Tesla has a huge incentive to deploy self-driving tech. But is the world ready?

Aug. 9, 2019 6 AM PT

Elon Musk frames his company’s aggressive push into driverless car technology as a moral imperative. Along with sustainable electric transportation, he views autonomy as a core element of Tesla Inc.’s “fundamental goodness.”

Humans will be freed of the tedium of driving, he told Wall Street last year. Millions of lives will be saved.

There is another incentive for Musk to put driverless cars on the road, though. The day he does that, hundreds of millions of dollars’ worth of stored-up revenue become eligible for a trip straight to Tesla’s perpetually stressed bottom line.

All Tesla cars built since late 2016 are equipped with sensors and other hardware that allow them to function without a human driver at the wheel, according to the company. Since then, buyers of Tesla Models S, X, and 3 have been able to pay $3,000 to $6,000 to eventually get what Musk calls Full Self-Driving technology, or FSD. (The price will soon rise to $7,000.)

Tesla has sold approximately 500,000 cars over that period. The electric-vehicle website Electrek has estimated that 40% of customers choose the FSD option. Owners who haven’t can buy it when available, albeit at a higher price.

Tesla cars will just need new lines of computer code beamed into the car to go full robot when the software is ready, the company says. Musk is aiming to make that happen by the end of the year.

But is Tesla anywhere close to ready with fully driverless technology? And what would that even mean?

The answers concern many in the auto industry, and not just for reasons of competitiveness. Auto executives worry that premature deployment of driverless technology would result in crashes, injuries and deaths and rile up politicians and regulators. It could also damage public trust in the technology — which surveys show is already low — and set the field back by years, they fear.

On Tesla’s website, where FSD is offered for sale, the company says that automatic driving will be available on city streets by the end of the year. FSD will recognize stop signs and traffic lights, it says. And Musk is aiming to release a self-parking feature by the end of the year. The technology, originally scheduled for a May release, would allow a car to drive itself around a parking lot, find an empty spot and park.

Tesla does not say how a car equipped with FSD might respond to a child crossing the street chasing a ball, or whether it would swerve over a double yellow line to avoid a bicyclist. It is “edge cases” such as these that Waymo — the autonomous-driving unit of Google parent Alphabet, and the acknowledged industry leader — and others say are taking them so much time to perfect.

Asked to provide a timeline for Tesla’s transition to totally driverless cars, a Tesla spokeswoman pointed to the Autopilot section of its website where the company discusses future use of Tesla cars without driver supervision. As for what Tesla means by “full” self-driving, she offered the following quote from a recent presentation by Musk:

“There’s three steps to self-driving. There’s being feature complete, then there’s being feature complete to the degree where we think that the person in the car does not need to pay attention. And then there’s being, at a reliability level, where we also convince regulators that that is true.”

The lack of clarity on FSD’s capabilities and timeline concerns the National Safety Council, a nonprofit health and safety advocacy group. “Most people don’t understand the technology that’s already in their cars,” said council Vice President Kelly Nantel. “It’s confusing to drivers. When you call something Full Self-Driving or Autopilot (Tesla’s driver-assist technology) you give the impression that the vehicle has capabilities it doesn’t have.”

Moving the millions collected from FSD customers onto Tesla’s bottom line could be enough to ensure a profit in the fourth quarter, which Musk told stock analysts last month he’s “pretty confident” Tesla can do. That would be huge for a company that is struggling to prove it’s not a perpetual money loser. Tesla hasn’t produced an annual profit since its founding in 2003. In the second quarter of this year, Tesla sold a record 95,000 cars but lost $389 million.

As of the second quarter, Tesla listed $1.18 billion in deferred revenue. The company doesn’t break out how much of that is for FSD. But if Electrek’s estimate is correct, then more than 200,000 car buyers have paid somewhere from $600 million to more than $1 billion for the option.

“The technical definition of ‘full’ means I can get into this car, fall asleep, and [the car] can take me from downtown Manhattan to the mountains of Maine in the wintertime.”

— Bryant Walker Smith, University of South Carolina

There are other advantages to flipping on the self-driving switch. Musk has said the “take rate” — the number of owners who will want the option — will jump when Full Self-Driving goes live.

Of course, Musk is known for his aggressive claims of actual and future Tesla technology. He boasted in April, for instance, that the company would have the first of a fleet of robotaxis on the road next year.

Some of the company’s claims have landed it in hot water. Consumer Reports found that a new Navigate feature that allowed unassisted lane changes “doesn’t work very well” and “could create potential safety risks.” Bloomberg reported Wednesday that the National Highway Traffic Safety Administration sent Tesla a cease-and-desist letter in October focused on a Tesla website claim that its newest car, the Model 3, has “the lowest probability of injury of all cars the safety agency has tested.” The safety agency also alerted the Federal Trade Commission to Tesla’s claims.

In addition, the highway safety agency hit Tesla with subpoenas for information on several crashes involving Tesla’s Autopilot.

