A Model X heading from Seattle to Yellowstone National Park crashed on a two-lane highway in Montana Saturday, veering off the road and severely damaging the front passenger side of the vehicle. No serious injuries were recorded, and the driver was cited for careless driving.
Tesla, according to CNN, said that its logs show no force was detected on the steering wheel for more than two minutes leading up to the crash, even after the vehicle advised the driver to retake control of the wheel. The automaker repeated guidance that the autopilot feature, which is an advanced lane control system but not autonomous driving, is best used on divided highways or at slow speeds and not while winding on a narrow road through a rural mountain valley.
The crash echoes many of the narratives that have been repeated in recent weeks as scrutiny has grown over the autopilot feature following disclosure of a fatal crash involving a Model S in Florida. While bears will cite this as another case of Tesla's technology running amok, Tesla and its fans can rightly argue that the company and its software can not technically be blamed if its customers are using it improperly.
The bigger fear for Tesla is winning the battle but losing the war.
While there is evidence to suggest the drivers in Florida and Montana, as well as the driver in a third recent incident in Pennsylvania, were using the technology incorrectly, that evidence also raises the question of whether Tesla is going too fast too soon in rolling out the alluringly named "autopilot" to customers.
Tesla isn't alone in developing early-stage self-driving technologies, but while more traditional automakers have mostly limited their development to the testing track Tesla has made autopilot, a company-described "beta" product, available to customers. That decision was controversial even before this recent run of bad headlines, with a Volvo safety engineer earlier this year warning that Tesla's system gives the impression that it does more than it is actually capable of doing, potentially causing drivers to lower their guard.
The autopilot feature is opt-in, and Tesla is careful to state in its manual that owners should keep their hands on the wheel at all times. As noted in the Montana crash the system also reminds drivers to stay alert.
But as Tesla is finding, human behavior isn't nearly as reliable as technology can be. And company CEO Elon Musk has at times been guilty of feeding the hyperbole surrounding autopilot, last year boasting that Tesla vehicles could go from San Francisco to Seattle almost entirely hands free and at times re-tweeting videos of customers marveling at the technology and testing its limits on public roads.
As far back as September 2013 Musk predicted that Tesla automobiles would be able to operate autonomously for "90 percent of miles driven within three years."
Even if Tesla is not legally liable for the crashes, and it quite possibly isn't, there is considerable reputational risk involved in these accidents. Tesla vehicles sell in large part due to the brand's reputation for superior technology. A spate of crashes involving one of Tesla's most jaw-dropping features, even if that feature is not technically at fault, could work to erode the brand's perception just as competition builds as other automakers ready their own electric vehicles.
More practically Tesla also gains added revenue from each autopilot sale, as the feature is a $2,500 to $3,000 add-on to most vehicles sold. With the company facing billions in upcoming capital expenditures as it seeks to build out its manufacturing capacity, add retail and charging locations, and potentially complete a deal for money-losing SolarCity (SCTY) , every dollar in revenue is important.
Tesla hopes to deliver 50,000 vehicles in the second half of 2016. If the percent of buyers who choose to activate autopilot falls from a hypothetical 75% of sales to 50% of sales due to negative publicity, at the more conservative $2,500 per activation number, that is more than $30 million in lost revenue. If the company succeeds in its outlandish goal of delivering 500,000 vehicles annually in the coming years but autopilot remains out of favor that is hundreds of millions in possible lost revenue.
The reality for Tesla is that each one of these crashes is potentially more significant than the actual data behind any of them individually would imply. Investors need be cautious, lest they be caught off guard should the stock drive off the road.