.
Cramming more paying passengers into the cars might be one solution, something that ride-hailing companies have rolled out in the form of Uber Pool and Lyft Line. But a fundamental problem with that model, Nunes says, is that steeper discounts need to be added to incentivize carpooling. Which cuts into revenue.
Costs associated with running teleoperations, in which remote human operators can intervene when self-driving taxis encounter confusing situations, will also hurt the competitiveness of the ride-hailing fleets.
Ride-hailing companies have made automation a central element of their long-term plans to find profitability.
But Nunes says it's difficult to see that path materializing.
"Their approach with the investment folks has been, 'Trust us, we'll figure this out and it'll be this great utopia where everyone is jumping from an Uber to a scooter to an air taxi," Nunes said. "The future may well be all those things. But you need to demonstrate you can offer the service at a price point that consumers are willing and able to pay. Thus far, they are unable to do so."
— Pete Bigelow