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

TSLA Market Action: 2018 Investor Roundtable

This site may earn commission on affiliate links.
Status
Not open for further replies.
I've been thinking about what this means:

"We are aiming to bring portions of Model 3 production to China during 2019 and to progressively increase the level of localization through local
sourcing and manufacturing."


What's the logical priority for China to complete first?

* Overhead. A factory is not just particular production lines, but all of the transport, utility, and other supporting infrastructure that the factory is useless without. Big focus for the next couple quarters.

* GA line. This is obvious - they have to complete the vehicles there to gain the local manufacturing benefits, so they need at least a single GA line. GA is also one of the production constraints at Fremont. Tesla demonstrated the ability to make a GA line in a matter of months with GA4. I expect a GA line to start up there shockingly fast - not Q4, but Q2-3.

* Paint shop. This is the other big bottleneck at Fremont. Tesla can divert BIWs straight to export to Shanghai. Wouldn't be surprised in the least to see a local paint shop open up in Q4. Maybe even Q3, but I'd bet on Q4; paint shops are pretty complicated and require a lot of fine tuning to get consistently good results.

AFAIK, Fremont has never been limited by stamping or plastics, and apart from one short period early on, not welding either. Giga seems to be capable of scaling easily and effectively indefinitely, so GF3 production of packs and motors is obviously not going to be a priority. But with a GA line and paint shop in Shanghai, they should be able to significantly expand their total global output.

Wouldn't be surprised to see Giga 4 construction starting much faster than people are expecting, using the exact same approach.

There's a reason Tesla has walked back focus on 10k at Fremont (I've noticed analysts keep missing this part): because they can start up production "not at Fremont" surprisingly quickly. Not full production, but it doesn't need to be. I see Tesla scaling Fremont to 7k/wk sustained by Q1/Q2, with future scaling Model 3 scaling focused first on Shanghai, then Germany. Whenever they can't squeeze more production out of a given line at Fremont without a significant capital investment, a corresponding line will be built at Giga 3 or 4 for it.
 
Last edited:
One thing to keep in mind while we are all watching what the stock will do today. the overall market took a huge hit yesterday. That fact made Tesla's surge even more impressive, but a weak overall market will limit any gains that may or may not be forthcoming. Just saying that I don't believe that a seemingly small gain or even a loss necessarily reflects on where the market feels Tesla should be. We'll see how it all plays out.

Dan
 
Disappointing report from Reuters. So clearly negatively biased.
That's just the first taste of several more negative spins we will see from the usual suspects. Maybe we get lucky and convert 1 or 2? I'm doubtful about that until after Q4 ER.

This will be a very interesting and difficult to predict day. Is it possible we could take out the AH HOD? I don't know, maybe with the market surging today. If we did, that could add some fuel to the climb. If I were a short and still bearish on TSLA, I think I would cover now and re-short after this climb. Is there really much money to be made shorting TSLA right now at this level? Probably only if the Nasdaq spirals down a lot further.
 
I just think it's irrelevant to the question of when we'll get full self driving, because they're not looking at the problem of how to teach the neural network to drive.
This is a very fundamental philosophical question in FSD, but my personal feeling is NN should never be used to drive.

NN is perfect for perception, because perception is too big a problem to define by rules, and it can tolerate failures.

But using NN to drive has a huge issue, it's that we don't have a framework to make NN explainable, also you could argue as soon as you make an NN explainable, you already converted it into a decision tree.

With the inherent lack of explainability within NN, comes huge uncertainty and liability, anyone who tries to use NN to do driving is just cutting corners, by throwing data to a problem they can't or don't want to define.

If you want to bring the "trolly car" problem into the discussion to make a point for NN, I will say we would have AGI by the time NN can solve that, and by that time, we don't need FSD cars anymore, tunnels and rockets takes you anywhere you want to go, if you still need to go places at all.

*EDIT: typos*
 
Apparently the new bear thesis is that Tesla is cooking the book.
I kid you not.

That's not the "new" bear thesis, it's the backup thesis they've been running to whenever their predictions didn't really match company's reports. Now it just became the main thesis.

And, yeah, when you're proven wrong in your "analysis" by the actual report, just yell "FRAUD!".
 

The analysis quotes are very strange. Even if ASP and margins decline a bit Q4, production and revenue will surely increase.

Plus production efficiency will improve each quarter for the next few quarters. They will do less dumb stuff. Production hiring will also likely decline. In the medium term Tesla will almost certainly make more cars more efficiently.

Mass production is a step function. Tesla is right in the middle of that step. This is the point where the financials look good as long as demand is good. These analysts should know the basics.
 
Regarding Tesla energy shortage of 2170 cells

Samsung is a major supplier for Tesla 2170 cells. The entire SA project was Samsung cells. There TE cells are supposedly better at some things then the Panasonic cells, though I do not know exactly what ways. I think Tesla has just been focused on Model 3 for obvious reasons and now Giga/Pansonic can start to open up more TE capacity and more refined cells for specific requirements. I think its great that Tesla also has Samsung to help soak up all the battery supply it can for its great products which include all the pack tech that makes it all work. The Cells are only 1/4th of the great tech.
 
