In defense of Musk I think he sincerely believed FSD would be solved within a couple of years of the Model 3 introduction. So buy the car today and get a robotaxi in a couple of years. Obviously didn’t work out that way. Such is life.
I’ve long asserted that we should think of electric vehicles as computers on wheels. In principle, free over-the-air updates should help Teslas retain their value, but that has been overwhelmed by the rapid advancements in battery technology.
When I bought my Model Y, I did so fully expecting it to become “obsolete” in a year or two. But I figured it was good enough for me to be happy driving it even if newer models charge faster with longer range. And by the time I want to replace it, I figure that the decline in trade-in value will be more than compensated for by tech-driven declines in new BEV prices.
It also wouldn’t surprise me if BEVs become like cell phones where enough money is made from subscriptions that the hardware can be heavily discounted. Something like agree to a three year FSD subscription and you get the $25K Tesla for half price.
Heh, I don’t think an increase in utility or functionality is synonymous with an increase in value.
My computer software seems to update itself every month with OTAs and I certainly can’t sell it for more than what I paid for it.
At some point, Microsoft is likely to tell me that they will no longer update my software. I might still get a lot of utility from it but I doubt it will have a lot of value for resale.
Hawkwin
Who wonders what would happen if Tesla eventually says it will no longer update the software for Model S that are 10 years or older.
No. No. No. For one, I won’t spend $50k on a cell phone or computer, and I won’t spend that type of money on a car that depreciates like one either. That’s just stupid. And I’m buying a car as a means of transportation. People who claim “cars are becoming computers” I just don’t understand.
Maybe you should show how this math would work. IIRC, FSD is $200/month. So 36 months would be $7200. And you’d get $12,500 of a discount and pay no interest on that? Tesla would have to be charging at least $350/mo just to break even. And probably $500/mo to actually have a net gain considering some full priced $25K car owners would have bought or rented FSD anyway.
This is clearly part of the phenomenon. I think carmakers will have to make certain critical modules upgradable someday. Tesla has done that to a small extent in that people with MCU1 were able to upgrade to MCU2 (and [some] people who signed up for FSD got the MCU2 upgrade included in the price).
I wonder if it will affect leasing more than buying? If you lease, most of the time you turn in the car after 3 or 4 years. But if you buy, you may keep it 8 years, and many keep it for 10 years.
If you are worried about resale value consider buying used instead. There is a certain amount of market irrationality regarding used EVs. One is that belief that next year’s model will be better (likely true, but likely not that much better) and worries about battery life, which not an issue. So there is a substantial discount for buying used in exchange for very little performance loss.
Maybe. A few months ago people here were saying the slump was because people were waiting for major improvements in batteries or Tesla was opening up its chargers or something.
At any rate, that rapid s-shaped adoption curve doesn’t appear to be as expected.
If you are just buying a car for transportation, why worry about depreciation? Do you worry about the resale value of your refrigerator or table saw? Just drive it until it doesn’t work anymore, at which point depreciation becomes mostly irrelevant.
Also curious as to what you think the Tesla depreciation rate happens to be. Sure, if you buy it at its highest price you probably won’t be happy in a couple of year, but that’s true of any vehicle. Now seems like a great time to buy a Tesla with price cuts and the tax break.
Here is the Tesla Model 3 depreciation curve according to Caredge:
It’s not that hard. People want adaptive cruise control, automatic emergency breaking, infotainment systems, backup cameras with proximity warnings, etc. Governments are demanding higher emissions standards that require computer control of internal combustion. That’s why about 40% of the cost of a new car is from semiconductors, computer chips, and software. For the past decade, the major improvements in vehicle performance have come from software. Electrics are accelerating that trend.
I’m guessing that an FSD that takes you to Level 5 will be worth twice as much as the current version, so let’s just round that up to $5000/year. And lets assume the car’s production cost is $20K. The customer can pay $25K for just the car or $27.5K for the car and three years of autonomous driving. I suspect most will pay the $27.5K. In return, Tesla loses $7500 on the hardware but gains $15K in software profit.
The difference is that a car costs tens of thousands of dollars more to own, and (at least) thousands of dollars more to maintain. Also refrigerators and table saws don’t go in and out of style the way cars do. Which is why depreciation is a real thing for car buyers, and not for table saw buyers. You can twist the argument if you like, but using equivalent analogies instead of inapt ones is likely to make for a better conclusion.
I don’t know where Caredge gets its figures, but KBB says a 2020 Tesla that was $40,000 is now worth $25,000. And a 2020 Toyota Camry that was $28,000 is now $22,000 so yeah, I see a difference. Kelly’s is pretty much the gold standard, isn’t it?
