Tesla Network / Robotaxi Business Model

[Buckminster]

Average length of car ownership in the US is about 6 years.... (71.4 months to be specific) so the point where it's "technically" more than 10k, over 71 months, would be $140.85 a month.

Even more technically it's slightly lower since you're not paying 10k up front and thus paying later with "less valuable" inflated dollars.

So $150 month would, for average length of ownership, be a "worse" deal but not hardly by much (especially since you derisk totalling the car and not liking how your insurance handles FSDs value or something if you bought it).

$200/mo bumps you to 14.2k over 71 months compared to 10k for purchase, a roughly 42% premium ignoring inflation impact.

Not sure how Tesla will proceed. Charging $10,000 or $200 a month leaves a ton of money on the table from people like me who would definitely not buy it. I am very interested in FSD but the most I would pay is $7000 or $100 a month.

to me the answer is obvious, FSD needs to be separated into two products, one personal, one commercial. The commercial addition is actually going to be worth way more than $10,000, possibly as much as $50,000. The personal edition will drive you all day long , park itself etc. it will not drive unattended and pick up your mom from the airport.

I've been posting about a separate commercial license/cost here for years. Surprisingly very few has thought this is gonna happen. I'm still almost 100% certain it will. It's the only way the can get max profits from robotaxis and still get significant amounts from all those not using their cars for taxis.

It is possible the commercial profit can be taken via a mileage cost. This will work better for those only interested in using their cars as taxis less than full time.

I've also posted about the benefits of subscription for as long. Almost nobody liked or thought that would happen either. Until Elon said it would some months back.

Over a 6 year period, a 42% premium for subscribing at $200/mo vs. buying up front strikes me as quite reasonable. The whole point is just buy FSD if you plan to own the car a long time. IMO, a breakeven at ~4 years makes sense.

The risk of subscriptions cannibalizing $10K up front vs. generating incremental revenue from additional FSD users is something Tesla is weighing. Methinks the initial monthly fee may be adjusted once the initial response is in.

That is possible, but I still see them selling FSD up front for a while, for the same reason that they are selling cars.
it is good to get a big chunk of cash up front, and mix that in with some longer term subscription revenue.

In terms of take-rate, customers have a choice, the conditions around subscription may cause some to buy outright.

The subscription and Robo-taxi revenue will grow slowly, and increasingly become a bigger chunk of the business,

Another trend is Tesla Energy becoming more important including some possible subscription revenue\e from Autobidder.

While they are growing fast they want to maximize revenues where possible, because that money can be invested in expansion.
Up front sales bring the dollars in the door.

Possible timeline:

Apr 21 - FSD Beta rolled out across USA etc.
May 21 - FSD price raised to $15k
Jun 21 - FSD subscription introduced at $250 / month
Jan 22 - FSD price raised to $20k following V12 in Dec (subscription maintained at $250)
Jan 23 - FSD price raised to $75k, Tesla start keeping 10% of vehicles
Jan 25 - 50% of Tesla vehicles are not sold, remainder are leased and kept long term by Tesla

FSD calc - 50% residual value = $7500
$7500/36 month lease = 208
I prefer $250

$250 will initially look expensive - this will provide a new aura to the product. It will very quickly look cheap which is a good thing as Tesla should avoid having too many long term FSD vehicles in circulation where it can't pickup up the majority of future Tesla Network revenue. Subscription could become the non commercial FSD route of usage.

 

[Buckminster]

https://twitter.com/truth_tesla/status/1366815403742294016

 

[Buckminster]

 

[Buckminster]

I'm becoming more and more bullish on robotaxi business model. See here on use of robotaxis and V2G.

It's simple:
Tesla = Tesla owned robotaxis = biggest power provider

It's a complete solution:

1. Energy
2. Human transport
3. Good transport - replace semis - factory direct to home might become possible which would enable
4. Replacement of logistics chains including warehousing

I'm thinking of doubling down on my simplistic model that gets TSLA to $125T market cap in 2030.