Tesla just can't be judged by normal car metrics, they are so much more than that. The fact that even the hard core fans here only ever focus on the vehicles themselves is an indicator of the loaded gun they have.

Also take note of the fact that 50% of trade ins were < 35k new. That's me to a Tesla "T". I didn't trade in to Tesla, but my previous car was 22k new. We've now sold my wife's ailing Subaru Forester (maybe 29k new? not sure) and are using my dads 20k 2007 civic that HE stopped using when he got his 60k model 3.

As much as i HATE when people say Tesla = Apple, in this case, yes, they are iPhoning the ever living **** out of the auto industry. And instead of rising the the challenge, the incumbents are pissing into the wind in an effort to turn it back.
This. My first phone was an iPhone. I have never in my life bought a new car, but that will change when I get a Tesla. I have no idea what my cars were worth new, but as old clunkers (still run though!) they are waaaaay below $35k. I know, I'm not typical of this forum. But this forum is not typical of the population as a whole, and for the mission the population as a whole matters. I see the trade in valuation as a clear signal that they are making inroads into the general population.

And unlike phones, it won't be so easy to copy a Tesla car. Batteries. Safety. EAP. All of it, these are advantages that are not easily surmounted. Which is why I see Tesla dominating the EV market for some time to come. VW seems serious, but the results are not evident yet. As someone else said, not the legacy, but the newcomers: BYD, Geely.
 
Just a reminder of Market Action the day after the 2018 Q2 Earnings Call:

TSLA Aug 02, 2018:

Open: 328.44
High: 349.99 (occurred near close at 15:58 hrs)
Low: 323.16 (MMD; occurred at 09:35 hrs)
Close: 349.54

Volume: 23,159,160
(from NASDAQ)


View attachment 346835

Since this seemed popular, I'll add TSLA Market Action from Aug 01, 2018 too:
(note the SP tapering fade during the After-hrs session)

capture_001_01082018_182913.jpg
 
  • Informative
  • Like
Reactions: PANN and TDog
This is a very fundamental philosophical question in FSD, but my personal feeling is NN should never be used to drive.

NN is perfect for perception, because perception is too big a problem to define by rules, and it can tolerate failures.

But using NN to drive has a huge issue, it's that we don't have a framework to make NN explainable, also you could argue as soon as you make an NN explainable, you already converted it into a decision tree.

With the inherent lack of explainability within NN, comes huge uncertainty and liability, anyone who tries to use NN to do driving is just cutting corners, by throwing data to a problem they can't or don't want to define.

If you want to bring the "trolly car" problem into the discussion to make a point for NN, I will say we would have AGI by the time NN can solve that, and by that time, we don't need FSD cars anymore, tunnels and rockets takes you anywhere you want to go, if you still need to go places at all.

*EDIT: typos*

This. Note that I really don't want to further a NN discussion in this thread, but: NNs are black boxes. They're almost impossible to analyze as to exactly why they made a given decision. This makes them fundamentally unfit for determining how to drive. The physics of "how to drive" is very straightforward, if you know the details of everything around you; this makes "how to drive" an obvious fit for human-coded algorithms, which are explicable and easy to control and tweak. It's the "knowing the details of everything around you" that's the sticking point, and the reason you need an excellent sensor suite backed by a powerful neural net.
 
Wow, it hit 304.44 after opening bell yesterday, and now it flirted with $310 in the premarket. 2% after an amazing quarter... shorts aren’t covering. I feel like it’s going to pull a Netflix unless we get some crazy surprise news soon. I’ve attached my position to the post so I don’t get called a troll for being negative. To be honest I went all in at 380 after Elon’s tweet and happy to be green thanks to calls. Last 10 weeks have been absolute hell... I’ve learned why Elon hates the media and shorts.
Now more than ever, the clock is ticking loudly against the shorts. Just a little more patience...
 
  • Like
Reactions: SpaceCash
That's 90k per quarter, or 364k per year assuming no closed weeks. Let's just say treefiddy a year with the current lines. Great!

7k with essentially no Capex required. 10k would require some downtime to reconfigure and probably insert some new stations to speed up slower parts of the line. Think about it this way.. You identify a bottleneck and you can either find ways to speed it up or if you cant, you just double with a copy of the station and go from 5k/w to 10k/w instantly. The amount of Capex required would be equivalent to the number of stations that need to be duplicated. Also a station could be sped up by adding more robots to the same station if there is room. So now 4 actions done by 2 robots can be done 30% faster by 3 robots or 4 robots. Maybe there is room or maybe they have to reconfigure the station, so downtime would be required to do all of that and my guess is that they have been doing this since hitting 5k burst at the end of Q1, but the limiter has been cells production and as soon as that clears, which it sounds like its clearing now with an additional line and better yields, you could see a jump to 7k/w almost instantly. At some point they run out of easy stuff to fix and have to start duplicating stations. Also, they probably knew they would have to when they originally designed the line, so they have space allocated in the line for duplication, but they still have to shut it down to put in the new equipment.
 
Status
Not open for further replies.