Autonomous Car Progress

[daniel]

I found this article interesting re Mercedes Benz partnership with Nvidia... Tesla Is Still the Car Company to Beat. Just Ask Mercedes.

This quote stood out in particular:

Morgan Stanley analyst Adam Jonas says Tesla is the only company “fully monetizing its autonomous driving assets at scale.” In other words, Tesla generates actual money from its internally developed self-driving solutions.

Monetizing full self-driving by selling the promise of it. All companies monetize what they sell by selling it. Other car companies monetize whatever driver-assist features they're putting in their cars, either by offering the features as extra-cost options or by pricing them into the base cost of the car. Only Tesla will sell you what they don't have yet.

Tesla has the best driver-assist features you can buy today as an ordinary consumer. But the distinction is the quality of those features, not that Tesla is the "only" one monetizing autonomy features.

 

[mikes_fsd]

Monetizing full self-driving by selling the promise of it. All companies monetize what they sell by selling it.

There is some of that, but as more features are added, that price is most definitely going to go up.
So, it is both supporting the "promised future features" but also getting a deal on the features if you're one to keep your cars around.

All companies monetize what they sell by selling it. Other car companies monetize whatever driver-assist features they're putting in their cars, either by offering the features as extra-cost options or by pricing them into the base cost of the car. Only Tesla will sell you what they don't have yet.

Please do show me which car companies monetize their driver assist features?
There are car companies that pass the money along to MobileEye, but there are no companies that monetize their own features ala Tesla.

 

[Knightshade]

Please do show me which car companies monetize their driver assist features?
There are car companies that pass the money along to MobileEye, but there are no companies that monetize their own features ala Tesla.

GM.

Supercruise is their own system.

 

[diplomat33]

Please do show me which car companies monetize their driver assist features?
There are car companies that pass the money along to MobileEye, but there are no companies that monetize their own features ala Tesla.

Every auto maker offers driver assist features on their cars which cost extra to the customer. The only big difference is that Tesla sells direct to the consumer whereas other automakers use dealerships to sell their cars to the consumer.

 

[mikes_fsd]

Every auto maker offers driver assist features on their cars which cost extra to the customer. The only big difference is that Tesla sells direct to the consumer whereas other automakers use dealerships to sell their cars to the consumer.

My point was not that they do not have the feature, it is that they just pass the money along to a 3rd party like MobileEye.
But @Knightshade has corrected me, GM's SuperCruise is their own.

 

[daniel]

... as more features are added, that price is most definitely going to go up.
So, it is both supporting the "promised future features" but also getting a deal on the features if you're one to keep your cars around.

Which is not really relevant to the claim that Tesla is the only company monetizing driver-assist features.

Tesla excels in the features you can actually get. And it excels in OTA updates (my autosteer is better now than when I bought the car and new features are occasionally added even to my EAP package). But the only companies not monetizing their technology are the ones who are unable to sell their technology. Calling Tesla the only one to monetize such technology is silly. But Tesla is the only car company to successfully monetize technology they don't have yet, by selling the promise of it, and even successfully downgrading the promises made to people who bought earlier. When and if "feature complete" becomes available as a Level 2 system, Tesla will pretend that it has fulfilled its obligation to people who bought when "no driver needed" was the promise.

I harp on this because it saddens me that the company making the best cars a consumer can buy today feels it needs to sell promises it may or may not be able to keep, rather than just selling what it has and telling the world what it hopes to accomplish in the future. I really hope that Tesla is the first to full autonomy. But I can imagine a time when some other company is selling fully autonomous cars and people who paid Tesla for "FSD" years previously are still waiting.

 

[Soda Popinski]

My point was not that they do not have the feature, it is that they just pass the money along to a 3rd party like MobileEye.
But @Knightshade has corrected me, GM's SuperCruise is their own.

I would think no car company would offer ADAS at a loss. Either they're using it to increase sales by having a more capable car, which makes money, or they're paying MobilEye less than they're charging for the feature, making a profit, or both.