 

[LiveLong&Profit]

I'm becoming more and more bullish on robotaxi business model. See here on use of robotaxis and V2G.

It's simple:
Tesla = Tesla owned robotaxis = biggest power provider

It's a complete solution:

1. Energy
2. Human transport
3. Good transport - replace semis - factory direct to home might become possible which would enable
4. Replacement of logistics chains including warehousing

I'm thinking of doubling down on my simplistic model that gets TSLA to $125T market cap in 2030.

"One reason why Elon has not yet fully committed to V2G is because he is waiting for robotaxi in combination with V4 supercharger etc."
Agree: Now is not the time, neither the volume of cars nor the extreme battery durability is there - yet.

I have trouble with 3: Aren't semis really useful - and self driving too? You need a lot of cars to replace one semi.

 

[Buckminster]

"One reason why Elon has not yet fully committed to V2G is because he is waiting for robotaxi in combination with V4 supercharger etc."
Agree: Now is not the time, neither the volume of cars nor the extreme battery durability is there - yet.

I have trouble with 3: Aren't semis really useful - and self driving too? You need a lot of cars to replace one semi.

I think robotaxis will first be used for pizza delivery. Then last mile logistics. Then farm or fishery to home where speed is of the essence. Eventually you get to a point where the cost of logistics (transport, warehousing, scanning, picking, unloading, loading etc.) is more expensive than the cost of shipping one item from factory to home.

Take Nike as an example. They have pulled their products from Amazon so that they can control the entire experience. Eventually, you will be able to order a pair of personalised sneakers and have them shipped overnight from a nearby micro factory direct to you in a Nike branded vehicle.

 

[LiveLong&Profit]

I think robotaxis will first be used for pizza delivery. Then last mile logistics. Then farm or fishery to home where speed is of the essence. Eventually you get to a point where the cost of logistics (transport, warehousing, scanning, picking, unloading, loading etc.) is more expensive than the cost of shipping one item from factory to home.

Take Nike as an example. They have pulled their products from Amazon so that they can control the entire experience. Eventually, you will be able to order a pair of personalised sneakers and have them shipped overnight from a nearby micro factory direct to you in a Nike branded vehicle.

Hm... ok..
Still think that semis and vans have a role.
But ... given a sufficient large amount of robotaxis roaming around, with the cost of 5 cents/mile for Tesla as fleet owner, that opens up a lot of room to disrupt traditional bulk transportation. Hm...
Kind of weird to imagine a world of autonomous couriers.
Maybe the correct mental model is something like TCP/IP packet routing. Yeah, that's probably it.

As others have pointed out, (Dave Lee had a 'last foot' video recently - haven't seen it but referenced in a chat with James Duoma) Tesla may go into robotics. Solving FSD and having dojo may open that door for Tesla. Maybe wide.
Also, and this is important, because Tesla is stacking up so many great business opportunities, having both cars AND robotaxis may be the killer combo.

Think of people in apartments, disabled, elderly - or just lazy.
If the human driver delivers you pizza at your door, with a smile, why should you settle for schlepping down to the parking lot, or out (in the cold ?) 20 steps to the curb at your front lawn.
Also covid or other pandemics: People may choose or be forced to isolate themselves. Robots to the rescue.

The last missing piece is starlink, providing global low latency internet connectivity worldwide.
Should a robotaxi or future robot get stuck, the human operator comes in and rescues it - or evaluates what measures should be taken for at later 'rescue'.

It doesn't need to be a big percentage wise to be a lot numerically. But a few operator-centrals on each of the 3 major customer continent with a 1000 operators at any given time would go a long way.
~50 complex cases per work-day per operator, ~3000 operators = ~150.000 'odd' cases handled per day. Given 100 mio cars/robots, that is a very low error rate of 0.15 % But, it may be possible for Tesla to get that low.
50 cases seem low.
But cars could get stuck and perhaps you need to monitor repair. Robots may have time-consuming special cases: Discovering people who are ill, need help, reporting or preventing crimes: A lot of things can happen when interacting closely with humans. Perhaps 0.15 error/operator rate is too low. Maybe 1% is more realistic.

