Tesla's approach to developing and selling self-driving technology could cause some hiccups for the company down the line, experts say.
The most divisive element of its strategy, arguably, is its refusal to use lidar sensors, which emit pulses of light that bounce off objects to determine where they're located, in its planned self-driving system. Lidar sensors are seen by Tesla's high-profile competitors, like Waymo, General Motors, and Uber, as a necessity, but Tesla CEO Elon Musk has said they're a "crutch," and that Tesla can build self-driving cars using just cameras and radars.
Gene Munster, a managing partner at the venture capital firm Loup Ventures, said while Tesla will be able to improve its semi-autonomous driver assistance system, Autopilot, without lidar, Tesla will have to use it to achieve Level 5 autonomy — which means a vehicle can drive itself without human input in all situations.
Munster said he is interested in Tesla's autonomous driving technology in part because of a 2016 video on the company's website, which depicts a vehicle driving itself on residential streets and handling situations, like left-hand turns at intersections, that present challenges for self-driving vehicles. The video suggests that Tesla's autonomous driving technology is close to consumer-ready, but Munster said that likely isn't the case.
"I think we're just further away than that video would make you believe," he said.
Adam Jonas, an auto analyst at Morgan Stanley who has a $291 price target for the stock, said Tesla will most likely use lidar within the next eight years.
"I would be surprised if, by the middle of the next decade, they did not have some form of lidar on the car," he said.
Tesla may have to refund a risky product
If Munster and Jonas' predictions are correct, Tesla will find itself in a difficult position. Since October 2016, the company has sold what it calls "full self-driving capability." For $3,000 (or $5,000 if purchased after delivery), customers can double the number of cameras on their vehicles from four to eight, which Tesla says will give them the ability to drive without human assistance once the necessary software is ready and receives regulatory approval. But on its website, Tesla says it doesn't know when that will happen.
Selling a feature without giving any timeline for its availability is unusual in any industry, and Munster said he doesn't think Tesla will be able to fulfill its promise that all vehicles sold since October 2016 will be able to support completely autonomous driving technology.
"I'm surprised they sell it," he said.
Munster said he believes fully-autonomous vehicles won't arrive for five to 10 years, and that Tesla will have to refund the customers who bought the self-driving hardware for incompatible vehicles, a number Munster said he thinks is relatively small compared to the number of Autopilot customers, or offer them discounts on newer Tesla vehicles.
Tesla's confidence in its ability to integrate new software into existing hardware may stem from its history of vertical integration. The company has long sought to design and manufacture as many elements of its vehicles as possible, rather than outsource them to third parties, a rarity in the modern auto industry.
That desire extends to autonomous driving technology. While many of its competitors, like General Motors, Ford, Toyota, Volkswagen, Waymo, and Uber, have sought acquisitions or partnerships, Tesla has been developing its self-driving software without outside assistance and has moved to decrease its reliance on third-party hardware. While the company's current vehicles feature software chips made by Nvidia, Musk said in 2017 that Tesla has been developing its own chip. During the company's second-quarter earnings call in August, Musk said the chip's processing power is "an order of magnitude" better than any existing options.
Going it alone could pay off
Tesla's go-it-alone strategy risks spreading the company's resources too thin or demanding unsustainable levels of investment, but it could pay off in the long run, Munster and Jonas said. Munster compared Tesla's desire for vertical integration to Apple's. Unlike some mobile phone makers, Apple sought to control the manufacturing and design of the iPhone, which has helped drive the company's enormous profits and historic market capitalization since its introduction in 2007.
For traditional automakers, who have spent most of their existence focused on vehicle manufacturing, a partnership with a tech company can accelerate the transition to a business model that is more reliant on software and services. Likewise, partnering with an auto manufacturer is easier for a tech company than learning how to make cars. But Tesla has emphasized both software and manufacturing expertise since its founding, which is the optimal route, Munster said, and makes it easier for the company to work on new technology without outside assistance.
Jonas said taking on autonomous driving hardware and software by itself will yield dividends for Tesla, even if it does end up partnering with another company down the line. By attempting to solve the most difficult problems in self-driving technology, Tesla will develop a wider base of knowledge than its competitors, which could help in identifying and working with potential partners, Jonas said.
Tesla's desire to create self-driving technology with minimal assistance is the latest in a history of bold, unconventional decisions. How well this one pays off will have a significant impact on the company's future.
Have a Tesla news tip? Contact this reporter at mmatousek@bjinnox.com.
Read more about automotive technology:
- I tried Tesla's Autopilot for the first time — and it was terrifying at first
- Car companies are pouring billions into self-driving tech they may never use, experts say
- Waymo could be worth as much $175 billion — here's a brief history of the Google Car project
- A Waymo employee reveals how the company is putting their self-driving cars to the test
- I rode in a self-driving shuttle in Times Square — and it showed what the early years of fully-autonomous vehicles might look like
- We tried Tesla's, Cadillac's, and Mercedes' semi-autonomous driving systems — and the winner is clear
