Tesla’s Approach to Self-Driving Cars Could Cause Problems Down the Road

Text size

Tesla

is a leader in self driving technology, but it’s taking a risky approach that could prevent the auto maker from realizing the benefits of scale.

Mass market self-driving cars aren’t here yet, but some autonomous driving features have arrived.

Tesla (ticker: TSLA) is already offering the higher levels of its autonomous driving features, dubbed full self-driving, on a subscription basis today—or for a one-time payment of $10,000. But whereas other car makers are doing autonomous driving with cameras, radar, and laser-based radar, Tesla is alone in using optical cameras without other sensors. That could be a problem if safety regulators decide down the road that multiple sensors are required for full autonomy.

First a quick primer on how self-driving cars are evolving. The industry talks about autonomous driving in terms of levels. Level 2 systems, which are available today, are impressive—they can change lanes, manage speed, and brake without human intervention, but require a driver to stay engaged, with eyes on the road.

Level 3, 4, and 5 systems aren’t available yet, but allow drivers to stop paying attention. Level 3 systems might operate on highways. Level 4 systems can drive without intervention most of the time. Level 5 systems—the Holy Grail—can handle every driving task, even in crummy weather, without a human.

Level 2 systems being rolled out now are nice and can generate revenue for car companies in the same way air conditioning did two generations ago. But the big revenue opportunity starts at Level 3. “Car companies see level 3 as a revenue opportunity,” says lidar maker

Innoviz

(ticker: INVZ) CEO Omer Keilaf. With Level 3, car makers can generate subscription sales by offering software-enabled, self-driving features.

Most auto makers believe lidar is required to deliver truly self-driving cars. They say that safety regulators will likely require sensor redundancy to allow Level 3 systems to operate on roads.

When Level 3 software exists any car with the right hardware—lidars and cameras—can instantly upgrade to Level 3 by downloading the newest software. With lidar “you have Level 3 ready platform,” adds Keilaf.

Tesla, of course, hopes that it can achieve Level 3 autonomy with only cameras. But if it can’t, it will have a hardware problem. Millions of Tesla vehicles won’t be able to be Level 3 because they have cameras only. That will limit sales.

And if other cars have lidars and cameras Tesla could fall behind in self-driving features. Tesla didn’t respond to a request for comment about its camera-only strategy.

The good news for Tesla is no high-volume production cars have lidar on them yet. Lidar is just starting up. And New Street Research analyst Pierre Ferragu believes Tesla can simply add lidar when costs drop to an attractive level.

Cars, up until today, weren’t really software-enabled devices—they didn’t get better the older they got. Autonomous driving software is changing that.

Hardware will still matter though. Don’t forget, not all iPhones can support the most sophisticated iOS update. Sometimes better hardware is needed.

Tesla stock is up about 6% over the past month even as the

S&P 500
and

Dow Jones Industrial Average
are off about 4% and 3%, respectively. Tesla delivered more cars than analysts had expected, but the company is also is rolling out the latest version of its self-driving software. That’s another factor that has improved investor sentiment.

Write to Al Root at allen.root@dowjones.com