That said, I think we can all agree that since Tesla and GM developed ADAS in house, they're not passing money to a third party. Though I wonder if GM is making a profit on each SuperCruise system installed yet. Right now you can only get it on relatively rare Cadillacs for as a $2500 option, which has to support the R&D and HD map generation. It's coming out in about 2 dozen cars over the next few years, so I'm sure it'll be more profitable then.

 

[mikes_fsd]

I would think no car company would offer ADAS at a loss.

I think you are missing the point here.
If you offer ADAS packages on your car from a third-party, you are not monetizing your own R&D efforts but rather the 3rd party provider.
Yes, they still will mark it up and make a profit off it.

 

[Soda Popinski]

I think you are missing the point here.
If you offer ADAS packages on your car from a third-party, you are not monetizing your own R&D efforts but rather the 3rd party provider.
Yes, they still will mark it up and make a profit off it.

I think we're splitting hairs here. I didn't realize you were talking about their R&D efforts, just the features ready for consumers. I'll agree with that (besides GM of course).

 

[mikes_fsd]

I think we're splitting hairs here. I didn't realize you were talking about their R&D efforts, just the features ready for consumers. I'll agree with that (besides GM of course).

I would say it is more like separating wheat from chaff.
If you are just re-selling a 3rd party product, you might be able to mark up and make $$ off it, but you are not learning or contributing to the FSD solution at all.
And with the original quote, if you are not able to monetize your FSD/ADAS package as a developer of it, you are going to go the way of Zoox and get swallowed up by someone who has an idea and path on how to monetize it.

 

[Cirrus MS100D]

Don't forget that Waymo is also going to purchase 20,000 i-Paces and deploy that fleet of self-driving EVs as soon as <checks calendar...> a year ago!
Waymo’s self-driving Jaguar I-Pace vehicles are now testing on public roads – TechCrunch

 

[diplomat33]

Don't forget that Waymo is also going to purchase 20,000 i-Paces and deploy that fleet of self-driving EVs as soon as <checks calendar...> a year ago!
Waymo’s self-driving Jaguar I-Pace vehicles are now testing on public roads – TechCrunch

From your article:

"Waymo plans to roll the I-Pace vehicles into its self-driving ride-hailing fleet in 2020. The deal between Waymo and JLR is for up to 20,000 modified I-Pace vehicles to join the robotaxi service in the first two years of operation."

The covid pandemic obviously slowed things way down. But the plan was to start the roll out in 2020 with 20,000 robotaxis joining in the next 2 years, ie by 2022.

 

[Knightshade]

From your article:

"Waymo plans to roll the I-Pace vehicles into its self-driving ride-hailing fleet in 2020. The deal between Waymo and JLR is for up to 20,000 modified I-Pace vehicles to join the robotaxi service in the first two years of operation."

The covid pandemic obviously slowed things way down. But the plan was to start the roll out in 2020 with 20,000 robotaxis joining in the next 2 years, ie by 2022.

It's nice to see others racing with Elon for who comes in most over their promised deadline for putting lots of BEV robotaxis on the road

 

[diplomat33]

It's nice to see others racing with Elon for who comes in most over their promised deadline for putting lots of BEV robotaxis on the road

I would say Elon is winning that one. Elon promised 1M robotaxis this year and still has ZERO.

Waymo may be behind their 20,000 robotaxi promise but pre-covid, they at least had 600 robotaxis on the road.

 

[Soda Popinski]

I would say Elon is winning that one. Elon promised 1M robotaxis this year and still has ZERO.

Waymo may be behind their 20,000 robotaxi promise but pre-covid, they at least had 600 robotaxis on the road.

This does raise the question of who will get 1 million robotaxies out first. Hypothetically, if Tesla gets robotaxies out 3 years behind Waymo, I doubt Waymo will have 1 million of them by then. Tesla would just have to flip a switch and a bunch of consumer cars can become robotaxies. Of course, they'd only be part-time robotaxies...

 

[daniel]

I don't care who gets a million robotaxis or one robotaxi on the road first. I care about when I can buy a car that's Level 4 (or even Level 3) on my roads.