But robotaxi and androids will be extremely profitable, so maybe having 20.000 operators online all the time is not really a big deal.

 

[Buckminster]

Yes: 50 mio robo-vehicles/year is a staggeringly wild assumption. Granted.

But ... we don't know how the transportation market will look like in 20-30 years.
There is a path for Tesla to get robotaxi cost down to 5 cent/mile. Lets say 7 cent/mile price. With some extrapolation of growth globally, and more and more people joining the (perhaps lower) middle class globally there might be a huge market for 7 cent/mile transport.

If you include vans and mini-buses you can perhaps get the price per customer below 1 cent/mile. At 1 cent/mile almost all people can afford some kind of robo-transport in 20-30 years. So around 9 billion customers. (depending on population growth)

 

[Buckminster]

Another reason that Tesla will not sell cars:

By running all vehicles 24/7 on the network (rather than having owners limiting their time on the network) they offer a better service - this will be key to breaking Uber's dominance in year 5 and keeping AV competitors out of the market.

In year 1, you will go on the Tesla app to find there are no vehicles in your area. You return to Uber. Year 5 - different story.

Mission is to accelerate EV adoption - not selling cars maximises this. Also, making the maximum profit will accelerate further.

 

[Buckminster]

Played around with back of the napkin numbers for using autonomous vehicles to deliver food. Just revenue, did not dive into margins. It's pretty obvious that margins would be far higher with autonomous vehicles, especially if small purpose built versions are created (such as an Arcimoto sized vehicle).

CAGR of the food delivery business is 4.3% and cheaper services would no doubt increase that. This isn't a huge business, but 5 billion is 5 billion, particulary if deliveries can be timed along with customer pickups. (car picks up pizza, drops off, and then picks up a nearby rideshare saving time/money)

View attachment 647368


How much of that is flexibility though? Say you like to drive to your vacation home or visit grandma 2 hours away. That's prohibitively expensive and a hassle with an Uber. Won't be nearly the same issue with RTs.

 

[Buckminster]

James slightly frustrated with Dave's conservative nature.
Neither expecting Tesla to own the robotaxis. Why would Elon leave that kind of money on the table? Why would Elon allow owners to lock away their vehicles preventing the removal of ICE Ubers from the CO2 runaway train?

 

[Buckminster]

Part 3 of James Douma valuation.

I really don’t get it. Tesla would basically be handing over most of the profits to fleet operators. I cannot see Tesla doing that. Much more likely would be a high upfront cost (e.g. every Tesla sold has a mandatory > $30k FSD package) PLUS Tesla gets 30% of the revenue, or Tesla operates their own fleet and keeps ALL the profit. Tesla has never ceded that kind of profit to any supplier or partner.

in James’ model a new Tesla buyer could purchase a $40k Tesla and earn $77k revenue which is mostly profit, the 1st year alone. Those money making opportunities just DO NOT exist in the real world.

Would love other people’s take. One other small nit is that he showed Global production, but then the rest of the calculations were on U.S. alone.

 

[kbM3]

I do think a Robotaxi available for work (/charging) 24x7x365 can make 30K per year as claimed by Elon...

It is easy to see why a Robotaxi owned by Tesla is worth 100K, that is essentially 3-4 years of operation.

For a privately owned vehicle what is the split between Tesla and the owner?

I say minimum Tesla take is 20% of 6K per year, in theory the owner might make 24K per year, but only when the car is available 24x7x365

In that model, cars are an appreciating asset but I'm not sure the FSD software license is worth 100K.

Consider a 50/50 split, Tesla would make 15K per privately owned vehicle....they could then lower the cost of FSD

Where the car is only used privately, or available to the Tesla limited hours, the situation is tricky... 100K for FSD seems excessive.

There is lots of potential here, but getting the right business model is tricky...

Consider instead if after a cut off date it was no longer possible to buy FSD outright and FSD could only be leased.
That 'cut off' date is a good way of getting buyers off the fence...

For a lease the price can be higher, I suggest $6K per year or $500 per month... with a $1 discount for every hour spent working as a Robotaxi in the Tesla network, but only in the Tesla network, up to $400. So 400 hours or more working as a Robotaxi, FSD only cost $100 per month. With that discount, Robotaxi income is split 50/50, but the discount is a strong incentive to go with the Tesla network.

For a private user $500 per month seems excessive, but if that owner is trading in a Tesla with FSD, a special exemption applies and they are able to lease FSD for a more reasonable price. We can extend this to some sort of referral program, where both parties get a discount on FSD.

Somewhere there exists a fair model where Tesla makes a lot of money, but customers particular long term loyal customers are treated fairly. You might not like my model, but it isn't easy to come up with a model.

FSD is worth $50k+ because an asset that can generate $30k profit per year is easily worth over $100k. An asset that can generate $30k profit per year at $50k would be an even more insane 60% return and waiting lists for those assets would quickly grow to 10 years plus.

If Tesla has a huge demand for commercial robo-taxi money printing machines, they would be violating their mission to sell consumer vehicles instead. A robo-taxi displaces 4-5 ICE cars and generates more cash for faster factory buildout.

Practically, I think Tesla will start making dedicated robo-taxi vehicles and de-emphasize (but not immediately stop selling) personal vehicles. Probably higher prices and forced FSD bundling. You can buy a Tesla for personal use but it will cost you.

The real question is market size at various price points.

We have data at $2.50 / mile from taxis/Uber/Lyft. I would assume the $1.00 / mile market would be many times larger, and $0.50 / mile would be larger than the entire personal automobile market.

From Autonomy Day, Tesla thinks total all in costs < $0.18 / mile. At their super conservative 50% utilization that becomes $0.36 cost per billable mile. assuming they can at least improve to 75% utilization that reduces all costs to $0.24 / mile

At $1.00 price that leaves $0.76 per billable mile profit (1M mile life * 75% billable * $0.76 profit per mile = $570k lifetime profit per car to be shared between the operator and Tesla) . I would assume Tesla is going to capture the lion’s share of that profit (maybe $0.50 / mile or more likely a higher upfront cost and a lower per mile fee.).

 

[kbM3]

Alternative view on costs.

What’s the applicability? Obviously from their statement that HAVs have a much higher upfront cost, they’re making the erroneous assumption that only the Waymo approach works.

 

[kbM3]

Dave and Rob talk about FSD deliveries. Getting the parcel to the front door could be a significant challenge. It might be that delivery drivers will remain for a good while - they can switch off and watch Netflix en route and then just deliver to the door. Not a bad gig.

Front door is not necessarily needed for a lot of functionality. Dave is just picturing humanoid robots displacing delivery humans. Reasoning by analogy.

A specialized vehicle that you schedule to show up in your driveway at a given time, which opens a compartment containing your food or package, solves much of the issue, in a much more economical and lower tech fashion.

 

[adiggs]

I think robotaxis will first be used for pizza delivery. Then last mile logistics. Then farm or fishery to home where speed is of the essence. Eventually you get to a point where the cost of logistics (transport, warehousing, scanning, picking, unloading, loading etc.) is more expensive than the cost of shipping one item from factory to home.

Take Nike as an example. They have pulled their products from Amazon so that they can control the entire experience. Eventually, you will be able to order a pair of personalised sneakers and have them shipped overnight from a nearby micro factory direct to you in a Nike branded vehicle.

