It’s all about optics

Musk has been wrong about self-driving for years. His track record is marked by missed deadlines and broken promises.

He said:

“Our goal is, and I feel pretty good about this goal, that we’ll be able to do a demonstration drive of full autonomy all the way from LA to New York, from home in LA to let’s say dropping you off in Times Square in New York, and then having the car go park itself, by the end of next year. Without the need for a single touch, including the charger.”

That was in 2016, and therefore, he claimed it would happen by the end of 2017. Today, in 2025, Tesla is still not capable of doing that.

Musk has claimed that Tesla would achieve unsupervised self-driving every year for the last decade. It has become a running gag, with many YouTube videos featuring his predictions and a Wikipedia page tracking his missed deadlines.

Famously, the predictions are about Tesla achieving self-driving “by the end of the year” or “next year.”

This time, Musk has set a clear deadline of “June” for Tesla to launch its robotaxi service.

With Waymo pulling ahead in the autonomous driving race, now operating in four cities, providing over 200,000 paid rides per week, and soon expanding with 2,000 more vehicles, Musk needs a win to maintain the illusion he has been pushing for a while: that Tesla is the leader in autonomous driving.

He recently claimed about Tesla’s self-driving technology:

No one is even close. There’s really not a close second. We felt like it was a bit of an iPhone moment — you either get it or you don’t, and there’s a massive gap.

This is becoming increasingly difficult to claim amid Waymo’s expansion. Still, Musk believes that the robotaxi launch in Austin will help maintain the illusion, even though Waymo has already been operating like Tesla’s plans in Austin for years in other cities and for months in Austin itself.

Moving of the Goal Post

We have often described what Tesla is doing in Austin with its planned “robotaxi” launch as a moving of the goalpost.

For years, Tesla has promised unsupervised self-driving in all its vehicles built since 2016. Musk explicitly said that customers who bought Tesla’s Full Self-Driving package would be able to “go to sleep” at the wheel of their vehicles and wake up in another city.

Now, Musk is claiming that Tesla has “solved” self-driving with its “robotaxi” launch, but it is vastly different from prior promises.

Tesla plans to operate its own small internal fleet of vehicles with dedicated software optimized for a geo-fenced area of Austin and supported by “plenty of teleoperation.” This is a night-and-day difference compared to deploying unsupervised self-driving in customer vehicles, as promised since 2016.

Musk himself is on record saying, “If you need a geofence area, you don’t have real self-driving.”

Now, Musk is on record saying that Tesla will only launch the service in a limited area in Austin and even avoid certain intersections that Tesla is not sure it can handle:

We will geo‑fence it. It’s not going to take intersections unless we are highly confident it’s going to do well with that intersection. Or it will just take a route around that intersection.

In addition to geofencing, Tesla is also utilizing teleoperation to control vehicles with human operators remotely.

We reported last year when Tesla started building a “teleoperation team.”

Despite Tesla originally planning to launch the robotaxi service on June 12, and now “tentatively” on June 22, the automaker posted a new job listing days ago for engineers to help build a low-latency teleoperation system to operate its “self-driving” cars and robots.

The use of geofencing and teleoperation results in Tesla having the same limitations as Waymo, which Musk claimed means it’s “not real self-driving and not scalable to the customer fleet as promised by Tesla for years.

‘Paranoid’ about Safety

Musk claims that Tesla is being “super paranoid” about safety, but you have to take his word for it.

We have pointed it out before, but it’s worth repeating: Waymo tested its self-driving vehicles in Austin for six months with safety drivers and then for another six months without safety drivers before launching its autonomous ride-hailing service in the city.

As for Tesla, it tested its vehicles with safety drivers throughout Austin for a few months. Then, Musk announced in late May, only weeks before the planned launch, that it had started testing without safety drivers.

Despite many people being on the lookout for these driverless Tesla Robotaxis, they were only spotted for the first time last week.

Since then, only two confirmed Tesla vehicles without drivers have been spotted testing.

Furthermore, several of those vehicles were spotted with Tesla employees in the front passenger seat. While Musk claims that there are “no safety driver”, these “passengers” pay attention at all times and have access to a kill switch to stop the vehicle.

They virtually operate like “safety drivers”, but they are on the passenger seat rather than the driver’s seat.

Tesla is currently still in the “testing” phase based on the listing with the state regulators, which also mentions “no” safety drivers:

To go back to the “optics” for a second, Tesla’s head of self-driving, Ashok Elluswamy, has shared this conveniently cropped image of Tesla’s “robotaxis” being tested in Austin:

The image crops out the passenger seat of the car in front, which would show a Tesla employee, and the driver’s seat of the trailing car, which would show a driver, as spotted in Austin over the last week.

There’s also no way to know precisely at what rates these safety passengers and remote operators are intervening on the self-driving vehicles.

Tesla has never released any intervention or disengagement data about its self-driving and ADAS programs despite using “miles between disengagements” as a metric to track improvements and Musk claiming for years that self-driving is a “solved problem” for Tesla.

As we have previously reported, the best available data comes from a crowdsourced effort. Musk has previously shared and misrepresented the dataset in a positive light.

Currently, the data for the combined two most recent updates (v13.2.8-9) on Tesla’s latest hardware (HW4), which is reportedly the same hardware used in Tesla’s “robotaxis” in Austin, currently sits at 444 miles between critical disengagements:

That would imply a high risk of an accident every 444 miles without a driver paying attention and ready to take control at all times.

Tesla is also currently actively fighting in court against organizations trying to access its self-driving crash data.

There are currently efforts to raise concerns about Tesla’s “robotaxi” deployment in Austin.

The Dawn Project attempted to convey the potential danger of Tesla’s upcoming robotaxi fleet by demonstrating how Tesla vehicles fail to stop for school buses with their stop signs activated and can potentially run over children on the latest public Supervised Full Self-Driving (FSD) v13.2.9: