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

[shrspeedblade]

Was just at the factory, and I think I’m ok to say there was TP in the bathroom and lots of cars loading up in the production lot. Guess I might as well show up tomorrow as bankwuptcy isn’t immediately imminent!

 

[erthquake]

Was just at the factory, and I think I’m ok to say there was TP in the bathroom and lots of cars loading up in the production lot. Guess I might as well show up tomorrow as bankwuptcy isn’t immediately imminent!

Could you tell if the cars loading were RHD?

 

[sparcs]

Tesla has some clear advantages. But so does Waymo. I don't see any here who acknowledge the latter. That's not smart investing.​

Waymo and Cruise are much more opaque compared to Tesla. Lot easier to criticize and dissect Tesla since their progress is getting released out to a lot of actual customers. Tesla literally had an Autonomy Investor Day in full public view.

If you want a real discussion, why not post a full discussion in the main investor forum? I would love to see some clear information with sources. Would be a lot better than assuming none of us want to acknowledge Waymo's progress. Some of us just lack the information or haven't stumbled on the article.

 

[MarcusMaximus]

15 million miles with a left turn every couple miles or so.

I very specifically gave an example where Waymo struggles despite ample training data. You ignored this point and trotted out the time-worn "Tesla has billions of miles" cliché.

You do, of course, realize that “billions of miles” > “15 million miles”? By at least 2 orders of magnitude?

From what I can see, you’re taking 15 million and assuming that’s sufficient data because it sounds like a big number.

 

[EVNow]

There's no evidence Tesla has some "tight integration" advantage.

Tesla has some clear advantages. But so does Waymo. I don't see any here who acknowledge the latter. That's not smart investing.

Well, Tesla makes cars - Waymo doesn't.

I think of Waymo as the current leader because they are running a "robotaxi" service. They can clearly do city + Freeway NOA, which Tesla can only do in Dev now. But they apparently have hit a plateau - we don't know why.

Tesla is definitely doing things differently and we've to see how good their city NOA is when it comes out.

BTW, any comparisons on NOA on freeways between Waymo & Tesla ?

 

[Electroman]

That Audi e-tron recall is really bad, according to local Belgian news: Audi moet 'Brusselse' e-tron terugroepen

“Er zijn nog geen dergelijke gevallen bekend, maar toch moeten alle wagens die gebouwd zijn sinds de productiestart in september 2018 en midden maart 2019 terug naar de dealer. Omdat de herstelling tot augustus kan duren krijgen de Amerikaanse eigenaars van een e-tron gedurende die periode een vervangwagen en een kaart met 800 dollar ter compensatie. ”

Translation: All cars built since production start 2019 en mid march 2019 need to be fixed. The repair can take untill august and the (1.644 ) American e-tron owners get a replacement car and a card with 800 USD compensation.

The part in question is an out-of-spec third party part.

Notice how Bloomberg soft peddles this: Audi Recalls First Electric Vehicle in U.S. on Battery Fire Risk
Would not want to upset one of their top advertisers. Look at the bulleted summary at the beginning of the article:

[*] New E-Tron SUV challenges Tesla’s dominance in electrics
[*] Company not aware of accidents or injuries related to flaw

And that report is from Dana Hull, to boot !! If this had been about Tesla, we can see Dana and Lopez writing the report something like this:

"Money losing Tesla has another big problem in its hand - Exploding batteries that could potentially to kill its occupants and is a safety hazard for other vehicles on the road.

With the recent massive fall in demand, and with the increasing cash burn resulting in 750M loss, the exploding battery is the last problem Tesla investors were looking forward to. With JP Morgan and other long term investors coming out with bearish calls, Wall Street analysts are wondering if bankruptcy is only a few months away. Bloomberg reached out to Tesla on this but received no comment. Meanwhile there is a new report that has come out on extreme dissatisfaction among Tesla employees on working conditions. One employee who preferred to remain anonymous said, "it is hell out there". You can read more on that here:

Scumbags. Third rate dishonest vermins.

