To The Moon

A large percentage of the Asymmetric Portfolio is invested in companies that could have major tailwinds from autonomous driving. My thesis is that many companies will make autonomous vehicles, leading to the modularization of components and technology, and aggregators like $Uber(UBER)$ ( ▼ 0.26% ) and $Lyft, Inc.(LYFT)$ ( ▼ 3.24% ) being huge winners as supply is commoditized.

The view of a more autonomous future is consistent with many investors, but how I envision that future is very different. The market still thinks that $Tesla Motors(TSLA)$ ( ▼ 5.92% ) — who first promised Teslas could soon drive across the country fully autonomously in January 2016 — will dominate the industry, and the Cybercab will be the ubiquitous vehicle.

So, I keep an eye on the companies I own and the competition closely.

Over the past few weeks, we’ve been hearing about a Tesla Robotaxi event that was to be held in Austin, Texas, on September 3. This could be huge! Details were sparse, but that’s not unusual for a Tesla event.

What actually took place was stranger than any Tesla event I’ve seen, and that’s saying something.

The first I remember watching was Elon Musk showing off the Solar Roof in 2016. It was a strange presentation because of its informality, and when I talked to people in the industry, they said this product wasn’t economically feasible, particularly in the configuration you see below. It turns out, it wasn’t, and Tesla shut down the Solar Roof with only about 3,000 installations.

Since then, we’ve seen outlandish after outlandish claims about FSD, Dojo, Robotaxis, and more at these presentations.

So, I was intrigued about what we were going to learn and how Tesla shaped up against my investments. But my guard was up for disappointment.

What we got was downright strange.

The Launch That Wasn’t

As the Robotaxi event drew closer, questions started to arise.

The event website disappeared.

People on Twitter started to ask what time the event was, and no one knew.

They asked for a livestream…and there wasn’t one.

We slowly learned that only a few content creators who were also Tesla supporters were invited. It was mainly employees there.

Elon Musk didn’t even show up.

What was announced? This video from David Moss shows…essentially nothing.

Tesla once again talked about how great vision-only autonomy is, how low-cost their vehicles are, and how great their manufacturing is. But they didn’t give launch plans or reveal safety data or say anything new.

That noted, we did get some news about Tesla’s Cybercab this week.

Tesla’s Questionable Business Model

Thus far, Tesla has been testing a few dozen Model Ys in Austin with small fleets in other cities. They’ve been handling the operations and maintenance of vehicles, but that’s apparently not the plan long-term.

As part of the Cybercab launch, Tesla asked people to express interest in buying and operating Cybercabs.

What exactly that business model looks like isn’t certain, but it’s likely that this copies the fleet owner model that Uber and Lyft have been testing. Someone buys a few dozen vehicles and then handles the operations (remote interventions), maintenance, cleaning, etc.

For a decade, Tesla has promised that individual vehicles will become money-making assets because they could be put into the Robotaxi fleet. That doesn’t appear to be the case if Tesla reserves fleet ownership for a smaller group.

But we may be a ways away from more Cybercabs hitting the road, anyway.

The Feds and Courts Come Knocking

There are two minor (MAJOR) items outstanding with Tesla’s Cybercab launch.

The first came out Friday morning when the NHTSA said it is investingating Telsa’s self-certification of the Cybercab.

On September 3, 2026, Tesla began commercial deployment with a small number of its Cybercab vehicles in Austin, Texas. Tesla notified the Agency that it certified those Cybercab vehicles as compliant with all applicable Federal Motor Vehicle Safety Standards (FMVSS). Tesla also notified the Agency that it plans to gradually expand commercial deployment of the Cybercab to include additional vehicles and locations. The vehicles lack permanently attached, conventional manual controls, such as a brake pedal, gas pedal, steering wheel, and mirrors. NHTSA is opening this AQ to examine the process and technical data on which Tesla relied when certifying the Cybercab and related issues. Among other things, NHTSA will consider the extent to which Tesla’s certification depended on determinations that certain FMVSS are inapplicable to the Cybercab.  

NHTSA

In the U.S., vehicles need to meet standards like having a steering wheel, brake pedals, and side mirrors, which the Cybercab doesn’t have.

To reduce regulatory burden, companies often “self-certify". If there are problems, the Feds come knocking, but most OEMs have been doing this for a long time, so they know what they’re doing.

This isn’t necessarily a big deal. But it might be.

