Tesla's Robotaxi Dreams Are Dead

And Musk is running away from the problem.

Photo by Pixilustration on Unsplash

Self-driving has been a core part of Tesla for well over a decade. It is what prompted investors to classify it as a tech company rather than a manufacturing company and is what has driven its valuation to dizzying heights through insane speculation. Cathie Wood has gone on record to state that Tesla’s self-driving technology will become its main income source by 2029, adding more than $7 trillion to the company’s already ridiculous value. But unfortunately, the wheels have fallen off, and this fever dream is dead on arrival. This is utterly devastating for Tesla, and how Musk plans to treat this self-inflicted wound will only make the situation worse.

In October, Musk announced that Tesla plans to expand its tiny experimental ‘Robotaxi’ fleet in Austin, Texas, to 500 vehicles. But, last month, Musk revised this to “roughly doubling” the number of Austin Robotaxis by the end of December, which would mean only 60 vehicles. That is an underperformance of around 88%!

And, to make matters worse, these aren’t really robotaxis. Legally, there still has to be a driver in the car who monitors the system.

Now, Musk wanted to expand fully driverless Tesla Robotaxi services in Phoenix, San Francisco, Miami, Las Vegas, Dallas, Austin and Houston by the end of this year. However, Tesla’s FSD (Full Self-Driving) system is apparently so unfit for purpose that it can’t pass any of these cities’ safety requirements and so has failed to obtain a permit to even operate supervised operations for these cities! This appears to be why Tesla has failed to deliver 500 Robotaxis to Austin, since they are unable to obtain a permit to operate that many of them, which is almost certainly due to how badly they drive. If you want to know just how horrifically Tesla’s Robotaxis drive, read my previous article here.

Instead, there are just a few dozen “Robotaxis” in Austin, each with a legally mandated driver in the passenger seat who has their finger on a kill switch. This makes them a “supervised” pilot program, rather than a fully autonomous system. To make matters worse, the driver has to switch to the driver’s seat during a more challenging route, something which I’m sure comforts passengers. There are also a few “Robotaxis” in San Francisco, but they legally have to have a driver in the driver’s seat ready to take control at a moment’s notice. You know, just like an Uber driver that shelled out for FSD.

So, technically speaking, Tesla has zero fully autonomous taxis and fewer than 100 augmented taxis. Oh, and these augmented taxis cost more to operate than conventional taxis, because the FSD AI is not free, even for Tesla, and you still have to pay the driver. This isn’t from lack of trying or lack of capacity. No, this underwhelming rollout is because legal authorities do not want a system this unsafe on their streets.

If you want to know why Tesla’s FSD system and its Robotaxis suck so much, read my previous articles here and here. But, in short, Musk single-handedly ruined the system by taking a vision-only approach and leaving zero redundant safety features in place. As such, the system relies on developing nearly 100% accurate AI to get even close to acceptable levels of safety. This is something that everyone is finally realising is totally impossible.

I just want to take a moment to compare this to Tesla’s main rival, Waymo. They have well over 2,000 fully autonomous robotaxis in the US, operating in five major US cities, with services being actively rolled out in six more, including London and Tokyo. This should give you an idea of just how far behind Tesla’s self-driving capabilities are. London is widely considered one of the hardest places to deploy self-driving technology, as it has complex, archaic streets, highly strict traffic safety laws, and locals who begrudge change. Yet Waymo is breaking into that market before Tesla even has a permit to run a single ‘unsupervised’ Robotaxi.

And it seems that it isn’t just legal authorities who don’t want to touch FSD with a ten-foot barge pole. Other manufacturers are staying well clear.