During yesterday’s Tesla investor presentation, informally dubbed Autonomy Day, Elon Musk did what Elon Musk does and said a lot of things—this time about the company’s aggressive and definitely controversial expansion plans for autonomous technology. And lot of the things Musk said were not so much facts about Tesla’s cars, but predictions of how they will be used in the near future.
What’s interesting here is these forecasts, by and large, are far out of line with industry standard time frames and even technological practices.
Given his predilection for over-promising and under-delivering, here are some of the most significant ones we may want to come back to in the near future and see how they played out.
The first half hour or so of the Autonomy Day presentation was dedicated to Tesla’s in-house chipset development, which the company claims delivers best-in-class performance for self-driving cars. The goal is Level 5—the highest form of full autonomy—by 2020. Which is (checks watch) next year. There are no Level 5 vehicles currently for sale.
In short, Tesla developed a chip specifically for self-driving cars, rather than adapting powerful chipsets by major manufacturers that consume more power for less computational output. That sounds smart!
But then things got weird when Musk started predicting what this will mean for Tesla in the near term. Mainly, fully self-driving cars by next year:
By the middle of next year, we’ll have over a million Tesla cars on the road with full self-driving hardware, feature complete, at a reliability level that we would consider that no one needs to pay attention.
Tesla has a long history of using confusing terms when referring to its driver assist technology, including the very name of the software package “Full Self-Driving,” when it is not full self-driving.
Moreover, Musk claims they can do this without the use of LiDAR, the laser-based technology that creates real-time 3D mapping around the car and is used by most every other company persuing self-driving, even though it adds considerable hardware costs.
While others, such as Ford’s Chief Technology Officer Ken Washington, consider LiDAR absolutely necessary for self-driving, Musk, uh, disagrees, saying yesterday:
LiDAR is a fool’s errand. And anyone relying on LiDAR is doomed. Doomed. Expensive sensors that are unnecessary. It’s like having a whole bunch of expensive appendixes.
Because Tesla thinks the company can achieve truly autonomous driving without LiDAR, it means they believe they can ship this capability via software update rather than having to build cars differently. It’s quite a claim that, in short, very few other people or companies in the industry buy.
Further, the timeline of rolling this out by next year is similarly out of whack with industry predictions. Most other companies, including industry leader Waymo, are revising their time scales backwards, not forwards, for a full autonomous vehicle revolution, if it ever comes to pass.
Instead, they’re looking at limited rollouts in specific geographic areas favorable to full autonomy like Phoenix. Of course, Tesla is not proposing any geofencing for their cars.
A corollary of Tesla’s Level-5-By-2020 prediction is even wilder.
“From our standpoint, if you fast forward a year, maybe a year and three months, but next year for sure, we’ll have over a million robotaxis on the road,” Musk said, per TechCrunch. “The fleet wakes up with an over the air update; that’s all it takes.”
There will be a fleet of over a million robotaxis next year for sure is definitely A Take. It obviously relies on the whole full autonomy thing working out, which is a huge if on its own.
But even granting Tesla somehow beat every other company to full, all-condition autonomous driving AVs by a matter of years if not decades, the robotaxi thing feels very half-baked. After all, Musk admitted as much during the Q&A:
Make no mistake: this would be a disaster for cities. In this scenario, hundreds of thousands of Teslas at any given time would be dispatched to the nearest urban core to make some bank for their Tesla owner.
This would be an utter nightmare from an urban planning and traffic perspective. It is also something we absolutely do not need, given the glut of for-hire vehicle (human) drivers clogging major urban areas every single day as is.
It would also involve a tremendous amount of deadheading, or driving without passengers inside, as the Teslas travel to/from urban centers for fares.
Notwithstanding issues of range, charging, your autonomous car dying of battery in the middle of the road somewhere, etc., it seems to me the only problem this (hypothetical) solution solves is providing some way to make Teslas “cheaper,” therefore making them available to a price point that cannot currently afford them. But even then, it would almost entirely benefit upper-middle to upper-class earners while clogging streets used by everyone, making it a de facto regressive policy cities will almost certainly attempt to regulate.
Beyond that, the economics of this simply don’t make sense any way you approach them. If Tesla were, somehow, magically able to push a software update that gave a million cars this capability, as he claims, the supply for rides would far outpace demand, total gridlock in dense urban areas where most ride-hailing occurs would make these rides unattractive prospects, and Tesla owners thinking they could make back a good chunk of their car payments with robotaxi fares would be shit out of luck.
That’s a thing Elon Musk said yesterday, before comparing buying any other car to buying a horse, which is very bad financial planning advice.
I wish I could remember where, but I once read a review for an Android phone where the manufacturer promised regular software updates, but the reviewer made the point that you shouldn’t buy a product with the expectation it will become a better product in the future. You should be happy enough with the product as it is out of the box or else you’re in for disappointment.
To that end, it is definitely not “financially insane” to buy a car that costs tens of thousands of dollars less than what a reasonably trimmed Tesla will run because you don’t make enough money to afford a Tesla. In fact, it would be financially insane to buy a car you can’t afford simply because you believe it will be a completely different car in the future.
Look, if I ran a company that sold things, would I say it’s “financially insane” to buy anything other than my company’s product? Maybe! But that wouldn’t make it true. (We’d also add that it’s generally insane to expect any car to rise in value over time, unless you’re buying a rare Ferrari from the 1960s or whatever.)
Teslas, by many accounts, are great, impressive cars that owners are generally pleased to own. And, as someone who really does want electric cars to take off in the United States for environmental reasons, that’s part of what makes all this bloviating hyperbole about self-driving futures and robotaxis so frustrating.
Tesla makes far and away the best electric car readily available in the United States and, in many respects, one of the better cars of any kind. Which is to say, they don’t have to do this. They don’t have to create all these false promises. They could just... keep making good cars, but more of them—focusing on scale and quality improvements and long-term sustainability. Is that really too much to ask from a company that makes cars?