And all those things are at virtual parity and have been for the past several years. They’re not “depreciating”, they’re the same. This is opposite of computers, which systems continuously improve, get faster, etc. The rear camera from 5 years ago could be switched with one from today and most drivers would have no idea they’d been switched. A computer needs to “stay current” and accept new inputs and devices; a car is a car. Almost nobody is switching out a 2020 car with a 2025 entertainment system (or drive-by-wire system).
The addition of FSD, if it comes, will certainly enhance the resale value of a car, but only if the cost is reasonable in relation to the used cost of the car, else people will just buy a new one. Relatively few will want to pay up bigly for an old car with new software when a new car is just around the corner, cost-wise.
This sounds high, even for EVs. In an ICE the dollar cost is between $500 and $1000. I’m sure it’s much higher in an EV, but I would be surprised to find it’s 30-40 times higher.
Your numbers only reflect the cost of the semiconductors. Those semiconductors require an infrastructure of wires, sensors, and other electronics. Consumers want cars to do more things for them to both increase safety and efficiency. One can imagine the electronics required in hybrids to coordinate two separate drivetrain systems.
Perhaps it is best said in a Motortrend artical:
“Once, software was a part of the car. Now, software determines the value of a car,” noted Manfred Broy, emeritus professor of informatics at Technical University, Munich and a leading expert on software in automobiles. “The success of a car depends on its software much more than the mechanical side.” Nearly all vehicle innovations by auto manufacturers, or original equipment manufacturers (OEMs) as they are called by industry insiders, are now tied to software, he said.https://www.motortrend.com/software-issues-hurdles-self-driving-cars-evs/
Well sure, if you’re going to include entertainment systems and the like. And yes, brakes and acceleration now flow through chips, not mechanical linkages - but again those are largely at parity. There is almost nothing my 2016 van doesn’t do that our 2024 cars do, save the Bose sound system.
Who clearly has no incentive to have the point of view that software controls the value of the car. Sheesh. I have to ask, if Chinese automakers are turning out $10,000 and $12,000 cars and disrupting Tesla from below, and making a modest profit, how are they doing that if the cost of software is already $20,000 of Tesla’s price?
That is also not surprising. ICE cars reached “feature parity” long ago. Everybody got AC, power windows, power steering & brakes, and so on. At some point your reach the end of the line in a particular form, and have to wait for a whole new paradigm to continue. (Think flip phones which were pretty much the same until smart phones came along. Then there was an explosion of new features —- until near parity, again.)
Frankly, with the large carve out of driver-assist features (including someday FSD) including lane-follow, smart cruise control, etc. we’re near that now. Bigger screen, smaller screen, automatic wipers, you can get that stuff in virtually any product line you want. The improvements are going to come in battery chemistry, charging speed and capacity (yes, all of which will require “software”) and in driver assist. That will continue for another 5 years or so (hence my believe in accelerated depreciation until that cycle is complete) after which cars will revert to marketing (price & image) and style. Again. As they have, repeatedly, for the last century.
Government subsidies and they aren’t making a profit. Not on BEVs.
Just want to add in edit that this article nicely defines what is meant by a software-defined car and a computer on wheels:
Today, manufacturers of electrified vehicles use software not only to improve safety, comfort, and infotainment features, but also to enable increased electrification, ADAS/AV features, shared-mobility services, and interconnected transportation systems. In the future, vehicle functionalities and the related software will be developed independently from the hardware on which it operates and will be updated throughout its entire vehicle lifetime – leading to the “software-defined vehicle” (SDV).
To make this happen, the automotive industry is learning software architectures and development methods well-known in the telecommunications and computer industries and adopting them in a more sophisticated way. The challenge is to transfer the advantages of the tech industry to more intensive requirements of the automotive industry, and at the same time to orchestrate the necessary design changes and hardware developments in a complex supply-chain environment. Zonal Electrical Architectures Cut Vehicle Wiring-System Cost, Complexity - Tech Briefs
I believe the Cybertruck is Tesla’s first foray into zonal architecture on a large scale. The lessons learned there will be incorporated into the next generation vehicle. I don’t believe anyone else is close to mass producing a zonal architecture vehicle, which should reduce complexity while increasing functionality. This is why Tesla is considered to be at least 5 years ahead in this technology.
Sure, but it only makes a big difference if you are trading in a 5 year old or younger car. After 5 years or so, most vehicles not named Toyota have lost almost half their value. And trading in such a young car is almost always a bad financial decision anyway since the bulk of the depreciation occurs in the first few years. You are far better off buying a car you like enough to keep for 8+ years than compromising for better depreciation then trading up within 5 years.
Would also point out that it is unlikely that Hertz is going to dump thousands of used Teslas into the market every year. Tesla’s depreciation over the last year or two was probably anomalously high.
That Model 3 depreciation graph is bogus. The oldest Model 3’s are 7 years old and the graph goes to 10 years. There are very few even 6 years old as well.
Doubtful…but even if true is many years away.
And it is not yet proven that people buying a low end car will pay top dollar to have the car drive for them.
