As sales of its electric cars have fallen, Tesla and its chief executive, Elon Musk, have sought to convince Wall Street that the company’s future lies not in the grinding business of making and selling cars but in the far more exciting world of artificial intelligence.
In Mr. Musk’s telling, one of Tesla’s main A.I.-based businesses will be driverless taxis, or robotaxis, that can operate pretty much anywhere and in any condition. Tesla is very close to perfecting such vehicles and will easily secure regulatory approval to put them on roads, Mr. Musk said last week on a conference call to discuss the company’s second quarter results.
Mr. Musk’s vision of autonomous vehicles, or A.V.s, is not limited to cars that drive themselves. He has also claimed that individuals who buy Teslas would be able to make money when they are asleep or at work by letting the company use their cars as robotaxis.
The robotaxi service will, Mr. Musk has said, catapult Tesla’s stock market valuation, around $700 billion now, into the trillions of dollars.
But first, a lot will have to go right.
His idea would require major advances in technology and fundamental changes in the way people view cars. The experience of driverless taxi services like Waymo and Cruise in Phoenix, San Francisco and other cities raises questions about when such offerings will become profitable and how much money they will make.
Tesla’s technology will face stiff competition from Waymo, a subsidiary of Alphabet, the parent company of Google; ride-hailing services like Uber and Lyft; and Amazon’s self-driving business Zoox. Carmakers including General Motors, which owns Cruise, are also pursuing autonomous driving, along with Chinese tech and auto companies like Baidu and BYD.
“It’s a super competitive business that’s barely profitable,” said Ross Gerber, chief executive of Gerber Kawasaki Wealth and Investment Management, which owns Tesla and Uber stock.
Tesla, which did not respond to a request for comment, became the world’s most valuable car company because investors believed Mr. Musk could achieve his goal of selling 20 million cars a year, compared with around 1.8 million last year. Toyota, the world’s largest automaker, sold more than 11 million cars in 2023. In the last six months, however, Tesla’s car sales have fallen.
Mr. Musk has recently said that the focus on the company’s sagging car business is misplaced and that investors should buy and hold Tesla stock because of its work on ambitious projects like the robotaxis and a humanoid robot called Optimus.
Some Wall Street analysts and many of Mr. Musk’s ardent fans see the pivot as evidence of his genius. But skepticism of the company’s plans has been growing, especially after the company last week reported a 45 percent drop in second-quarter profit and delayed the unveiling of its robotaxi design to October, from August.
Tesla’s stock price is down about 6 percent since the earnings report.
Mr. Gerber suggested that Mr. Musk is trying to divert attention from the company’s core business. “It’s a convenient reorientation when you’re having a hard time selling cars,” Mr. Gerber said.
Still, some analysts believe that robotaxis and self-driving software could generate significant profits for Tesla.
The entire robotaxi market may someday be worth $5 trillion, Tom Narayan, an analyst at RBC Capital Markets, said. And though he added that “regulators aren’t going to allow one company to have a monopoly,” RBC estimated that the economic potential of Tesla’s robotaxis already accounted for over $400 billion of its stock value.
Tesla offers two types of driver-assistance software for the cars it sells now. One is called Autopilot and the other, a more advanced option, is marketed as Full Self-Driving and costs $99 per month. Some analysts say those systems could also be a huge source of profits.
The company has been working on autonomous driving for years, and Mr. Musk has repeatedly said it was just months away from creating a system that can operate cars flawlessly without human assistance. But Tesla’s two driver-assistance systems, which can accelerate, steer and brake on their own, require drivers to be prepared to retake control at any moment. Some users report that they have to intervene often.
Accidents attributed to the software have prompted numerous lawsuits and an investigation by the National Highway Traffic Safety Administration. The agency has said that the technology does not adequately ensure that drivers remain attentive. In April, it issued a report linking hundreds of accidents and at least 29 fatal crashes to use of Tesla’s driver-assistance systems.
Experts who have studied autonomous driving systems have said that Mr. Musk has chosen a more difficult approach to self-driving cars than other companies. For instance, Tesla uses relatively affordable cameras to detect objects and people rather than investing in more costly ones or Lidar sensors, which use laser pulses and some experts say are more capable.
Tesla is also trying to build robotaxis that would operate in any environment, including places for which the company has not created detailed maps. By contrast, Waymo vehicles operate in places that the company has thoroughly mapped.
“You’re trying to solve the hardest problems,” said Philip Koopman, an associate professor of electrical and computer engineering at Carnegie Mellon University, a leading center of autonomous vehicle research. “I’d be surprised if Tesla could pull off a ‘no kidding’ robotaxi in the next 10 years without sensors and maps.”
Other robotaxi services offer a cautionary tale. Last year G.M. halted its Cruise driverless taxi service after one of its cars hit a pedestrian and dragged her for 20 feet in San Francisco. Cruise recently resumed road testing of its cars in three cities.
Waymo, which currently provides rides to paying customers in San Francisco, Los Angeles and Phoenix, has had many years to improve its technology, which uses Lidar.
Other companies that are building self-driving vehicle businesses “lack Waymo’s unique recipe, honed over 15 years and millions of engineering hours and billions of miles of on-road and simulation driving,” said John Krafcik, a former chief executive of Waymo and a board member of the electric vehicle company Rivian.
Yet some analysts said that they still did not know how Alphabet would make money from Waymo. Will it manage a large fleet of self-driving cars over the long term or try to sell or license Waymo’s technology to ride-hail companies like Uber?
“We don’t know what the business model is,” said Mark Mahaney, who heads Evercore ISI’s internet research team.
And investors don’t know when Waymo will generate profits for Alphabet. The company records Waymo’s revenue and operating profits in a category it refers to as Other Bets, a collection of experimental businesses. Other Bets recorded an operating loss of about $2 billion in the first half of this year, but Alphabet did not disclose how much of that stemmed from Waymo. Mr. Mahaney said he assumed Waymo accounted for a significant chunk of that loss.
A representatives for Waymo declined to answer questions about its plans, strategy and financial performance. Alphabet did not respond to requests for comment.
Mr. Krafcik, who left Waymo in 2022, said he expected the business to be profitable this year in markets like San Francisco, after accounting for the cost of vehicles, maintenance, energy, hardware, insurance and personnel. His definition of profits did not include research and development and some other overhead costs.
In a vote of confidence in Waymo, Alphabet said last week that it would invest another $5 billion in the business.
Another challenge for Tesla will be its plans for persuading car owners to put their vehicles in its robotaxi service — an approach that Mr. Musk recently described as “Airbnb on wheels.”
Services like Turo and Getaround allow people to rent out their personal vehicles. But it’s unclear how many people would lend their cars to strangers. If the number is too large, the profits are likely to be small.
“Say everyone who owns a Tesla can rent it out,” said Christopher Robinson, a senior director at Lux Research who has studied robotaxis. “You’re going to flood the market, and prices are going to come down.”
Mr. Robinson noted that robotaxis, while eliminating drivers, would require plenty of human labor. Cars will need to be cleaned, maintained and repaired. A driverless taxi service would have to employ customer service agents, engineers who remotely monitor cars and technicians who fix and retrieve vehicles that have problems.
“It’s a slightly better taxi,” Mr. Robinson said. “I don’t think it’s as disruptive as people think it is.”