Seems to me, that is unlikely. It takes a huge effort just to geofence one city. Including Two Bumps, Nevada seems undoable. Not to mention some obscure back corner of Serbia. Note that the robotaxi concept doesn’t require covering everything since it only has to work in the places one wants to offer the taxi service.
Interesting comment. With lvl 4, I can do something to get around this. What will lvl 5 do? With no steering wheel, I might be stranded.
If that was true than Tesla would no longer be gathering data. They would already have enough and instead be working on perfecting it all. But since they are still gathering data there must be some clear need for it.
Andy
A Level 4 car can travel anywhere. It just can’t self-drive everywhere.
If Tesla sells their Level 5 option for the same price as other automakers sell their Level 4 option, then of course will be a superior option. But not large enough to drive people’s consumption choices far from where they would otherwise be. Which is the context of this discussion - whether Tesla can get 8-10 million people a year to buy the same model, when decades of consumer preferences show that they prefer variety in their purchases.
If Tesla was the only car that had self-driving, then maybe that could happen. It’s easy to see people whose first choice wouldn’t be a Model Y deciding to go with it anyway, if it’s the only self-driving car out there. But in a world where most cars could self-drive nearly everywhere you’d want to go, that’s just not going to be that big of a pull.
How much Tesla can make from either licensing or as a premium to their own products depends, of course, on what consumers will pay for the feature. In a world where Tesla was the only one that could offer that feature, they’d probably be able to get a fair amount of money based on their monopoly access to the option. But in a world where multiple automakers are competing with each other on “close enough” versions, you’re just not going to get as much of a premium. The effect on their market cap will be dramatically lower if Tesla doesn’t have that exclusivity.
You assume because you have a steering wheel you can go around the branch? Exactly what a level 5 car would do only it would do it for you.
Andy
Right. That’s why I said that if the geofence is big enough to encompass everywhere (or virtually everywhere) you might want to drive, there’s no material difference.
Very few people will need to drive to Two Bumps or an obscure back corner of Serbia - ever. Fewer still will have that be a material part of their normal driving routine. So a Level 4 system that covers most major metro areas and the major routes between them will be good enough - the handful of times you might need to drive outside that area, you just won’t be able to self-drive. Your car will still get you there, it just won’t do it by itself.
Between any two points within the fence, though, your driving experience in a Level 4 car will be indistinguishable from a Level 5 car.
I believe if Tesla get’s to level 5 first, and licenses that to other automakers , that everyone else will just stop trying to perfect theirs. Especially if they license it at the correct price. Just like their charging network. While there are some companies still building out charging networks, Tesla, is the standard and the others have pretty much given up.
Andy
That’s certainly plausible. But that’s along the lines of what I said upthread. For Level 5 to be a massive money-maker for Tesla, they have to get there first. After all, if someone else gets there before them, and someone else is already out there licensing to all the other automakers, then Tesla hitting Level 5 isn’t going to do much for them. And indeed, in that scenario Tesla could be the one giving up its efforts and just licensing from the other company.
I think the same holds true for a reasonable robust Level 4 system. If Tesla isn’t fast enough in getting to Level 5, and Level 4 self-driving becomes widespread and commonplace, they’ll lose much of the economic benefit to getting to Level 5. Not all of it, of course - but they won’t get nearly the economic boon as if they had the only self-driving system.
And even if they are first, your observation that the license has to be “at the correct price” is important. The amount of money Tesla can get from self-driving isn’t unlimited. It might not even be astronomical. Consumers will only pay so much (hi, reduced FSD price announced today!), and other companies will only pay so much for a license because of that. If the price is too high, Tesla will have the market for that option to itself for a time, but other companies will keep doing their best to come up with a competing product.
If the actual price for self-driving ends up being closer to the $2-4K that advanced and adaptive cruise systems go for today, rather than the $15K or more that Musk has alluded to from time to time, FSD won’t be a massive enough home run to triple the stock price.
Uh, why would Tesla license to anybody? It is already known, once Tesla has a documentable Level 5 AV, they will stop selling cars to the public. Why crush their own system of making money by licensing the Level 5 system to anyone? They won’t. TaaS will become the de facto standard because it will be cheaper than buying a car (along with paying all the associated costs of owning and operating a car).
They will always need more data to find smaller and smaller edge cases.
1 Like
The people doing geofencing are also relying heavily on maps.
People may not need to go to Two Bumps NV as a destination very commonly, but they need to go through the empty space between urban centers.
Not an issue. Those empty spaces are actually among the areas that are being mapped first. Ford’s BlueCruise advanced driver assist relies heavily on high-definition mapping - and they chose to commission hi-def maps of all the nation’s limited access highways. It’s a hands-free assist for highway driving. Okay, not all the highways - apparently they’ve got about 97% of them mapped. But close to all.
Even besides that, given the enormous commercial applications of self-driving for long-haul trucking, we would expect that the major routes running through the empty space between urban centers to continue to be among the most well-mapped areas of the country.
As long as the “urban centers” are connected via a major roadway (highway, freeway, whatever it may be called), there will be enough daily traffic to maintain the “in between” areas. That was the whole idea of those major roadways between urban centers–to AVOID the small town slowdowns and keep traffic moving as fast as possible.
What about the word “geofenced” do you not understand? A given place is either inside the fence or its not. If it can’t operate outside the fence, then it can’t go there. Even if one supposes a system of no driver inside the fence, driver outside the fence … how attractive is it going to be to buy a car capable of driverless operation and then not be able to use it on long trips … one of the easier driving environments to automate (not counting amusing edge cases like the moose crossing the road).
Once airplanes became available how long did it take for people to adopt them for travel? How many people were saying they would never fly? This will be much faster in todays age and time.
Andy
It is if one is using an automated system that relies on a map like the current geofenced systems.
I don’t have any idea how well the Cybertruck will sell. Maybe a lot, maybe not. If not a lot then I think they will modify it and/or come out with a more mass market offering, certainly by ~2030.
As for all this level 4 and 5 discussion. I think the importance of this is way overblown. As difficult as it is to do the self driving software, eventually it will be a small-ish dollar value compared to making the cars themselves no matter what Musk or others say. Either several companies will all have good enough level 4 or 5 or the winner takes all company will license it out, IMO.
Mike
I guess you didn’t read the article I posted in the string.
Andy
What is said and what really happens are two different things. One key point was NOT made: If Tesla would license it to others AND help make their cars also Level 5 AV. If there are limitations, what are they? Not mentioned anywhere. Thus, take the claim with a truckload of salt.
It doesn’t change the problem all that much if you assume four models, rather than three. To get close to 20 million total sales, you’d still need more than 5 million units of sales for each of them - and since they won’t all sell equally well, probably closer to 6 or 7 million for at least one. The most popular cars in the world - ever - have never sold anywhere close to those numbers. The best single year for any model was about 1.6 million. Those figures are close to the size of the total market for each of their vehicle segments. I mean, the entire global pickup truck market today is “only” about 3.8 million units (about half in the U.S.). Is Tesla really going to capture 100% of the global market share of pickups within six years? With just one model of truck?
For Tesla to achieve these kinds of sales volumes with only four models, they’ll need to have a completely different pattern of consumer behavior than has predominated for the last half-century.