Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[kabin]

IMO FSD won't affect TSLA until it's level 5. Because of the long history and because few people understand V12, FSD will have to be actually in place and driving while everyone in the car naps or works on their laptop before it will make a difference to TSLA. Or maybe level 4 will be a trigger. But not optimism before they accomplish level 4. Elon just said execs at OEM's don't believe FSD is real.

In my uninformed opinion.

Yep. Elon said himself without FSD TSLA is just another car company. There's plenty of reasons auto manufacturer won't sign up for the FSD crowd sourced sucker train.

 

[uscbucsfan]

There's really no such thing as level 4 with a L5 FSD design(to go anywhere). Level 4 can happen any time with FSD as you just need to define when/where and under what conditions it can be driver free. Find a well mapped light traffic area 3am in the morning and have a Tesla runs in circles 10000 times without disengagement and that is your L4. It's almost like the nonsense Mercedes came out with their Level 3.

People likes to front run potential and we already awarded Tesla with a generous valuation due to many believing FSD will be solved even with little evidence it will be. I am hoping V12 actually starts providing an abundance of that evidence for the not so faithful.

A L4 system that has no mapping, but a limit on conditions would still be considered solved FSD. Like if the car couldn't drive in heavy/built up snow or severe weather conditions, but could go anywhere. That's a more realistic goal. Level 5 goes beyond just great driving, it's the capabilities of the car.

 

[kabin]

The fact that humans will no longer be writing code for V12 while performing as well if not better than V11 is what gives me hope that it will be solved. The more human involvement when it comes to NN based software the worst outcome.

Hook, line, and sinker, that's Elon's narrative. Time will tell.

 

[uscbucsfan]

Yep. Elon said himself without FSD TSLA is just another car company. There's plenty of reasons auto manufacturer won't sign up for the FSD crowd sourced sucker train.

There was no reason to believe that any car manufacturer would OEM FSD at this point. It's not even a reliable level 2 system and they would need Tesla computers, cameras, and to feed data back to Tesla. Tesla would have to make adjustments for their brakes, size, control, etc.

Many said that was BS over a year ago. Solved FSD, sure, but not the Beta product we have now that 5-10% (optimistically) of the time drivers are forced to take over. My car uploads as many as 6 gb in a night after long drives. That's not feasible for most manufacturers.

 

[uscbucsfan]

As a tester and someone who loves FSD Beta, I think Tesla is more likely to launch the M2 and sell 3+ million cars this year than to "solve FSD".

 

[Singuy]

Hook, line, and sinker, that's Elon's narrative. Time will tell.

It's not a narrative, it's literally how NN based software work and hence why it can do certain things beyond the capabilities of human coding. I think people didn't realize how much human based coding was in FSD before end to end. The way Tesla's team talking about solving edge cases which requires NN training, shadow mode, disengagement feed backs, hundreds of thousands of video clips, etc etc...everyone thought they were doing end to end already but turns out most of the NN based software stack were just perception and never planning.

 

[Singuy]

There was no reason to believe that any car manufacturer would OEM FSD at this point. It's not even a reliable level 2 system and they would need Tesla computers, cameras, and to feed data back to Tesla. Tesla would have to make adjustments for their brakes, size, control, etc.

Many said that was BS over a year ago. Solved FSD, sure, but not the Beta product we have now that 5-10% (optimistically) of the time drivers are forced to take over. My car uploads as many as 6 gb in a night after long drives. That's not feasible for most manufacturers.

Reliability is obviously not the bar legacy auto set for L2 software. If that was the case, then we should see all of their AEB and lane keep assist being as good if not better than Tesla's. So far objectively they are all worst...so no reliability is not the limiting factor for them to adopt FSD as if they cared..lol. Their AP equivalent can't do hard turns and run people off the road daily. That's why no one uses them.

 

[uscbucsfan]

Reliability is obviously not the bar legacy auto set for L2 software. If that was the case, then we should see all of their AEB and lane keep assist being as good if not better than Tesla's. So far objectively they are all worst...so no reliability is not the limiting factor for them to adopt FSD as if they cared..lol. Their AP equivalent can't do hard turns and run people off the road daily. That's why no one uses them.

Blue Cruise and Cruise are better than AP. Less calable than FSD, but better/more advanced than AP.

Irrelevant to the FSD licensing convo, too.

 

[Singuy]

Blue Cruise and Cruise are better than AP. Less calable than FSD, but better/more advanced than AP.

Irrelevant to the FSD licensing convo, too.