I don't see what's the big deal about robotaxis except as a showcase for where we're at with autonomous cars, or, if you're lucky enough to be somewhere they're operating, the chance to ride in one for fun. I don't like to drive, and the older I get the closer I get to being unable to drive safely. For the practical matter of transportation, a robotaxi is no better than a Lyft. Is the robotaxi going to be able to pick me up sooner or get me where I want to go faster? Probably not. A private car is a convenience. A private autonomous car is a convenience that I don't have to drive. Like having a chauffeur, but cheaper. If I was rich enough I'd have a chauffeur and I wouldn't need an autonomous car. But I'm not, so I want an autonomous car. A million robotaxis won't do me any good unless they are so abundant at my location that the waiting time is less than Lyft. And that ain't happening in my lifetime.

 

[diplomat33]

Frankly, 1M robotaxis is an arbitrary goal. The 1M robotaxis comes from Autonomy Day when Elon was touting that Tesla will have 1M cars on the road. And Elon was trying to make a point that the 1M Teslas all have the potential to become robotaxis as soon as Tesla finishes the FSD software. But we could just as easily set the goal to 100k robotaxis or 2M robotaxis.

Elon claims that they just need to flip a switch and every car will be a robotaxi but that of course depends on two big IFs. It depends on the current hardware being good enough and it depends on Tesla finishing the FSD software. Tesla still has a lot of work to do before the software is robotaxi-ready.

Waymo already has robotaxis but they need to mass produce them and expand then safely to new areas.

Personally, I don't think anyone will have 1M robotaxis any time soon. Not Tesla, not Waymo. Realistically, in the next few years, I think we will probably see Waymo deploy thousands of robotaxis though in various areas (but not 1M yet) while Tesla will have semi-autonomous driving that works everywhere with driver supervision but is not ready to be a robotaxi yet.

 

[Soda Popinski]

Frankly, 1M robotaxis is an arbitrary goal. The 1M robotaxis comes from Autonomy Day when Elon was touting that Tesla will have 1M cars on the road. And Elon was trying to make a point that the 1M Teslas all have the potential to become robotaxis as soon as Tesla finishes the FSD software. But we could just as easily set the goal to 100k robotaxis or 2M robotaxis.

While I agree it's arbitrary, it's also a good way to illustrate the scaling potential of their model, compared to robotaxies built and operated by a non-car maker.

 

[daniel]

While I agree it's arbitrary, it's also a good way to illustrate the scaling potential of their model, compared to robotaxies built and operated by a non-car maker.

Of course, a car-maker's business model is to sell cars. The more the better. And autonomous tech is a way to increase sales of cars. And of course if Tesla can achieve full autonomy on the current hardware (which I doubt) then, yes, the flip of one switch makes every Tesla a potential robotaxi in the length of time it takes the servers to send the update to that many cars.

A company that develops autonomy technology but does not build cars has an entirely different business model, which is to lease the tech to car-makers to put in their cars. Such a company could potentially lease the tech to many car makers. The lag time is much longer, because those car companies need to tool up and build the cars. But if Tesla cannot reach full autonomy on the present hardware, those other companies collectively could beat Tesla to the one-million mark.

What Tesla has going for it is its attitude of no geofencing. If Waymo or MobilEye need to closely map and analyze every locale before their cars can operate there, it's possible that the cars running their technology are on sale years before Tesla but Tesla has a car that can operate where I live years before those others. A car that's Level 4 in Phoenix, San Francisco, and new Your City does me no good if it's not also L4 on Maui.

 

[2101Guy]

I harp on this because it saddens me that the company making the best cars a consumer can buy today feels it needs to sell promises it may or may not be able to keep, rather than just selling what it has and telling the world what it hopes to accomplish in the future..

As someone who worked at a large financial that went/is still going through a transformation attempt to be a "tech company" or a "tech company that happens to be a bank" or (whatever their slogan is these days), I can say that Tesla isnt alone in overpromising/ underdelivering. It's preached at many tech orgs...the "shoot for the moon, even if you miss you'll be among the stars" mentality, is how they look at it.

Aggressive forward movement vs traditional method of measure many times, cut once. (Lets be real...no other traditional car company, for example, would take such risks and release critical driving features to the public as "beta".)

Not saying I fully agree, just sharing some background