-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
- Status
Fact Checking
Well-Known Member
I'm afraid I agree with @diatz and disagree with @neroden regarding FSD: Tesla is way ahead in the FSD game today. It's just that not even competitors like Waymo appear to have realized this:
- The Tesla AI chip is a game-changer: designed for NN and only for NN processing gives them significant advantages compared to NVidia - and Tesla's first-principles design gives them an advantage even over Google's third generation Tensor Processing Unit (TPU).
- Tesla's chip was designed by two chip design legends: the guy who designed the first x86-64 CPU at AMD (Jim Keller) which CPU has beaten Intel at their own game, and by the guy who designed Apple's first own CPUs, including the first 64-bit Apple CPU (Pete Bannon). Bet against these guys in the AI chip design space like you'd bet against Elon in the reusable rockets space. (Note that after finishing the AI chip design Jim Keller went to Intel's server division, probably pursuing another dream job - Pete Bannon has productized the HW3 design, implementation and roll-out.)
- The next generation Tesla neural network layers are already in the V9 firmware, probably testing-only in shadow mode. Here's @jimmy_d's tear-down of that network: "I didn’t expect to see a camera agnostic network for a long time. It’s kind of shocking." Note that V9 includes that new network but it's probable that only HW3 can run it efficiently. I think the new network was included to support field-test units of Tesla HW3 with v9 firmware as-is, and perhaps to run it in shadow mode on HW2.5 at very low frame-rates.
- Another quote from jimmy_d:
- "This V9 network is a monster, and that’s not the half of it. When you increase the number of parameters (weights) in an NN by a factor of 5 you don’t just get 5 times the capacity and need 5 times as much training data. In terms of expressive capacity increase it’s more akin to a number with 5 times as many digits. So if V8’s expressive capacity was 10, V9’s capacity is more like 100,000. It’s a mind boggling expansion of raw capacity. And likewise the amount of training data doesn’t go up by a mere 5x. It probably takes at least thousands and perhaps millions of times more data to fully utilize a network that has 5x as many parameters.
This network is far larger than any vision NN I’ve seen publicly disclosed and I’m just reeling at the thought of how much data it must take to train it. I sat on this estimate for a long time because I thought that I must have made a mistake. But going over it again and again I find that it’s not my calculations that were off, it’s my expectations that were off."
- "This V9 network is a monster, and that’s not the half of it. When you increase the number of parameters (weights) in an NN by a factor of 5 you don’t just get 5 times the capacity and need 5 times as much training data. In terms of expressive capacity increase it’s more akin to a number with 5 times as many digits. So if V8’s expressive capacity was 10, V9’s capacity is more like 100,000. It’s a mind boggling expansion of raw capacity. And likewise the amount of training data doesn’t go up by a mere 5x. It probably takes at least thousands and perhaps millions of times more data to fully utilize a network that has 5x as many parameters.
- The unnecessary and uneconomical LIDAR detour the others are taking might take 1-2 years for them to undo. Elon correctly saw it that the LIDAR approach is a dead end and Tesla now has a significant head start with 8 cameras in every consumer car, and probably up to 30 in the Tesla Semi.
Last edited:
Buckminster
Well-Known Member
Scores on the doors:
ER - 8/10 (perfection regarding the manufacture of free cash flow but some factory pics would have been nice for the family album)
Call - 4/10 (dare I say - a little dull... I will show myself out)
SP up to 316.80 - 4/10 (I feel like I'm on a different page to most of you here)
I expected profit, they gave us profit - am I wrong to have assumed more? Elon made a comment about a year ago regarding keeping his powder dry. It still needs to breathe otherwise it will get mouldy!
I know what you are going to say, its great that Elon didn't make promises he can't keep. Post profit, that is so last weeks problem. Tesla are now officially unstoppable and Elon is invincible.
ER - 8/10 (perfection regarding the manufacture of free cash flow but some factory pics would have been nice for the family album)
Call - 4/10 (dare I say - a little dull... I will show myself out)
SP up to 316.80 - 4/10 (I feel like I'm on a different page to most of you here)
I expected profit, they gave us profit - am I wrong to have assumed more? Elon made a comment about a year ago regarding keeping his powder dry. It still needs to breathe otherwise it will get mouldy!
I know what you are going to say, its great that Elon didn't make promises he can't keep. Post profit, that is so last weeks problem. Tesla are now officially unstoppable and Elon is invincible.
Fact Checking
Well-Known Member
Elon's notion of dialing up the Q1 earnings call to 11/10 wasn't well received, so he abandoned that concept. Sometimes visionaries are way ahead of their times and have to adjust. To quote a classic: 'But your kids are gonna love it'.
So 'dull' earnings calls are actually the new 10/10!
Last edited:
Europe: +9,04%
I feel like anything less than 12% is a disappointment.
Fact Checking
Well-Known Member
It might take a few days, weeks and months for the market to catch up to the new normal and add that zero to adjust it to +120%.
Europe down 2% now in pre-market. Thx for 50 more shares...
I think if he buys options he'll also be screwing the wash sale rule. IANATE but I believe the wash rule applies to stocks and options.