 

[MartinAustin]

I picked up my Raven P100DL Model S on Saturday. The new suspension is incredible. On standard mode there is still no significant body roll, but the car just eats up bumps on the road like a real luxury car. Very quiet too. When I change it to Sport setting, which matches what my previous Model S performance vehicles had, the ride is much more bumpy (you just get used to that as a driver when that is all you have). The ride of the Raven is much, much smoother and nicer than our Model 3 vehicles (which drive like go-karts). I think the new suspension will bring new customers who thought the previous Model S did not have a comfortable enough ride for a luxury vehicle.

I test drove a Raven S recently (for 160 miles!!!) and concluded the exact same thing. Model S is already the sales #1, but now the ride is something the critics won't be able to complain about when comparing with (for example) S-Class.

 

[Electroman]

Aaand he is gone ... this is almost poetic. Congratulations everyone (including the concern trolls that I have put on ignore)

Twitter

Why is this troll's account still there?: Mark B. Spiegel (@markbspiegel) | Twitter

 

[Beltsbear]

That's a remarkably blasé claim. A quick literature search didn't turn up any studies for me that compare collision rates of Anisopterans to autonomous vehicles. Perhaps I'm looking in the wrong place...

It’s called outside.

 

[Mike Smith]

Tesla Gigafactory 3 complex enters initial tooling phase as new jobs open for the site

 

[urarthur]

Notice how Bloomberg soft peddles this: Audi Recalls First Electric Vehicle in U.S. on Battery Fire Risk
Would not want to upset one of their top advertisers. Look at the bulleted summary at the beginning of the article:
And that report is from Dana Hull, to boot !! If this had been about Tesla, we can see Dana and Lopez writing the report something like this:


Scumbags. Third rate dishonest vermins.

So basically all E-trons are now recalled and Audi provides a loaner that could last until August. Let me guess, the loaner is not an EV.

 

[acoste]

You do, of course, realize that “billions of miles” > “15 million miles”? By at least 2 orders of magnitude?

From what I can see, you’re taking 15 million and assuming that’s sufficient data because it sounds like a big number.

Pretty difficult to predict who is going to be the best 5 years from now.

Tesla has more miles recorded. That's good for debugging their own system, learning the behavior of other drivers and for collecting unexpected objects on the road.

Waymo doesn't have nearly as many miles on the road. However for training their system they can use simulators with predefined situations. It's not that difficult to come up with many possibilities. And their street view cars have collected thousands of unexpected objects on the road, which can be used as well and fed back into the simulator. They don't have much data for driving at night I guess.

Well, Tesla makes cars - Waymo doesn't.

I think of Waymo as the current leader because they are running a "robotaxi" service. They can clearly do city + Freeway NOA, which Tesla can only do in Dev now. But they apparently have hit a plateau - we don't know why.

Tesla is definitely doing things differently and we've to see how good their city NOA is when it comes out.

BTW, any comparisons on NOA on freeways between Waymo & Tesla ?

Waymo already has a factory where they are going to modify the Pacifica and the I-Pace. I guess they picked these because these platforms have the features needed or the potential to add these features.

Waymo robotaxis don't drive on the highway as of now.

 

[EVNow]

Waymo already has a factory where they are going to modify the Pacifica and the I-Pace. I guess they picked these because these platforms have the features needed or the potential to add these features.

Waymo robotaxis don't drive on the highway as of now.

Modifying a few thousand cars is not the same as building hundreds of thousands from scratch. But I guess it serves Waymo’s narrow needs.

Are you sure the robot axis don’t go on freeways? I read reports of people who use their robotaxis talking about how Waymo misses exits etc.

 

[Fact Checking]

So first you claim:

Self-driving teams today are not limited by data. To the contrary, they can't fully solve cases they see every day.

But then, when your broad claim is challenged, you claim:

Tesla has an undeniable advantage collecting edge case data.

