Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

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

This site may earn commission on affiliate links.
This looks too good to be true!

tsla.PNG
 
This kind of talk buy bulls strikes me as rationalizing. Trying to make excuses for what is otherwise a clear mistake.

It is only a mistake if Tesla had free resources to do something major to Model S/X, and elected not to do it.

More likely their short term priorities are Model 3, Model Y, FSD and Cells.

Within the available budget they did what they could with Model S/X, and that is the Raven refresh..

I think further improved versions of Model S/X are coming it is just taking longer than people would like.

Yes, demand has dropped a but probably more than they are happy with, but there are lots of reasons for that.
If you step back there only reason there is a problem with Model S is they did such a good job on the Model 3.

They were also a larger company with more resources to expend when developing the Model 3, and that shows in the end result..
Eventually a lot of Model 3 improvements, or more likely improved versions of those improvements, with be in Model S/X.

I have little doubt the plan is to eventually make Model S/X a lot better, but that is the last thing Elon can say publicly now..

It is also likely the budget for improvements is not unlimited and the likely future revenue stream is a consideration. IMO there is no need to change the exterior shape of the cars at all.
 
I didn't see this posted here yet:


Self driving part starts at 26:45.

A lot of very interesting bits, including many comments about Tesla:

  • "Tesla is 100% going to be first to level 5 autonomy. Comma.ai will be 2 years behind them."

  • "Tesla will be the iOS of self-driving, our plan is to be like Android."

  • "Usually I think Elon's comments are spot on, but in the case of not needing driver monitoring he is wrong."

  • "My conversations with Elon are the reason for Comma.ai's existence. I wouldn't be doing this if not for those conversations."

  • "Waymo is 3-5 years ahead of everybody else using Lidar. Too bad for them it's not the correct strategy."

  • "GM buying Cruise for 1B$ was brilliant. It's an insurance policy against Waymo."

  • "I don't see Cruise leapfrogging Waymo. I also don't see us leapfrogging Tesla unless they mess up."
He also talks about how he plans to launch car insurance, and a couple of advantages he would have over other insurance companies. I'd imagine all of these will apply to Tesla too.

There's a bunch more, and it's an interesting watch/listen. I feel like George can be overly confident, and a little disconnected from reality at times, but it's also clear this guy lives & breathes self-driving, and he knows a lot about the tech.
 
This looks too good to be true!

View attachment 438730

I think this has been in place for over a year. Hong Kong has very high car registration taxes, which rapidly scale with car value (40% on the first HK$150,000 of vehicle taxable value, 75% on the next $150,000, 100% on the next $200,000 and 115% on the remainder), so despite this potential HK$250k incentive, Teslas are still very expensive. Also, i presume given the high registration taxes, car scrappage is relatively rare.
 
Last edited:

George basically says that everyone’s approach to self driving aside from Tesla’s is wrong and just will never work. My impression was that he convinced Lex of this fact as well, though obviously Lex did not come out and say it.

George also said that everyone else’s approach stems from that early code that successfully passed the DARPA challenge early last decade (which I didn’t know). It kind of explains why Tesla is taking a different approach from everyone else, and all the others may just kind of be locked in.

It just seems like there’s more and more evidence piling up that Tesla really is alone in the self-driving lead, contrary to appearances and industry consensus.

I really struggle to see how anyone with neural net experience who has spent a lot of time thinking about the nature and the difficulty of the self driving problem cannot see deep down that Tesla's Data Heavy approach is the only one that will work. The longer it takes other companies to pivot in Tesla's direction, the further ahead Tesla is going to get.

We have now seen George Hotz saying Tesla will win Level 5 and Waymo/Cruise/everyone else's approach will not work.
We've seen Anthony Levandowski admit that Elon was right about Lidar all along.
We've also seen Alex Krizhevsky (behind Alexnet which kicked off the deep learning revolution in 2012) admitting Tesla is in the lead (after working at Google and Waymo). “I think Tesla has the unique advantage of being able to collect data from a very wide variety of environments because there are Tesla owners with self-driving hardware all over the world,” “This is very important for machine learning algorithms to generalize. So I would guess that at least from the data side, if not the algorithmic side, Tesla might be ahead.”

