I don't think they will sell FSD on the 3 unless it has surpassed regular AP capabilities. If that doesn't work they will be in big trouble with the sales of the 3. Those buyers will be even less patient than the current crop because for them, a $4k option will be a much tougher pill to bite if the functionality doesn't work.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
pilotSteve
Active Member
Kind of like the current class action lawsuit that is being handled by the same firm as dieselgate...
They only have to make it work on M3 with some HW (shipped initially or upgraded later). The small number of HW2 MSs and MXs don't matter at all and the small lawsuits can be settled easily.
I bought FSD up front in order to commit Tesla as much as possible, betting that they could make something work with the sensors and knowing that it is likely that a processor upgrade will be required. I see two cases where I win: 1) Tesla delivers with AP2 HW as shipped, in which case the upgrade price will be raised dramatically, or 2) Tesla needs to do HW upgrades, in which case it is possible they will do the upgrades for pre-paid customers and will simply withdraw the upgrade option (or price it prohibitively high).
If they never deliver FSD for AP2, I figure they'll have to give the money back, so not a terrible outcome. But I think it is existential for the company to make something that can be labeled "FSD" work on M3 now. I only hope AP2 on my MS gets L4 of some sort.
Tesla has made pretty strong statements on their website that the AP2 sensors are sufficient to support full self driving, contingent upon software validation and regulatory approval. It's very likely Tesla's development group did enough testing of the new sensors to have confidence to make this statement. I've been responsible for major software projects - and that's what I would have done before making this type of commitment (and the FSD video on Tesla's website could be a demonstration of that testing).
The biggest unknown is the amount of software/processing required for real-time object recognition and implementing the rules for handling that information and generating driving commands - at a safety level higher than a human driver. If the NVidia Drive PX2 processor board doesn't have enough horsepower, it should be relatively easy to swap that board with even faster processors in the future.
We purchased the FSD option for our new S 100D because we believe Tesla has a high probability of getting there with the AP2 hardware - eventually. And even before Tesla gets regulatory approval, the FSD software will likely run in a super EAP mode, using all of the sensors, and doing most of the driving, based on navigation routing, only requiring a driver to maintain contact with the steering wheel. And, if Tesla does need hardware changes to get the regulatory approval, pre-purchasing FSD should protect us from any future price increases for activating FSD.
The biggest unknown is the amount of software/processing required for real-time object recognition and implementing the rules for handling that information and generating driving commands - at a safety level higher than a human driver. If the NVidia Drive PX2 processor board doesn't have enough horsepower, it should be relatively easy to swap that board with even faster processors in the future.
We purchased the FSD option for our new S 100D because we believe Tesla has a high probability of getting there with the AP2 hardware - eventually. And even before Tesla gets regulatory approval, the FSD software will likely run in a super EAP mode, using all of the sensors, and doing most of the driving, based on navigation routing, only requiring a driver to maintain contact with the steering wheel. And, if Tesla does need hardware changes to get the regulatory approval, pre-purchasing FSD should protect us from any future price increases for activating FSD.
Excellent point on exposing / not exposing yourself to the cost of a potential hardware upgrade. When I did the financial math, I also included property tax. In CT we pay a percentage of the value of the car in property tax, around 3% annually. With pre-purchased FSD, the purchase price of the car is higher and hence the property tax as well. Even with that included, the Net Present Value (NPV) of pre-paying was still more attractive.
I do think there is one alternate scenario where it may not be beneficial: right now FSD is $3,000 prepaid and $4,000 afterwards. I can envision how FSD, if very successful, will get an increased price point (up from $3K/$4K) because of the additional value it offers, but I can equally envision a scenario where the FSD prices get dropped. Why? Two-fold:
A) it would unlock a lot of 'free' money for Tesla: current owners upgrading would be an immediate pickup in the quarter. Similar to the 60kWh to 75 kWh upgrade price drop, I bet a lot of current owners jumped on that (me included) -> immediate pickup for Tesla. A pickup like that is hard to say no to for any listed company.
B) once the software is developed there is no additional cost to Tesla to install or not install it on a given car, it's all sunk cost. Therefore it becomes a simple optimization, numerical example: do I charge $3,000 and get 100K customers ($300M revenue), or do I charge $2,000 and get 300K customers to adopt it ($600M revenue). My costs are the same, but my revenue outcome can be very different. Obviously these numbers are made up, but I'm sure they'll do a total revenue optimization exercise like the above, and the answer of that exercise may well be a lower price point.