Which is in contradiction with your first claim.

For example, Waymo recently hired a retired police horse so they could train and test their "horse-handling" because they don't encounter horses that often in normal driving. Tesla can leverage their fleet to capture much more horse data than Waymo. Advantage Tesla. Yay!

So for the case of "horses" Waymo was data limited - and they are probably also data limited for a large number of other cases that Tesla encounters.

Tesla is probably even able to identify edge cases that Waymo doesn't even know about exist with statistically relevant probabilities.

You are proving my point really.

But we're not talking about horses. Unprotected left turns are not an edge case. Waymo has plenty of left turn data. "Billions of miles" is not a magic wand.

So my objection to your claim was that you incorrectly generalized from the fact that Waymo is not data limited for unprotected left turns, that self-driving teams are not data limited at all.

Self-driving teams are obviously data limited in a number of key areas:

• Flow of data defines the quality of corner case, edge case, "six nines" data sets.
• The iteration and testing speed of FSD software versions depends on fleet size and data size: the larger the fleet, the faster is a given level of confidence achieved.

Waymo has plenty of left turn data. "Billions of miles" is not a magic wand.

You still don't seem to get it: Waymo might have 10 million miles of data for their OLD networks and OLD driving response logic.

But when Waymo modifies their driving software, they have to re-test it with NEW millions of miles of real driving where that new software makes the decisions, to have full confidence in it.

Tesla is able to expose their new networks and new driving logic to 3 orders of magnitude more data: millions of miles every week.

This is Tesla's primary advantage: they can iterate and gain confidence in a new FSD software version so much faster than Waymo, GM Cruise and others.

 

[acoste]

Modifying a few thousand cars is not the same as building hundreds of thousands from scratch. But I guess it serves Waymo’s narrow needs.

Are you sure the robot axis don’t go on freeways? I read reports of people who use their robotaxis talking about how Waymo misses exits etc.

They may test them on the highway, but as far as I know the cars deployed in Phoenix don't use the highway. Also in the Bay Area I see them testing on local roads only.

based on this article it sounds like an on and off thing A Waymo One rider's experiences highlight autonomous rideshare's shortcomings
"The first time I took it, it got onto the freeway on-ramp"
" Carl said it was several months before he was in another Waymo vehicle that attempted to take the freeway"
"It has never taken the freeway since."

 

[SO16]

Tesla Uberbulls:

#2 ranked place millenials want to work? Yay! So important.

Massive drop in Glassdoor ratings? Meh who cares.

One metric has been consistent for years. The other, a few months right after layoffs.

 

[acoste]

You still don't seem to get it: Waymo might have 10 million miles of data for their OLD networks and OLD driving response logic.

But when Waymo modifies their driving software, they have to re-test it with NEW millions of miles of real driving where that new software makes the decisions, to have full confidence in it.

That's where the simulators come in handy.

 

[urarthur]

Anybody thinks macros will go down tomorrow because there was no "deal" with Mexico?
Or as long as Trump thinks there is and does not impose tariffs the existence of the deal should not matter?

huh? there was a signed deal with Mexico over the weekend... Next big macro is if Xi will join G20 and meet with Trump.

 

[CyberDutchie]

I picked up my Raven X on Saturday, as well. I am not super experienced with fancy suspensions so I didn't pipe up earlier--was waiting till see what others thought.

But I agree completely with @rdalcanto on this. The standard adaptive damping is supremely smooth. Sport feels a lot like the car is on coils, which is rather amazing for an air suspension.

The range upgrade got most of the Raven press--and for good reason. But the suspension is fantastic. It is definitely a step up from the prior offering. You get to have your cake and eat it, too.

Loving the car so far.

View attachment 417849

Have you timed 0-60 with Ludicrous yet?

 

[shrspeedblade]

Could you tell if the cars loading were RHD?

Sorry no, was just cruising by on the way to do the factory tour and noticed a lot of activity out in the lot.