I think the industry is finding it so hard to pivot to Tesla's approach for several reasons:
  • Many of the self driving project teams are led by roboticist/Lidar experts who do not have a deep understanding of neural nets.
  • Waymo etc have already invested so much time and resources into their approach it is difficult to admit they were heading in the wrong direction.
  • All so called "self driving experts" consulted by corporates, investors and the media originated from the DARPA challenge and therefore have experience in the Hardware Heavy, Data Light approach, and have also invested their whole careers into it. This has convinced CEOs and investors etc to bet on the wrong strategy.
  • Everyone else is so far behind Tesla in their ability to role out a mass produced self driving data filtering, processing and collection consumer fleet, now their only hope is to somehow solve self-driving without the data.
  • Some people really think 10s of millions of miles of real world driving experience is enough to train the car and to verify safety. Elon is betting it will require 10s of billions of miles of real world driving.
  • Some people really believe simulation can substitute for real data and somehow think a software developer in silicon valley has the imagination to program all possible 1 in 10 million driving scenarios, and has the ability to accurately program artificial conscious beings in the simulation.
  • Some people really think it's impossible to get distance and velocity measurements from cameras alone (despite all the evidence to the contrary).
There seems to be a false narrative spun by the industry that Elon's rational for choosing the Hardware Light approach was just to save on manufacturing costs and/or an arbitrary rejection of lidar. This is completely false. There are two mutually exclusive choices for self driving strategy: 1) Hardware Light, Data Heavy or 2) Hardware Heavy, Data Light. It is impossible to choose a Hardware Heavy, Data Heavy approach because a car fleet of Tesla's size using this approach would cost the company towards $100bn capex and development costs. Elon choose his approach because he correctly understood that data is key to neural nets, solving edge cases and verifying safety. Elon knew there is a proof of concept (us) that driving can work through vision alone, but he took a gamble at how long it would take to solve distance/velocity with vision rather than lidar. Fortunately everything is suggesting Tesla (and other AI researchers) are making rapid progress on this (if not solved sufficiently already).
 
Last edited:
I think this has been in place for over a year. Hong Kong has very high car registration taxes, which rapidly scale with car value (40% on the first HK$150,000 of vehicle taxable value, 75% on the next $150,000, 100% on the next $200,000 and 115% on the remainder), so despite this potential HK$250k incentive, Tesla's are still very expensive. Also, i presume given the high registration taxes, car scrappage is relatively rare.

Yes, appears to be old news, reposted today by Vincent, usually a reliable source. It seems I no longer have option to delete the post though. Thx.
 
  • Waymo etc have already invested so much time and resources into their approach it is difficult to admit they were heading in the wrong direction.

Same reason some Japanese companies are still going after fuel cell vehicles. Especially the Japanese really don't like to admit they were wrong and bring shame upon themselves.

  • All so called "self driving experts" consulted by corporates, investors and the media originated from the DARPA challenge and therefore have experience in the Hardware Heavy, Data Light approach, and have also invested their whole careers into it. This has convinced CEOs and investors etc to bet on the wrong strategy.

I too get this impression when I see car manufacturers throwing money at self-driving startups like Aurora and Cruise. They don't really seem to understand the problem, and just throw money at things hoping it's the winning horse.
 
It looks like July has been weaker for EV sales in China post subsidy cuts (which Tesla never had access to).
BYD July EV sales were +16% yoy and -50% mom to 9.5k. PHEV sales were -35% yoy and -7% mom to 6.5k. I'm not sure how much of the subsidy cut was passed on to consumers.
I'd imagine weaker EV players will have performed even worse in China, i think full numbers will be out next week.
It's a great opportunity for Tesla to take market share in China - Tesla can significantly reduce their pricing with local production of GF3 Model 3s while pricing in the rest of the industry is likely to significantly increase.

First data (via @DKurac) has China July NEV sales down 6% yoy and 43% mom to 74k. EV battery capacity up 40% yoy and down 29% mom to 4.7 GWh.

EBW4nTeUIAEHSqJ
 
Same reason some Japanese companies are still going after fuel cell vehicles. Especially the Japanese really don't like to admit they were wrong and bring shame upon themselves.



I too get this impression when I see car manufacturers throwing money at self-driving startups like Aurora and Cruise. They don't really seem to understand the problem, and just throw money at things hoping it's the winning horse.

The problem car manufactures are primarily trying to solve is making it look like they have a plan in the self driving space, even if just for appearances sake for shareholders and Employees. There is also a bit of a safety in numbers thinking going on it seems, as in if all the legacy automakers are partnering in the same or similar efforts, then they are all in the same boat if the tech does or doesn’t work. (As long as they all fail at it together there is less chance one specific company goes under in a hurry.)
 