James slightly frustrated with Dave's conservative nature.
Neither expecting Tesla to own the robotaxis. Why would Elon leave that kind of money on the table? Why would Elon allow owners to lock away their vehicles preventing the removal of ICE Ubers from the CO2 runaway train?

A few issues I see standing in the way of the particularly bullish forecasts for robotaxis and robotaxis for last mile delivery (and new delivery models).

The first is that some of these forecasts look like a significantly larger number of vehicles on the road. Vehicles on the road are dependent on the roads being subsidized by society. An obvious reaction by society is to stop externalizing this cost and shifting the value to Tesla (or other robotaxi operators) through an annual or per-mile tax.

To be clear this isn't the road damage subsidies that society provides to the trucking industry. These would be effectively new subsidies for the incremental congestion.

If we see robotaxis directly replacing current road usage with little incremental new usage then I see actual road usage and congestion going down as vehicles can do a better job of going the same speed on the roads and providing more complete usage of the roads.

Mostly I mention this as friction for use cases around robotaxis replacing chunks of the trucking industry. There is a similar friction for robotrucks getting their costs down low enough to replace train freight.


The next is it looks to me like this is all predicated on no other company bringing this functionality to market, thereby enabling Tesla to charge effectively whatever it wants for robotaxi functionality. I DO see a "golden era" of robotaxi functionality competing with ICE car with driver functionality, enabling the pricing power talked about here.

I don't see it lasting. Transportation is a commodity business which will drive prices down. In today's trucking industry, profit margins are closer to 2-4% with some annual spikes to 6%, and some lower.

More broadly transportation has always been a commodity business. If nothing else, even if the overall functionality is cheaper, the sort of monopoly on transportation being contemplated here will be resisted by society. I see the resistance in several forms, but the two biggies that immediately come to mind are antitrust and taxes. Antitrust is interesting as the way I understand it, different countries define it differently. In the US the key test is harm to consumers - a lower cost of transportation will be hard to argue US consumers are being harmed. It is my understanding that in Europe harm to competition is also considered and that will be an easy argument to make.

The other obvious mechanism will be taxes. Just as we're contemplating a monthly subscription model for FSD I would expect society to start a fixed tax ($x / year / vehicle) or a per-mile tax (better accounts for partial and full usage in a robotaxi network). Either way it will simplistically provide for society to share in the financial benefits. I'd expect it to be sold as road and infrastructure improvements that are critical to robotaxi functionality, and maybe even as avoiding externalized costs of the functionality.

 

[Doggydogworld]

What’s the applicability? Obviously from their statement that HAVs have a much higher upfront cost, they’re making the erroneous assumption that only the Waymo approach works.

Their calculations assume $15k vehicle cost. Their medallion cost is way too high IMHO (250k, I think), but it's silly to think cities won't take a cut. Especially since Robotaxis almost double total road miles. They also have (very high) safety/oversight costs and include costs for cleaning, deadheading, insurance, etc. which Tony Seba types tend to ignore. I find comprehensive studies like this to be a useful starting point. I can easily adjust their numbers vs. starting with a weak study that ignores or wishes away many of these ancillary costs.

 

[MC3OZ]

If Tesla has a huge demand for commercial robo-taxi money printing machines, they would be violating their mission to sell consumer vehicles instead.

My hunch is they will do both.

Just taking the US as an example, I expect Austin to produce all of the specialised Robotaxis. (EU and China would use local factories).
So the ideal mix is Austin churns out 2,000 Robotaxis per day with LFP batteries, Model 3 and Model 2.
So the Robotaxi fleet expands at the rate of 2,000 per day, with each city location starting out with 2,000-10,000 Robotaxis.

Taking the 2,000 per day model, these need :-

• Quick clean once per day - 8 hours shift - 200 per hour
• Deep clean once per week or as needed - 8 hour shift - 50 per hour

Even if they have fancy Robot based cleaning, they need to install it in every suitable location.
Robotaxis without cleaning will rapidly get a bad reputation.

