Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

This site may earn commission on affiliate links.
You never know. They’ve thrown legacy owners a bone now and again. Maybe a two week window where you get it for 1/2 price.

I’m leaning toward not going to happen. Do they ‘need’ those vehicles and drivers to add to the data pool? That’s the $64,000 question. From the graph we just saw because free trial - nope, doesn’t seem they need legacy. But I will sacrifice a chicken for you today and speak with the voodoo bosses.

Out of curiosity, did they split out parts and labor on that estimate?
haha, then people need to use the correct stat of how many Tesla's on the road are actually able to use FSD.

No, they did not split it out. It seemed like it was for both parts and labor. I was going to get it yesterday, as the app says it's just $1,000 for the hardware upgrade, but found out that's only if you buy FSD. Just the hardware upgrade is a little over $1,900.
 
Considering how much the medium term valuation of this company is becoming dependent on robotaxis, has anyone modeled what the potential losses / profits are dependent on accident rate? This is the main issue right? Critical disengagement rate matters ultimately because of the cost / mile driven it may affect. We should get more insight into this.

I just created a small spreadsheet, simple analysis. Gonna assume certain miles driven, but everything calculation is scaled with miles, not assuming fixed costs.

Let's say a robotaxi drives 60,000 miles per year. 75% of those miles are revenue generating. Revenue at $0.5 per mile. Operating costs of $0.2 per mile.

Accident rate: Let's assume 10,000 miles per "accident" to start.

Accident costs: These is the big variable that people can argue / give input on what is the best assumption. I googled and costs are all over the place. Let's assume $5000 payout cost Tesla would owe per accident.

Based on these inputs, total revenues would be $22,500 per year, and costs would be $42,000. A loss of $20,000 per year. Most of that due to accidents. If Tesla started with 10,000 robotaxi vehicles, they would lose 200 million in a year. 2 billion for 100k. And so on.

Let's now assume accident rate is 100,000 miles per accident. Now operating cash flow turns positive to $7500. So for 100k robotaxis they would earn a bit under 1 billion dollars per year, and so on.

Under this presumed $5,000 cost, at what accident rate would Tesla be breakeven?

I get about 30,000 miles per accident.

If cost per accident is $2000, then breakeven is ~ 12,000 miles per accident.

If cost per accident is $10,000, then breakeven is about 60,000 miles per accident.

Accident rates for human drivers are less frequent then this, so Tesla should be profitable at a lower level of reliability.

So I don't think the problem will be operating costs in ramping up a robotaxi network, it will be human injury rates that people may object to.

If Tesla can prioritize reducing high speed accidents more than fender benders, this could be the biggest benefit.

So Tesla could probably start some robotaxis in urban areas (no highways) when accident rates are around 30,000 miles per accident. With a small fleet, total number of accidents may not be too high or harmful. The costs should be okay.

If we assume a critical disengagement = accident, then the current rate is ~ 300 per mile. So Tesla needs to 100x the miles per disengagement to have a viable product to start a robotaxi network.

TLDR your investment mostly now depends are one metric: miles per critical disengagement. Expect a big improvement this year. But pay attention to if / when the rate of improvement starts slowing down again (as in any model, it will). If it starts slowing down not close to 30,000 / disengagement, that could be problematic.

Tesla would never launch a service that had an accident every 10,000 miles. That’s way worse than the average driver.
 
Informative post. I would be interested to hear more on how the construction zone flag person hand signal situation will be handled. School crossings guards etc.

This morning a nice polite gentleman across the street coming out of the Tim Hortons was adamant that I cross and turn first even though he had tight if way. . Lots of hand signals etc. I’m sure this is in the realm of AI but would still enjoy seeing comments on how this sceenario will be handled. Also accident scenes with police directing traffic.

Thanks.
Well, how the scenario is eventually handled is that both of you are in Teslas (or Tesla FSD licensed vehicles 🤪) and the cars execute the situation regardless of the sign language of the human occupants. Seat dance for all the car will car.

I’m not dismissing the myriad of edge cases or imaginative one off scenarios that everyone can come up with. What I’m getting at is that the vast majority of them likely don’t need to be individually solved for by programmers or AI or the NN or whatever. But rather once there’s enough FSD vehicles on the road, people will just let the cars do their thing. Additionally, as word spreads people are going to know that others are in self-driving cars, so should begin to understand that things like not obeying the rules of the road, even out of an abundance of Canadian politeness, isn’t going to work.

I have no idea when this tipping point occurs. Heck, we’ve been trying to determine the EV tipping point for years. I think the latter one is nearer than imagined because of all the hard scrambling going on to tell the world (North America 😉) recently that nobody wants EVs and watching some OEMs back up like their faces are on fire.
 
Informative post. I would be interested to hear more on how the construction zone flag person hand signal situation will be handled. School crossings guards etc.

This morning a nice polite gentleman across the street coming out of the Tim Hortons was adamant that I cross and turn first even though he had tight if way. . Lots of hand signals etc. I’m sure this is in the realm of AI but would still enjoy seeing comments on how this sceenario will be handled. Also accident scenes with police directing traffic.

Thanks.
RE hand signals,

This is a solved problem via neural networks. Waymo has had the ability to do it for awhile:

And no, Waymo does not have any special sauce that Tesla doesn’t have to do this. Tesla’s just handling things in priority order of severity and frequency, and things like hand signals, school zones, and stopping for emergency vehicles is lower down the list. But I think we’ll see all of these situations handled withon 6 months or so.
 
haha, then people need to use the correct stat of how many Tesla's on the road are actually able to use FSD.

