I'm curious what percent of Tesla sales are primarily because of full self driving capability.

 

Historically Tesla sales don't rely very heavily on FSD because most buyers don't want it enough to buy it (somewhere around 20%). But with the rapid advances closing in on full autonomy, that's bound to change. It won't be too many years before a car that can't drive itself is considered outdated.

Tesla's FSD, even with it's relatively low take rate, does boost profit margins considerably.

 

That is so effing delusional I want to puke

 

The current full-attention capabilities of my 2nd gen R1S are great, and I really appreciate them when I'm on a mapped freeway. But the leap to autonomous driving on freeways by 2026, let alone in any setting, is completely out of the question given the capabilities of the current system and the complete lack of training they do in the real world.

I'm with @Zipper on this one, RJ has to go. This is truly disqualifying stuff. Well ok Elon did it, but he had market advantages that compensated for his autonomous insanity. RJ doesn't have that benefit.

 

I'm with @Zipper on this one, RJ has to go. This is truly disqualifying stuff. Well ok Elon did it, but he had market advantages that compensated for his autonomous insanity. RJ doesn't have that benefit.

Elon's error with FSD was not realizing how long it would take to get where they are now, launching autonomous robotaxis using regular Model Ys. He was just too optimistic (like he is with many of his outsized initiatives). If you want to do what most people consider impossible, you have to be optimistic. Elon knows how to get stuff done, and part of that is expecting it to get done yesterday.

I do think Rivian will be able to buy and implement some level of autonomy, but not by 2026, probably not until 2028, and they will have to price it low enough to get enough takers that it won't add to their bottom line much, if at all, once they pay the people who provide it. It's just something most automakers will need to do over the next few years in order to simply stay in the game.

What I find interesting about RJ is that when they were a young company he appeared to take special pride in making it look like they were doing some things different from Tesla (even though most decisions were lifted right from Tesla). Most of those deviations from Tesla's playbook came back to bite them, like thinking they could overcome the added battery and range expense of not making car designs that were aerodynamic (particularly when batteries were more expensive than they are now). But as the financial noose at Rivian tightens, he seems more willing to take plays out of Elon's playbook, even though he desperately wants it not look like that.

RJ is a follower, not a leader, and when he tries to lead, he pays the price.

 

Maybe they're working on partnership with Google for that. Google have a profound amount of mapping data and autonomous navigation software. Rivian are already using Google maps.

 

Google doesn't have training data. Waymo of course does, but only in limited settings. Tesla has tons of data from their users, which are millions, in many settings, though not as specific and reliable as Waymo. Where would Rivian get the training data? It's a pipe dream.

 

Yeah, I didn't take those comments to mean they would do autonomy in-house, although it was presented in a manner that will lead many to believe that's what he was talking about. Most people that don't understand what is needed to make autonomy work. Massive data and massive compute. Rivian doesn't have that, nor can they afford it. Even if they had the money it would be difficult to get the expertise and necessary leading edge hardware to make it happen. And certainly not by 2026. That's the key to understanding he was talking about licensing it from others.