Time will tell
B) once the software is developed there is no additional cost to Tesla to install or not install it on a given car, it's all sunk cost. Therefore it becomes a simple optimization, numerical example: do I charge $3,000 and get 100K customers ($300M revenue), or do I charge $2,000 and get 300K customers to adopt it ($600M revenue). My costs are the same, but my revenue outcome can be very different. Obviously these numbers are made up, but I'm sure they'll do a total revenue optimization exercise like the above, and the answer of that exercise may well be a lower price point.
Time will tell
That's a really good point that I didn't consider. I must admit: I'm blinded by my base assumption that as soon as FSD delivers any sort of technology, I'd immediately desire it. I'd imagine TACC/AP stopping at red lights and stop signs would roll out first, along with increased intervals of hands off the wheel on the freeway, etc.
TexLaw
Member
I'd hardly ever advise anyone to buy vaporware at much any price. However, I went ahead and ponied up for FSD when I placed my order. If actual FSD does happen, then $3,000 could be a real bargain. If it never shows up at all, then I'm fairly sure TSLA will make good on it through nothing worse than a refund (short of TSLA bankruptcy--not out of the question).
I also figured that, hell, I've already paid in large part for FSD. It's not like Tesla put all that hardware in the car for free, crossed their fingers, and hoped to goodness that someone might pay for it somewhere down the line. I've probably already paid for much of the software development cost, as well (especially to the extent it overlaps with EAP). The $3k is just to throw the switch on additional functions.
I also figured that, hell, I've already paid in large part for FSD. It's not like Tesla put all that hardware in the car for free, crossed their fingers, and hoped to goodness that someone might pay for it somewhere down the line. I've probably already paid for much of the software development cost, as well (especially to the extent it overlaps with EAP). The $3k is just to throw the switch on additional functions.
Perpendicular parking and auto wipers and speed limit sign reading are NOT EAP features. They are AP features. AP1 had these. People that paid for EAP are expecting things like summon in complex environments and auto lane changing on the highway with no human interaction, as well as exiting the highway.
Tesla needs to deliver an EAP feature before they do FSD or people will be really pissed.
Remember that Tesla charged differently for AP and EAP for 20K customers that bought when AP1 existed but were given HW2.
The car I am trading in has HW1, and I am driving a loaner car with EAP. It just recieved an update last night that seems to have given it the majority of HW1 functionality including perpendicular parking and auto high beam. Today trying the autopilot in traffic there was more lane wandering than HW1, and I think they will obviously continue to refine that down like they did with HW1. Guessing that they would incrementally release functionality as it has been validated, how soon would you venture that you would begin to see a divergence in functionality between HW1 versus HW2? Would you think it is realistic to expect that as early as this summer is a possibility?
I think it is realistic to see that. I would say they are about 90-95% to AP1, and given the rate of updates, I would guess that by July (in line with Model 3 launch), AP2 will be 105%+ of AP1. And possibly even start to see some FSD features start to pop up, and finally include all the additional cameras (still using 1, maybe 2?).
They also charged more for FSD on top of EAP. I don't see any need to give EAP something on top of AP(technically, they've already done this in allowing activation of AS at 0mph and identification of stationary objects) before giving the FSD folks anything.
I'm a FSD owner that doesn't care whether EAP features come or FSD. Just send them. I would LOVE it if my car allowed local driving and recognized stop and red lights. That would be so awesome and would allow me to really feel happy with the $3k premium. I don't really like visiting malls or large parking lots (I'm the guy that parks away from everyone else and walks). I don't mind the walking even laden with lots of groceries. I won't really care that much about advanced summon (I mean it'd be nice but stop lights/signs has so much more utility).
I think that reliably responding to stop signs and red lights is probably one of the hardest features to implement. After all, if it doesn't work perfectly at every intersection, it's just plain dangerous. Wouldn't anticipate that this would be an early-add feature.
I am only speculating about his availability because of early versions of ap2 autopilot contained stop sign recognition and the full self-driving videos clearly had that capability. I agree it is difficult but easy things like Auto dimming high beams have taken a long time where is more difficult things such as high-speed auto-steer came sooner.
RobW2428
Member
Gotta say, overall this thread has made me happier about buying FSD with my car . So thanks for that. This thread seems to have a nice mixture of facts and balanced opinions. Nice job everyone!
. So thanks for that. This thread seems to have a nice mixture of facts and balanced opinions. Nice job everyone!It's a bit off topic, but my educated guess is late summer / early fall, definitely by the end of the year. I would guess some new features would start to show up independently of AP2 vs AP1 "stability parity".