No, they did not split it out. It seemed like it was for both parts and labor. I was going to get it yesterday, as the app says it's just $1,000 for the hardware upgrade, but found out that's only if you buy FSD. Just the hardware upgrade is a little over $1,900.
I agree.

Like buy for life or buy at least a one month subscription? If the latter, couldn’t you just say - um…everyone else getting a free one month trial. I want that after I have you upgrade by hardware. That seems reasonable.
 
All these models seem to assume that the robotaxis will be privately owned, and therefore Tesla has to share some of the profit. Why would tehy bother to do that rather than just own them themselves?

I know Elon proposed a "use your car as a profit center for us" model years ago. But he's proposed all sorts of things over the years that he pivoted away from. IMHO, if Tesla really is going to do this, they'll own and operate all the cars. And while they may have a dual-purpose model 2, all the manufacturing will be soaked up by the Robotaxi use for quite some time. Its not that they would be 'canceling' the small personal car -- its that producing it isn't the best use of manufacturing capacity so it would be low on the priority list. I see it as a backup plan... if regulatory approval takes longer than they'd like, they may produce and sell the personal version until it gets done.
 
While I understand the thought process and where we are going with this, I've seen a lot of people just lump all the Tesla cars on the road into a statistic like number 4 on the list.

Reality though, is that there are thousands of owners like me who have 2.5 hardware and would have to upgrade first (which is a little over $1,900 for the hardware upgrade - got quoted yesterday at the service center), so the round numbers people are throwing around need to lowered.

So that's first $1,900 + and then $200 a month for the legacy customers. Unless Tesla wants to upgrades us for free to drive the mission... :)
Change your though process slightly and you'll see that doesn't matter.

production has been on a exponential curve. Just assume that all new cars sold after robotaxi goes live become robotaxi and zero existing cars become robotaxi and you still get a curve like he shows.

Assume every HW 2.5 and even HW 3 car isn't in the pool and the curve is still exponential.

Add them in if you like but it doesn't change the shape of the curve after the first couple of years.

Or said another way, when in doubt zoom out!
 
Last edited:
Maybe the Robotaxi is the van not the M2. You heard it here first.

RT would work well as a van with the versatility of the Canoo style that can be easily configured/reconfigured for a variety of tasks.
A single platform easily adapted for carrying People, Packages, Food, Cargo, etc. would be ideal.

In my dreams, I imagine a rig that robotically makes a pizza to order while driving to a location to deliver it. Then, when the customer arrives at the vehicle it deploys the pizza, boxed and ready to go, from a slot similar to the Starlink Pez dispenser on Starship.

Maybe I should get something to eat now... 🤔 🍕
 
Change your though process slightly and you'll see that doesn't matter.

production has been on a exponential curve. Just assume that all new cars sold after robotaxi goes live become robotaxi and zero existing cars become robotaxi and you still get a curve like he shows.

Assume every HW 2.5 and even HW 3 car isn't in the pool and the curve is still exponential.

Add them in if you like but it doesn't change the shape of the curve after the first couple of years.

Or said another way, when in doubt zoom out!

meh, maybe. haha. I get it. Us legacy folks are:

 
  • Like
Reactions: wtlloyd
hahaha, woooooow. ok, then Elon. So what is the point of the acceleration then? Just mind your business and let it slowly develop...

It's always been about us...the planet will be fine, but it's us humans that will suffer....F*CK.

The CEO just said this. THE CEO of Tesla.


Tesla's mission statement:

Accelerating the world's transition to sustainable energy.”
 
So basically, you don't believe that Tesla is anywhere close to having the ability to run a robotaxi service.

....Oh no. Tesla can't take on the liability. Uber drivers take on this liability through their insurance companies. And Tesla runs its own insurance company.

If you want to argue intelligently that Tesla is not close to starting a robotaxi network then you need to first look at Waymo. Waymo has robotaxis today. So what is Waymo doing that Tesla can not do?


FWIW I agree with you on most of these points, but I think there's pretty near 0 owners that will take liability on themselves for their personally insured cars to drive around without a driver... If Tesla with 30B in the bank doesn't trust the product to do that, why should an individual owner? doubly so when you use Waymo as an example, and Waymo takes all responsibility for all their self-driving cars.

Tesla does own an insurance company- though the # of states it's actually legally underwriting insurance can be counted on your hands, with fingers left over- and that hasn't expanded in quite a while, and at last report was losing money.

Ideally that's the correct path- Tesla has their own insurance company and uses it to cover (on them) the liability of self driving RTs on their network... but it's a long way between here and there, especially in the dozens and dozens of states they can't even underwrite regular insurance yet--- THERE is your "it might take years to get through regulators" thing, unless Tesla instead insures with someone else while they're working on it, or just waits those years to deploy.



These Robotaxi models are all grossly conservative, ARK’s bull case is $613billion in net revenue from autonomous ride hailing within three years

Other than a random guess at a price target, once, that was still wrong on target date by years, and wrong on the specifics of how it got to that target, ARKs models are consistently, hilariously, wrong about nearly everything.

Why do we need to keep reminding people of this?


With tons of awful analysts throwing out targets all over the place SOMEONE was bound to accidentally "predict" a high price for Tesla-- Cathie was lucky enough for it to be her--- but it would've been a terrible idea to give her your money to manage as her actual results, over years, make clear.

Their financial performance has been awful- worse than just buying an S&P500 index fund, over a period where their one "win" (Tesla) has massively beaten the S&P500 over the same number of years.... because everything else they've invested in has done horribly.

They got lucky, once, picking Tesla early on- and just about everything else they've done since has been terrible.