There's a few things I can think of:
  • A bit over 1/3 of the LEAFs produced have been delivered to the US, whereas a bit under 1/3 have been delivered to Japan and Europe each. Have Model 3 deliveries even begun in Japan yet (I know orders have)? And various factors have caused the majority of Model 3 deliveries to AFAIK go to the US (and Canada).
  • There's 9 year old LEAFs, so yes, attrition is much more of a factor. (Especially with the well-documented LEAF battery degradation problems.)
  • I know the LEAF was one of the least compliancey of the legacy automaker modern BEVs (in that it was a serious attempt even with the major issues with it, and it was always sold in 50 states AFAIK), but deliveries may have been prioritized to ZEV states (Washington not being one), whereas Tesla delivery prioritization within the US seems to (after the initial cars, anyway) purely be based on logistics within a quarter (so everyone gets them based on demand, not based on credits).
Just to throw some general numbers out there....
500,000 Leafs total? 1/3 to the US? That’s 166K of those little guys in the US.
250,000 Model 3s? 1/2 to the US? That’s 125K in the US.

If those numbers are close then there should be about as many Leafs as 3s.
I definitely see a lot more 3s. Leafs actually seem to be disappearing. I think people are trading them in for 3s. I used to see more Leafs.
 
Thank you for expressing your opinion, that while irritating at times (he knows he is), that you would like @neroden back.:cool:

I have also expressed that opinion but can't find my previous posts to that effect.
I have deleted my post because I think I was wrong. I agree with my perspective on disagreements as a learning opportunity but not on the specific one I stated. Frankly I think I failed to consider the scope of cultivation for fudsters of recurrent harping on the negatives. I reviewed several earlier exchanges on AI deployment including NN. When one posts repeatedly the same content as if one is an expert when factually agreeing that such expertise was acquired only by osmosis one begins to think of more famous people who post layman opinion as objective fact. That is never good.

I hope we continue to have lively informed debate while maintaining respect and civility. Personally, I try to distinguish topics on which I think I have sufficient expertise to have a solid opinion and ones in which I have less confidence in my knowledge. I think most of us do that. Hopefully I do that myself. One part of that is to be straightforward when I have been popular in my wrong statement. This is one of those times. I am no fan of censorship, but I am a bigger fan of truth. I was wrong. Mods: please accept my apologies.
 
I'm not saying that the S completely defines Tesla. But it is not ONLY a means to and end. In that case, why Roadster 2.0? Why rocket packs on Roadster 2.0? Tesla is more than about just making mass market affordable EVs. It is that, but it is also much more.
There are just people who want a larger car. Particularly if they go on a lot of road trips. A larger car is just better for that. I really like the way we can travel without having luggage visible. I also like how I can lay my pants and shirts down in the frunk, keep all the small stuff in a duffel and Denise's two suitcases below the parcel shelf. Of course, the new S kind of ruined that with the minuscule frunk, but it's really nice in my car.
 
IAll so called "self driving experts" consulted by corporates, investors and the media originated from the DARPA challenge and therefore have experience in the Hardware Heavy, Data Light approach, and have also invested their whole careers into it. This has convinced CEOs and investors etc to bet on the wrong strategy..
The way consultants keep their jobs is to first do an interview and determine what the person hiring them wants. Then they do that, even if it doesn't work and they know it won't work, because if they say otherwise they are unlikely to get the job even if other solutions are better, lest costly, or both.
 
AFAIK The only car company CEO in the world who has any computer programming experience is Elon Musk.

I'm a coder. there are TONS of situations where the explanation makes no sense UNLESS you are at least partially a coder.
CPUs and GPUs are seriously quick right now. What you need is decent code, not more hardware tech. Most code is running at <1% of the true capability of the hardware, even if its generalist hardware, even in the supposedly optimized field I work in (video games).

Every car company CEO is a hardware person. Suddenly the problem is not hardware but software. No surprise none of them get it.
When they do get it, they will have already lost, a long time ago.
 
AFAIK The only car company CEO in the world who has any computer programming experience is Elon Musk.

I'm a coder. there are TONS of situations where the explanation makes no sense UNLESS you are at least partially a coder.
CPUs and GPUs are seriously quick right now. What you need is decent code, not more hardware tech. Most code is running at <1% of the true capability of the hardware, even if its generalist hardware, even in the supposedly optimized field I work in (video games).

Every car company CEO is a hardware person. Suddenly the problem is not hardware but software. No surprise none of them get it.
When they do get it, they will have already lost, a long time ago.
You got it. Specialized ASICs with code designed for them will run circles around a general CPU every time.
 
  • Like
Reactions: mekberg