The reason to mostly use Model 3 and Model 2 is costing < 30,000 they should make 30,000 per year.
Austin might start out with a run rate of 1,000 per day and eventually ramp up to 4,000 per day.

The remaining cars Model Y/Model S/X, Cybertruck etc, are still mostly sold to private buyers, this puts a floor under the quarterly earnings while the Robotaxi fleet ramps up. This approach also de-risks the adoption rate, if 2,000 Robotaxis become available in a new city, they don't hit 100% utilization on Day 1, it takes some time for customers to become aware of the service, decide to try the service, and decide they like it. Tesla could speed up Robotaxi adoption via advertising. The most cost effective advertising is a wrap on each Robotaxi advertising the service, rates, wedsite, app etc.
So when Robotaxis don't have jobs they can cruise around, advertising... they can park in visible locations, advertising.

With Tesla eventually making up to 20 Million vehicles per year, we may find that 10 Million of these are Robotaxis, or that there is a steep part of the ramp where Robotaxis are more than 50% of total car production.

Using LFP batteries for Robotaxis makes sense on many levels, but in particular it allows higher volumes of production with little risk of being cell constrained or hitting issues with raw materials supply.

 

[MC3OZ]

I'd expect it to be sold as road and infrastructure improvements that are critical to robotaxi functionality, and maybe even as avoiding externalized costs of the functionality.

Another edge (use case) Tesla Robotaxis wil have is operating in Boring Co tunnels.

The can possibly avoid road taxes for for some Robotaxis if the never/rarely travel on surface roads, or perhaps they can deduct underground miles travelled.

The secondary advantages are:-

• Travel by tunnel is faster and more predictable.
• Travel by tunnel avoids surface congestion, and reduces surface congestion.
• Less need for upgrades to surface roads.

So I think Tesla/Boring co is well place to negotiate a deal with many cities that mutually beneficial.

But I do think your point about taxation and monopolies stands, I also agree that there will be competition, and margins will be lower than many expect.

We should also see that selling entertainment and services to customers travelling in Robotaxis is profitable.

Overall Robotaxis accelerate the mission, they diversify income streams. more income is recession proof recurring income, overall the company is less dependent on making and selling products.

 

[Buckminster]

Good watch for Super bulls. First half is most relevant.
Warren is becoming the super bull standard bearer.

Dave got owned by Warren here imo.

Dave focused on $0.25/mile. Tesla will divert robotaxis to the more expensive cities initially. There were 48 worldwide cities (in 2017) where the cost of a taxi is more than it is in NYC. Highest is $7/mile = x28 higher than Dave’s long term estimate of $0.25/mile…

World taxi prices: What a 3-kilometer taxi fare is in 88 big cities in 2017
$9.08 – $11.80 Tokyo, Japan
$10.11 – $13.48 Berlin, Germany
$10.39 – $15.58 London, England
$10.39 – $10.39 Amsterdam, Netherlands
$11.00 – $15.00 Honolulu, United States
$11.00 – $15.00 San Francisco, United States
$11.22 – $11.28 Copenhagen, Denmark
$11.24 – $16.85 Paris, France
$11.24 – $16.85 Helsinki, Finland
$11.24 – $15.73 Monaco, Monaco
$11.69 – $18.18 Edinburgh, Scotland
$11.78 – $15.31 Oslo, Norway
$12.00 – $16.00 Los Angeles, USA
$12.16 – $19.26 Reykjavik, Iceland
$13.48 – $16.85 Nice, France
$15.00 – $20.00 Miami Beach, United States
$18.56 – $24.74 Zurich, Switzerland

Robotaxi price per mile could be more than a normal taxi.
1) FSD will be fun - like going to 6 Flags for many many rides - for $7bn people...
2) Not having a taxi driver is a blessing. More freedom, privacy etc.