Blue cruise enabled cars so far cannot beat a Tesla in AEB test. Ford pays mobile eye for a hardware/software suite that runs into dummies during Euro NCAP AEB test. You see how legacy auto has zero problem licensing driver assist suites that are unreliable or less reliable than Tesla's. You would think there's a line out the door after NCAP gave Tesla a 98% for passing all of their AEB test while competitors were hitting dummies left and right...you know for safety reasons. Yet we have crickets because....no one is interested in reliability.

 

[GhostSkater]

Now with the full curve, but to be fair, this is how Cybertruck charging curve has to look to hit the charging times also

https://twitter.com/x/status/1753530938837881186

 

[Usain]

IMO FSD V12 determines a lot of Tesla's valuation regardless of earnings this year. Everything prior to V12 people are either in the camp of it'll never work to it'll never leave L2. V12 is the final foundational iteration of FSD. Hopefully it turns people from never to most likely...and with each update goes from 10 years away to 5 years away to 1 year away. It could also hit some kind of AI training maxima that proofs certain things cannot be trained into working. Either way this is it...no more rewrites, there's nowhere to go after end to end.

It could also hit a maxima due to hardware limitations.

But as I said earlier today, V12 seems like the version that will not max out any time soon.

I agree that V12 is probably the one that proves if FSD will ever work or not. And I think we will get a good idea of that this year.

If you are buying TSLA you are betting on FSD and/or bots paying off in the next few years. As for me, I continue to buy every two weeks no matter the price.

 

[gene]

I was wondering the same recently. He is ok, still active on other forums, like here: JRP3

That's too bad.

 

[Singuy]

It could also hit a maxima due to hardware limitations.

But as I said earlier today, V12 seems like the version that will not max out any time soon.

I agree that V12 is probably the one that proves if FSD will ever work or not. And I think we will get a good idea of that this year.

If you are buying TSLA you are betting on FSD and/or bots paying off in the next few years. As for me, I continue to buy every two weeks no matter the price.

The good thing about bot is that it's Elon's back up bet. Since human life is mostly not at risk, that project is more about manufacturing than perfecting the software to solve all edge cases and be reliable 100% of the time.

 

[Usain]

The good thing about bot is that it's Elon's back up bet. Since human life is mostly not at risk, that project is more about manufacturing than perfecting the software to solve all edge cases and be reliable 100% of the time.

That's true, but I don't think FSD has to be as good as a human driver to do robotaxi. Like Waymo, it only has to be as good as a human driver in one geofenced area.

Tesla can train V12 to be truly great in a relatively small geofenced area. And unlike Waymo, Tesla can do it profitably.

I really think Tesla will start out this way. It has to be goefeced to start. And with V12, it shouldn't be that hard to accomplish.

 

[⚡️ELECTROMAN⚡️]

Has anyone ever broken down the relative cost to send an email? Because it is not free. Would be curious.

Seriously you guys don’t find the humor in this post. /s is mandatory or you think it’s serious? I thought it was pretty hilarious

 

[MartinAustin]

Looks like good ol Toyota is changing course again.

Toyota says bye to electric cars. They will create ammonia engines

Toyota has always chosen a different path in its quest for cleaner and more sustainable transportation. While it is true that the Japanese automaker is

lagradaonline.com

Never heard of that website before today. I checked some of the other articles. I would not call it a strong source of information

 

[MartinAustin]

Hey Tesla, I know you dominate everywhere else, but hey, take some notes, or at least give us a clue when/if/where Cybertruck will charge this fast

https://twitter.com/x/status/1753460897949811186

...when Teslas charge as fast as this one-off, not-for-sale Porsche prototype? I remember when the Mission E was unveiled in 2015 with 350kW charging. The 350kW number got a lot of coverage all over the media. When the car came out it was 270kW or lower, sometimes 220kW. When the Taycan is on sale, and a production car is delivered to a buyer and tested, let's see what the benchmark is then

 

[⚡️ELECTROMAN⚡️]

It could also hit a maxima due to hardware limitations.

But as I said earlier today, V12 seems like the version that will not max out any time soon.

I agree that V12 is probably the one that proves if FSD will ever work or not. And I think we will get a good idea of that this year.

If you are buying TSLA you are betting on FSD and/or bots paying off in the next few years. As for me, I continue to buy every two weeks no matter the price.

I thought we were betting on interest rates coming down.

 

[Paul_SF]

Crazy to think Meta added 1/3 of TSLA to their market cap in 1 day.... must be all those people betting with their stimmies, right? Right....

I can't wait to see what the hedgies and MMs are reporting as profits this year... Imagine if the game was rigged, they would have made even more!

 

[Ipcress]

Yes, a good listen. I like Chamath's take on this. Delaware has gone to *sugar*. Big time.

I stopped listening to that pathetic bunch when SVB went belly up and the former "get government off our backs" gang went wailing like terrified lambs for a Government bailout. Disgusting.