Musk’s own statements indicate Tesla’s Full Self-Drive technology remains in the early stages of development. Just several months ago, in a conference call with stock analysts, Musk talked about the difficulty of programming a car to reliably read traffic lights. “It’s easy to recognize stop signs. Traffic lights and intersections will be the next really tricky one,” he said. That is one of the capabilities the company says are “coming later this year.”

In the same call, Musk said navigating multilevel parking lots at a shopping mall “with lots of traffic and pedestrians … is where things get real tricky.” But the Tesla website claims, “your parked car will come find you anywhere in a parking lot. Really.”

On Twitter in June, Musk said that “we’ve been working on curbs” and that “excessive shades of gray” are making concrete seams hard to identify.

Even Tesla’s instruction on current technology seems problematic. A few weeks ago, Tesla’s head of manufacturing, Jerome Guillen, said that the “No. 1” reason for owner visits to Tesla service centers is to learn “how to use Autopilot.” Musk himself has been shown several times driving with his hands off the wheel while using Autopilot — most famously in a “60 Minutes” segment last year — even though the Tesla driver’s handbook warns against that.

In the robotaxi fleet Musk envisions, Tesla owners would send the driverless cars out to pick up passengers Uber-style and make money on the side for themselves and for Tesla — so much money that Tesla cars would appreciate in value, Musk said, a claim that prompted widespread skepticism.

A big hurdle to any self-driving push is that driverless cars are not yet legal in most locations. Musk said regulations are so tough in Europe that he’s not even thinking of deploying there. Florida and Arizona, both with ultra-light regulations on driverless technology, are two likely locations for an initial rollout, although Tesla’s Autopilot has been involved in at least two fatal Florida crashes, and in Arizona an Uber driverless test car with an inattentive test driver hit and killed a woman walking her bicycle across the street.

Such cars are technically legal in California, but companies that deploy them must undergo a permit process more complicated and time-consuming than the requirements in more liberal self-drive states. Driverless cars are being tested in California but have not been commercially deployed or made available for sale.

Tesla’s aggressive stance flies in the face of the go-slower approach of all other major companies. Waymo, Ford, General Motors’ Cruise and others have lengthened their development timelines and say that safe deployment of driverless cars is taking longer than expected.

In January, carmakers, technology companies, insurance firms, advocates for the elderly and disabled, safety groups, the association of state governors and the U.S. Chamber of Commerce formed a group called Partners for Automated Vehicle Education, or PAVE, to provide “clear, factual information to the public and policymakers” about driverless cars. Tesla and Uber were not invited.

PAVE’s Twitter feed could be read as taking on Tesla, if not by name. “It is damaging to public discussion about advanced vehicle technologies — and potentially unsafe — to refer to vehicles now available for sale to the public using inaccurate terms,” PAVE tweeted in April. “This includes terms such as “fully automated,” “full self-driving,” “fully autonomous,” “auto pilot” or “driverless,” which can create an inaccurate impression of vehicle capabilities that can put drivers and other road users at risk.”

Bryant Walker Smith, a law professor at the University of South Carolina who specializes in autonomous driving, agrees.

“The technical definition of ‘full’ means I can get into this car, fall asleep, and [the car] can take me from downtown Manhattan to the mountains of Maine in the wintertime,” he said. “The concern ... with language like ‘full self-driving’ is that people will be too confident in the technology” and the industry “will lose credibility and trust.”

The industry itself is at least partly to blame for the confusion, though. While PAVE is studying autonomous vehicle nomenclature with the aim of achieving greater clarity, there are no agreed-upon terms that define and describe the technology. The industry relies on a system of “levels,” from 1 to 5, with no automation on one end and total automation in all road conditions on the other. Adaptive cruise control, for instance, is Level 2.

Designed by the Society of Automotive Engineers, that ranking methodology offers a rough guide for technologists but hasn’t proved a useful tool for general public understanding. It’s one reason why many people think a term like Autopilot means the car can drive itself. Tesla’s driver’s handbook makes clear it doesn’t.

At the same time, Tesla has not made clear what it means by Full Self-Driving. The term seems to imply Level 5 full driverless automation. But when asked for clarification, Tesla’s public relations department sent this quote from a recent Musk presentation on the subject: “We expect to be confident enough from our standpoint to say that we think people do not need to touch the wheel, look out of the window.”

Adding to the confusion, Musk told analysts in July that “we already have full self-driving capability on highways.” After several high-profile crashes in which Autopilot might have been activated, however, Tesla issued statements saying the driver must pay attention at all times.

The lack of clarity extends as well to how Tesla is treating the FSD funds on its balance sheet. Tesla may have moved some deposit money off the balance sheet already. Its financial statements are opaque on the matter. Tesla’s chief financial officer, Zach Kirkhorn, told stock analysts recently that FSD revenue will be recognized “upon the release of additional features.”

“Everybody’s waiting to see what Tesla does,” said Nantel at the National Safety Council. “Federal regulators will be watching.”