Buckminster
Well-Known Member
Are we still on for 007's prediction a year ago? (12th Oct 17):
$550 is merely a pit stop
By mid 2019 this could very possibly be a $2900 stock or 8 X current SP of $355
That's exactly how I'm positioned
Let the games begin
Who do I need to call to make this happen?
$550 is merely a pit stop
By mid 2019 this could very possibly be a $2900 stock or 8 X current SP of $355
That's exactly how I'm positioned
Let the games begin
Who do I need to call to make this happen?
The biggest takeaway from Q3 conference call is that FSD and Tesla Network are still coming. Tesla’s FSD approach makes perfect sense. I think this part of the company should worth $100B today. It will worth much more when it’s deployed. We will see how it goes.
I will hold my shares tight and continue to add more shares when I have new cash in the account. I’m not a fan of using margin, will wait when more cash is ready. Have some 2021 leaps, will hold those. In the long run, how many shares do you own, that’s what really counts.
Just a few days ago, I explained to my wife why I don’t like the idea of diversifying into Amazon, Apple and Google. In my view, in the next 10-15 years, Tesla is highly likely to become one of those giant companies. If those largest ones can double in that time frame, Tesla would have to gain 30 fold to reach the same size. If Tesla buy back half of the shares in the process, the per share price could go up 60 fold. I see a clear path how this is coming, so not going to diversify. Instead, I will continue to buy more shares with new cash.
I will hold my shares tight and continue to add more shares when I have new cash in the account. I’m not a fan of using margin, will wait when more cash is ready. Have some 2021 leaps, will hold those. In the long run, how many shares do you own, that’s what really counts.
Just a few days ago, I explained to my wife why I don’t like the idea of diversifying into Amazon, Apple and Google. In my view, in the next 10-15 years, Tesla is highly likely to become one of those giant companies. If those largest ones can double in that time frame, Tesla would have to gain 30 fold to reach the same size. If Tesla buy back half of the shares in the process, the per share price could go up 60 fold. I see a clear path how this is coming, so not going to diversify. Instead, I will continue to buy more shares with new cash.
Mark Spiegel and David Tamberrino, and convince them all to go long.
Loved the last sentence: the moderator stated that this result came out of nowhere, and Gene Munster immediately corrected that by saying that was not correct, that it was always the plan of Tesla to get to this result, but that nobody believed it.
‘Nobody’ definitely doesn’t include a lot of the members here.
Nicely put!
I remember 4 years ago, after the debut of AP1, back when it could read stop signs and lights, come to a complete stop, drive highway to highway; at the first customer demos, it actually stopped at the stop sign at the end of the track. And then shortly thereafter Musk made some comments that FSD was just a couple years out, with vision and no lidar. I was just a noob here at TMC during that time. Talks on the investor threads were talking of the coup Tesla was going to pull off with FSD. The daydreaming and excitement was real. Tesla was so far ahead of every other automaker, the data it could gather because every car was learning from every other car. FSD seemed it really was just right around the corner. Those were exciting times, but there was one person that really was quite the buzzkill. He kept saying that everyone that said FSD was only a couple years away was WRONG. I remember being quite annoyed that he spoke with such unfounded authority. He could’ve at least said it was just his opinion, right? And at times it would take over the conversation of the investor thread. This one poster telling most of us here we were wrong, and no matter the quantity of people that tried to dispute his claim, many experts in similar fields could not dissuade him. Eventually it appeared the board just agreed to disagree. Long story, long, that dissident was @neroden.
I really don’t know how far out FSD is. 4 years ago, I thought it would be here already, only caveat being government approval. I also feel like the AI team at Tesla is perhaps the best in the world. It does appear they are on the right track. And in 6 months, there will probably be routes that could be publicly FSD ready. But as @neroden said 4 years ago, the edge cases are the real challenge. The cop directing traffic. The one lane bridge. Etc. Etc. I feel Tesla hasn’t even begun to tackle these problems.
Anyways, I guess to make this market action relevant I only caution against over optimistic views and timelines especially as they regard to investing.
Fact Checking
Well-Known Member
Called it a couple of days ago:
and here:
"Tesla has a ~40% tariff disadvantage, so I believe they might start next year with a Model 3 assembly line which would get parts (body in white, battery packs, etc.) from the U.S.
Then as the Shanghai factory and the local Chinese supply chain expands and matures they'll gradually substitute more and more parts with locally made ones.
In the end everything is going to be made in Shanghai.
This would be a viable plan to both scale up GF3 production gradually, and to avoid the 40% tariffs.
Then as the Shanghai factory and the local Chinese supply chain expands and matures they'll gradually substitute more and more parts with locally made ones.
In the end everything is going to be made in Shanghai.
This would be a viable plan to both scale up GF3 production gradually, and to avoid the 40% tariffs.
Note that the Chinese tariffs mostly apply to finished goods, not to car parts/bodies/panels - so even if they start with final assembly like they are doing it in Europe they already improved their costs by 40%.
That 12 to 18k is only form Fremont and China. Euro production not included.
It seems we all knew what’s going on, but the analysts don’t know.
- Status
Similar threads
- Locked
- Locked
- Locked
- Poll