The Cybercab isn’t just any vehicle. It doesn’t have the standard car parts like a steering wheel and pedals that are required by law.

Some companies get exemptions for vehicles — as Zoox has gotten for its autonomous vehicle — and even then, the exemption can take years to get, and production volume is limited.

Tesla has apparently self-certified without getting an exemption to standard rules.

Zoox robotaxi

We won’t know the findings until the NHTSA’s investigation is complete, but this could stop Cybercab scaling in its tracks.

The other, more thorny issue is safety. One of my biggest criticisms of Tesla is the company’s opaqueness when it comes to safety.

In 99.9% of cases, Tesla redacts the crash narrative, software version, and whether a vehicle was operating in FSD mode at the time of the incident. It’s hard to make the case that Cybercab will be safe if most of the safety data is redacted…

This became especially fishy to me years ago when I started following California’s autonomy data. Regulators collect a plethora of data and publish it regularly (albeit on a several-month delay). And as I followed the reports, I could see Waymo’s miles driven growing and how many fewer incidents Waymo had than, for example, Apple.

Tesla’s data was the most fascinating…but first.

Here are the tools available today, including the archived tools and data going back to 2014. I recommend browsing if autonomous vehicles are of interest. There used to be a disengagement report, which logged all disengagements and the reason, but that’s been discontinued recently.

You can see that Waymo drove 22.4 million miles in 2,664 vehicles through November 2025. That’s likely much higher today.

There have been crashes and incidents in Waymo vehicles as well, which we can dig into further.

During that time, there were 297 crashes in autonomous mode.

And this is where it gets interesting. Go to this link, and you can see details of each of these crashes (column BA tells the story of each crash).

I haven’t gone through each one, but the vast majority are human errors by other drivers or oddities like a sign falling on a car.

I believe Cruise has still driven the second most miles in California despite being shut down for nearly two years. This is now GM’s autonomous driving subsidiary.

Apple even drove nearly a million miles before discontinuing its program. Its safety record was never close to Waymo’s.

Pony.AI is another company that’s put in a lot of miles to test.

And then there’s Tesla.

Despite having a permit to test fully autonomous vehicles going back as far as I can remember, the company recorded only 12.2 miles of autonomous driving in a single car.

That’s not a typo. 12.2 MILES!

Why wouldn’t Tesla test in California?

I mentioned above that it would have had to record all miles, crashes, and disengagements in California. So, Tesla does its testing in Texas, where the reporting and rules are much more lax. You can see Austin’s autonomous vehicle incident dashboard here, but this is hardly every incident or disengagement given the videos we’ve seen online.

So, the Feds are investigating Tesla, and safety is…questionable, at best.

But sure, Cybercab is only a few months away!

The Autonomous Vehicle Industry Moves On

I’ve been skeptical that Tesla’s vision-only approach to autonomy will ever be safe enough to launch fully autonomous vehicles. Sure, vision can handle 99.99% of cases, but they get 1 incident in a few million, and someone dies.

And Tesla has been settling case after case in court over Autopilot and FSD deaths, so I’m not making this up to scare anyone. Ironically, Tesla settles before these cases get to court and become public…so ask a lawyer why they do that.

Meanwhile, the rest of the industry moves on. Most companies are solving safety first and figuring out cost later because…the cost of technology always comes down over time.

Waymo has now driven over 200 million miles fully autonomously.

Zoox is slowly scaling its fleet.

Nuro is expecting to launch soon with Lucid on Uber’s network.

Then there’s AVride, Weride, Wayve, May Mobility, Motional, MOIA, Mobileye, Pony.AI, Baidu, and more.

They’re all going through regulatory approval and scaling operations around the world.

Uber expects autonomous vehicles to be operating in a dozen cities by the end of this year and more than double that next year.

If there was a first-mover advantage for Tesla, it’s gone! And that may be the biggest takeaway from this week.

We’re only mildly closer to a million Cybercabs on the road than we were in June 2025 when they first launched in Austin. And in that time, Waymo is adding a few hundred vehicles to its fleet a month.

Any delays, whether related to regulators or safety, will further allow competitors to scale, improve operations, and reduce costs.

This isn’t a winner-take-all market, the way investors thought it was a decade ago. And even after 10 years of talking, Tesla somehow disappointed even the most ardent supporters with its “Cybercab Event.”

I think my thesis about the future of autonomous vehicles is intact.