CarlK
Active Member
You did not miss anything other than the second step would happen is six months (Elon's time of course). Although I think that will be done is smaller steps.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Autonomy Investor Day - April 22 at 2pm ET
- Thread starter diplomat33
- Start date
You did not miss anything other than the second step would happen is six months (Elon's time of course). Although I think that will be done is smaller steps.
This is not a startup pitching for financing. This is a publicly traded company discussing product plans for the on-going year. This is what makes the story big.
I agree the market now ponders whether or not it is true or it is a lie.
Eno Deb
Active Member
Well, he did clearly say "yes" when someone asked about level 5 by the end of the year. But perhaps it was a misunderstanding.
Daniel in SD
(supervised)
Level 3 would be a massive accomplishment. Arguably no one has deployed a level 3 system.
Eno Deb
Active Member
I think that's exactly what that was. I've seen this many times.
So far pretty light on that. This was an IP showcase to appease the market in my view. The closest to that we got was that Tesla ditched HD maps and that their data collection triggers have evolved. We also got some nice insight on how NoA has evolved. The demos may of course provide more insight.
Then it would be securities fraud on a massive scale so let’s hope not.
I of course agree it was an event overall to appease investors and improve market sentiment. But Tesla can’t go pitching yarn like to venture capitalists as a publicly traded company.
caligula666
Member
The number of TOPS is not really that relevant. What matters is how many neurons you can calculate in a single second. They were quite clear on this is the reason they went and made a specialized chip (data storage between steps, SRAM buffer etc...).
They could achieve the same performance as HW3 with normal GPU's, but they would need a lot more and 500W-1kW of power. Causing a very significant and difficult cooling requirement, plus way too much cost per vehicle. And a range-hit altough 1kW is just 5-7 km/hour so not that important.
Daniel in SD
(supervised)
It's not a lie if you believe it.
I feel like it was a good presentation of the infrastructure they have setup to pursue FSD. They emphasized their competitive advantage over other players int he market. It's hard to argue that their timeline is wrong when they're trying to do something that no one knows how to do or has successfully done. Maybe it is possible for them to do it in a year? It's impossible to disprove.
if somebody takes over driving it would turn the taxi into a rental car. Rental cars exist. If you want to assume liability, go ahead.
A robocar would be priced to reflect the value of the car.
If production capacity is less than or equal to taxi demand, the value would be extremely high and all would be taxis.
If production capacity is greater than taxi demand, the value of the excess would be relatively low and the company will make and sell as many as possible. A bird in the hand has a much greater capital return than two in the bush.
EDIT: Plus, owner of the vehicle pays for parking. Cost of parking is high.
Had Tesla not outlined a far more ambitious story then sure. But that does not fit the robotaxi plan.
Tesla has set the expectations, they did so in 2016 and now they have doubled down on them by saying what people buying cars today should expect: Level 5 no geofence with cars making maybe $30,000 a year in the fleet starting next year at least in some locations. Those are high goals but Tesla set those themselves.
People will now buy tons of Model 3s with this expectation rightfully bundled into the purchase. Tesla must deliver. Sure, some delay can and will be allowed, but not a fundamental deviation.
The other auto makers better not wait too long to find out the answer to that question because if Tesla even achieves L3, it will be too late.
I liked the presentations. Lots of nerdy info so geeks like me understand where they really are.
Feature complete in just 7 months given they are just now starting to involve neural networks in driving decisions seem extremely ambitious. I will be impressed if they pull that off.
Feature complete in just 7 months given they are just now starting to involve neural networks in driving decisions seem extremely ambitious. I will be impressed if they pull that off.
This is an all-in or nothing move, really. There is no going back to Level 2 ADAS from this.
It is robotaxis (next year) or bust. I like that.
My Level 5 capable hardware is waiting.
@ItsNotAboutTheMoney
I think we are saying the same thing. A hire car is what we call a rental car in the UK.
If I was being more specific, I guess it would be stop selling cars for a modest price. If you were confident in the feasibility of the robotaxi business you would want to start charging so much that people would be buying the car as a robotaxi or just use the cars in our own robotaxi business. You wouldn't be selling them as personal cars. Of course, you might also hope that speculation on the possibility of this ability to make money in the future would help sales of cars today. Then people are speculating on whether or not it actually happens.
Robotaxis could drive out of town and park somewhere cheap if not being used. Indeed one investment aspect of FSD is that the need for expensive city centre parking could collapse.
Of course the numbers only work if only Telsa has FSD. If every car maker had it then the value of FSD would collapse. I was somewhat taken by their argument that their fleet gives them some time to market advantage.
I think in general the high technical capabilities of the autopilot 3 were known. I would like to see the drives!
I think we are saying the same thing. A hire car is what we call a rental car in the UK.
If I was being more specific, I guess it would be stop selling cars for a modest price. If you were confident in the feasibility of the robotaxi business you would want to start charging so much that people would be buying the car as a robotaxi or just use the cars in our own robotaxi business. You wouldn't be selling them as personal cars. Of course, you might also hope that speculation on the possibility of this ability to make money in the future would help sales of cars today. Then people are speculating on whether or not it actually happens.
Robotaxis could drive out of town and park somewhere cheap if not being used. Indeed one investment aspect of FSD is that the need for expensive city centre parking could collapse.
Of course the numbers only work if only Telsa has FSD. If every car maker had it then the value of FSD would collapse. I was somewhat taken by their argument that their fleet gives them some time to market advantage.
I think in general the high technical capabilities of the autopilot 3 were known. I would like to see the drives!
True Tesla is setting the bar very high. I guess I am just looking at it from the point of view that even L3 would be huge.
But I am starting to wonder if maybe L4/5 is possible for Tesla because of the fleet learning advantage. As Tesla produces more cars, the fleet learning accelerates. It truly is exponential. And if Tesla gets 1 million cars on the road by end of 2020 (I think that was the stat), that's a massive amount of cars feeding into the machine learning to solve edge cases.
Similar threads
