Musk has a vision for Tesla’s Robotaxi. Others can’t see it.
Summary
Carmaker plans to show off its automated vehicle this summer, but tech and regulatory hurdles remain before it can become reality.
Tesla Chief Executive Elon Musk has made clear that he is making robotaxis a centerpiece of his long-term strategy. The hard part will come next.
This summer Tesla plans to show a future model—which is expected to have no steering wheel and pedals—that the automaker can deploy in its own, Uber-like ride-hailing service. Musk last week floated the idea of allowing owners to rent out their robotic Teslas, comparing it to Airbnb.
The strategy amounts to a sizable bet on a yet-to-be-built car underpinned by technology that remains under development. A combination of regulatory ambiguity and technology hurdles make widespread deployment of driverless taxis any time soon a long shot, analysts say.
“Robotaxis are likely years away from mass-market adoption," said RBC Capital analyst Tom Narayan.
It is also a concept that is years beyond its deliverable date: Musk and many tech and auto executives once predicted that robotaxis would be ubiquitous on U.S. roadways by now. But Musk—a self-described optimist who has missed deadlines in the past—is leaning into the autonomy story. And investors are cheering him on.
Today, a complicated, evolving patchwork of rules govern autonomous vehicles. Musk’s plan to deploy self-driving cars on public roads is expected to set up a face-off with federal regulators, which already are looking into the automaker’s advanced driver-assistance technology and how it has been rolled out to customers.
Technological hurdles will take years to overcome, many experts say. Today’s most sophisticated robotaxis require remote human operators to step in when the vehicles encounter situations they cannot handle.
Musk’s robotaxi focus during Tesla’s first-quarter earnings call came amid stagnating sales and price deterioration that had hammered the automaker’s stock so far this year. Investors bid up shares as Musk outlined a tantalizing vision of future revenue growth, at profit margins a regular carmaker could only dream of.
“If somebody doesn’t believe Tesla is going to solve autonomy, I think they should not be an investor in the company," he said. The company didn’t respond to a request for comment for this article.
On Monday, the stock rallied again after Musk won approval from China’s government to roll out Tesla’s most advanced driving technology available, called Full Self-Driving.
Few dispute the growth potential for robotaxis deployed on a broad scale. In 2019, UBS pegged the potential revenue opportunity for robotic cars that can be beckoned like an Uber or cab at $2 trillion globally by 2030, far exceeding that of the entire U.S. auto market.
Tesla last week teased what its ride-hailing app could look like. Musk said an owner could deploy their robotaxi—he referred to Tesla’s planned model as “Cybercab"—to ferry passengers around when it isn’t in use. Musk has said Tesla could split fare revenue with owners.
“The economics of the system are just insanely positive," the CEO said last fall.
‘Safety harness’
No automaker sells a self-driving car that requires zero input from passengers inside. Tesla’s Full Self-Driving controls speed and steering in certain situations, but it still requires drivers to pay attention at all times and doesn’t make cars autonomous.
Tesla relies on data gathered from owners’ use of the system to help develop driverless tech. In 2016, Musk said the company would need to log 6 billion miles of autonomous driving to secure approval from regulators worldwide. Today, Tesla says that owners have driven more than 1 billion miles using its Full Self-Driving feature.
“There’s a huge difference between mountain climbing with a safety harness and mountain climbing without one, and human drivers are currently Tesla’s safety harness," said Bryant Walker Smith, a University of South Carolina law-school associate professor who studies autonomous driving.
Waymo and a few other players are further along than Tesla at deploying Uber-like driverless car fleets in a few U.S. cities. But those services are limited in coverage area, and regulatory pushback has limited their expansion more broadly for now.
Congress has yet to pass a law that would create rules for deployment of autonomous vehicles. That void has left developers to shoehorn their plans into existing motor-safety laws and subjected them to a jumble of state and local rules.
Autonomous-vehicle companies have tried to get a federal nod to put driverless cars on the road and faced setbacks. General Motors’ Cruise division for years tried unsuccessfully to get special approval to deploy vehicles without a steering wheel or other manual controls.
Many states, meanwhile, have adopted a hands-off approach to driverless cars. Tesla executives suggested last week that they would focus on launching Full Self-Driving that doesn’t require driver supervision in limited U.S. areas to start.
Still, both state and federal regulators have cracked down on companies when their vehicles have wound up in serious incidents.
In October, California regulators yanked the autonomous-vehicle permit from Cruise after a single incident in San Francisco. A Cruise car struck a pedestrian who had landed in its path after being hit by another vehicle, and then dragged the person at slow speed as it pulled to the curb. The company later suspended its driverless-car operations.
Federal auto-safety regulators last week tied Tesla’s Autopilot assisted-driving system to crashes that resulted in more than a dozen fatalities and said they were investigating the adequacy of a recall Tesla conducted in December to address safety concerns.
Camera debate
To make its technology a reality, Tesla has bucked industry norms by deploying a system that primarily relies on cameras and artificial intelligence to see and drive on the road. Other driverless-car companies use expensive radars and laser-based sensors that many in the industry consider crucial to help driverless cars navigate.
Some researchers have questioned Musk’s approach. While camera-based systems are cheaper, they can be prone to false negatives that can’t necessarily be addressed through software fixes, said Raj Rajkumar, a professor at Carnegie Mellon who specializes in autonomous vehicle driving.
Musk disagrees. “It is obvious that our solution with a relatively low-cost inference computer and standard cameras can achieve self-driving," he said recently.
Write to Ryan Felton at ryan.felton@wsj.com
