No, Tesla's Robotaxi Won't Be as Cheap as a Bus Ticket

Common sense and a little math go a long way to debunking Elon's ridiculous claims...

Let’s come right out in the open. This quote from Elon Musk is going to be quickly debunked if the Robotaxi ever sees the light of day:

“Also, busses will not be needed when Tesla rolls out unsupervised full self-driving, as they will take people point to point for a similar cost to a bus ticket.”

It’s important to level set a couple of things. First, what is the average bus fare in the United States? If you believe Statista (and I have a hard time using them as a single source), they claim it’s roughly $1.52. That sounds about right, and that price can drop for volume passengers that buy things like daily, weekly, or even monthly passes. I wasn’t content on using Statista, so I found this rollup from the American Public Transportation Association’s (APTA) 2023 Fact Book:

$0.90 revenue per unlinked trip

That is certainly far lower than the $1.52 cited by Statista. But I believe that’s because they are counting all unlinked trips (daily, weekly, monthly passes as well as single trips). Average trip length is 3.7 miles.

So, busses on average are $0.90/trip for 3.7 miles. Or $0.24/mile.

Comparing to Ride Shares (Uber)

Comparing to Uber/Lyft is tricky as location, demand, and a multitude of other factors come into play. If we try to get to the bare minimum of price per mile, Uber on the lowest level is around $2/mile. NetCredit attempted to calculate cost per 10km ride at many cities throughout the United States (and abroad). FWIW, 10km is roughly 6.2 miles. Here are some headliners (and I’ve adjusted them to be per mile):

• El Paso - $12.56 ($2/mile)

• Baltimore - $27.27 ($4.4/mile)

• New York City - $34.74 ($5.6/mile)

Finding how much Uber gives their drivers is pretty much impossible. I’ve seen some places say as high as 75%, with some drivers claiming they get as little as 50%. Once again, there are a myriad of factors at play here. But let’s just grab the average here and say that Uber pays their drivers ~63%.

For the numbers above, if we took the drivers out of these things entirely and just treated the vehicles as some autonomous unit driving around picking up folks, here are the revised prices:

• El Paso - $4.65 ($0.75/mile)

• Baltimore - $10.09 ($1.63/mile)

• New York City - $12.85 ($2/mile)

So, at the cheapest, we are looking at robotaxis being 3x the fare of a bus. At the highest? 8x.

That isn’t a “similar cost to a bus ticket”. Sorry, Elon.

But the thing is, do we really need to guess what the cost per mile would be for a robotaxi? Doesn’t Waymo exist? Well, yes, yes it does.

Comparing to Waymo

Waymo is, by all accounts, Tesla’s competitor in the robotaxi space. Well, that’s if/when Tesla ever launches their robotaxi. Truth be told, Waymo has been operating fully autonomous vehicles for customers since October 2020 (and for years prior with safety drivers).

So, yeah. Waymo has a bit of a head start. Today you can grab a Waymo in Phoenix or San Francisco with Los Angeles “onboarding riders” and Austin having a wait list.

Mario Herger wrote an article on his blog “The Last Driver License Holder” in which he calculated the costs of Waymo rides. It is a sample size of about 70 rides, so I think it has a decent set of data points to include here.

Some of the main data points from that blog post:

• 90.2 miles totaling $842.75

• Averaged 1.32 miles/trip

• San Francisco had a 2% surcharge for traffic congestion

• Average fare without surcharge was $11.84/mile

• Cost varied based on time of day. Most expensive times were 1pm, 3pm, 4pm and 6pm averaging nearly $14/mile

That is obviously incredibly expensive. We are talking about nearly 2-3x as much as Uber.

I dug a little bit into this.

In Phoenix, a lot of riders are claiming that on average it is cheaper than Uber/Lyft, but not tremendously so. Some riders in San Francisco claim it can be cheaper than Uber/Lyft, but during peak times it’s often more expensive.

Either way, we are talking about some exorbitant prices that are often on par or more expensive than human ride sharing services.

Can Tesla Do It?

There are three sub-questions and answers to this.

Can they create a robotaxi? To that…yes.

Can they scale a robotaxi? To that…maybe?

Will the robotaxi cost similar to the cost of a bus? To that…no. Not anytime soon.

So, here are some things to consider:

Elon Musk has pushed Tesla to go vision only as their preferred mechanism for autonomous driving. This has the benefit of reducing costs but comes with safety issues. There lacks any meaningful redundancy in the event of, say, camera malfunction (due to dirt, electrical issues, etc.). That means no backup hardware like LiDAR or RADAR for autonomous driving purposes.

We’ve seen Tesla make numerous safety concessions often just for the sake of novelty. The Cybertruck, afterall, uses steer by wire with almost no redundancy. On a $120k+ vehicle.

But even still today, Teslas have “Supervised Driving”. I refuse to even call them “Full Self Driving”, even with the newly, legal team added “Supervised Full Self Driving”. And that’s because not only does a driver have to be in the driver’s seat, but they need to be monitoring/hands on wheel at all times.

There isn’t much in the way of regulation at the federal level for autonomous vehicles. A nice writeup from global law firm Norton Rose Fulbright can be found here.

The NHTSA does have a framework defining various levels of automation. Here’s a snapshot:

• Level 0 (L0): No automation

• L1: Function specific automation (ex: adaptive cruise control)

• L2: Combined function automation (ex: supervised driving)

• L3: Limited self-driving automation (ex: mostly unsupervised driving, driver needs to be available in certain situations)

• L4: Full self-driving automation (ex: vehicle handles everything)

Now, that’s a framework. But I do anticipate that state and local governments as well as (hopefully) the federal government will have strict regulations for L3 and above.

As things stand today, Tesla only has L2 capabilities.

Can they get to L3 without backup hardware/solutions and only relying on vision? Call me doubtful. If for nothing else as I anticipate there being state/local regulatory hurdles and public perception issues abound since Waymo, Tesla’s competitor, relies on a myriad of hardware and not just vision. Even Mobileye, which does have a licensable vision only solution, also has multi-modal solutions that, yes, include LiDAR.

But all that aside, can Tesla get their price per mile to be at a “similar cost to a bus ticket”? Well, Waymo can’t. And even without all the necessary hardware, even if they cut their prices in half, they’d still be magnitudes higher. Even if they cut their rates 3/4, still magnitudes higher.

I don’t anticipate a publicly traded company like Tesla to take a hit on profit margins, especially since Elon has desperately hung his hat on the robotaxi being the next big thing for Tesla. And I don’t anticipate them being able to compete with busses on price.

To me this is more just wishful thinking from Elon to push the hype train for the robotaxi event.

All that said, I do think Tesla will eventually get into the robotaxi game. It’s not a matter of if, it’s a matter of when. But as with anything that Elon says, it’s usually years after the fact.