-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
- Status
mmd
Active Member
Article also says, Powerwall 2 production will start later, in the next few weeks. Then, where are the 2170 cells being produced right now, and for the next few weeks, going? All the Powerpack projects already seem to be complete. Can anyone explain?
BTW, if the GF capacity is 3.6 GWh/year (a very optimistic figure, as I showed earlier), how is the 'economy of scale' achieved with this?
Also, this only equals 60k 60 KWh packs a year for Model 3.
dennis
Model S Plaid
They won't need to build extras for their test fleet - they can just source the unsold ones on dealer lots.
I have yet to see a Bolt in the wild here in Silicon Valley. And I live in the same town as Woz.
mmd
Active Member
How can you even THINK that, when the projects ARE getting reported as soon as someone knows about it? Check the links. The Kaui project was known back in Feb 2016. From Tesla's comment, it just completed. The SCE project was known since Sep 30.
So, again, how can you think that? Can you show examples of large projects that were known much later than when Tesla started working on them?
Also check the part I quoted. Tesla is saying, the bottleneck for Powerpack is behind them. If the GF will produce @300 MWh cells a month, that's 900 MWh a quarter, for 64k Powerwalls @14 KWh! At $5500 a powerwall, that's $353M in quarterly revenue! So, obviously, my 300 MWh/month cell production is a huge over estimate, since Straubel said TE revenue won't be material enough to be shown separately for a long time.
PS: I will just choose to ignore the personal attacks from certain other posters. Such comments have negative value for the thread. I take it, these posters have nothing meaningful to add..
Last edited:
MartinAustin
Active Member
Chevy is desperate to claim "firsts" before Tesla. Listen to Motorweek laud it as the first long-distance BEV under $30,000 for example. I'm sure Chevy's marketing department is pushing that claim.
MartinAustin
Active Member
Home page live player - Flash
Live coverage of a NASA/SpaceX presentation w/ Gwen Shotwell re: Pad 39A
(duh - also on NASA TV right now)
Live coverage of a NASA/SpaceX presentation w/ Gwen Shotwell re: Pad 39A
(duh - also on NASA TV right now)
Last edited:
Cash is king. Tesla wants to generate as much cashflow as possible to fund Model 3 and GF ramp-up.
Orders for Model X are very strong, which is part of the equation. But my speculation is that they can generate more overall cashflow by prioritizing Model S production -- or perhaps you could look at it as not slowing down S production to make room for more Model X.
Supporting this theory -- they kept taking orders for March US delivery for X100D and XP100D for one week after X75 and X90 deliveries were pushed back to May on the website.
My hunch is that because S production has been fine-tuned over 4.5 years and S takes substantially less build time than X, they can generate more total cashflow from one day of S production than they can from X, even though X's ASP is higher.
Will be interesting to hear what they say on the earnings call re internal cash generation v. potential need for a cap raise for Model 3 production. Cash generation in Q4 should be solid (before CapEx), but will be held back a bit by poor exchange rates and deliveries delayed into Q1.
Q1, however, should be very strong for cash generation, assuming they do not lose too much production time due to Model 3 equipment installation:
- Much higher deliveries than Q4 (my SWAG is 28K v. 22K)
- Significant price increases outside the US to correct for exchange rates
- AP2/FSD cars delivered for the entire quarter v. partial quarter
- X60 almost completely phased out versus partial phase out for Q4
- $68K S60 for entire quarter v. partial quarter
- Improved production efficiency
- Production tilted toward high margin vehicles -- S, X100 and XP100D
Last edited:
Fwiw, youll get more like 10-20x performance out of a gpu if the code for cpu is likewise optimized to use avx extensions and memory well. 1000x won't happen in any fair match. When i did some gpu coding nearly 10 years ago, simply learning cuda had me acutely aware of how to optimize for it, but when i went back and learned similar tricks on cpu, the difference narrowed. Also training mode usually gets better multiples than online mode because you aren't waiting for more data to fill the various pipelines.
Form the Tesla blog dated October 27, 2016:
"To date, nearly 300 MWh of Tesla batteries have been deployed in 18 countries, and we anticipate the impact and growth rate of energy products around the world to be far greater than that of electric vehicles alone."
On October 27, 2016 neither Kaui, nor Mira Loma BES were deployed, so go ahead and prove @MP3Mike being wrong by saying that "We don't hear about most of the in process projects." Please do so with links, because you did NOT so far, just by listing two projects. So list 300MWh worth of BES projects that Tesla deployed before October 27, 2016.
According to executive quoted by Electrek a while ago Tesla is in negotiation for several large scale projects each of which could more than double everything what they installed so far. So we are talking about several projects on the scale of 600MWh each. They also have close to 38,000 orders for power wall.
So in summary we do not know a lot about the TE pipeline at this time, but if history is a guide, it certainly does not look as ominous as one would conclude reading your posts.
BTW, saying "I will chose to ignore" does not qualify as ignoring...
Last edited:
racer26
Active Member
For a typical GPU, yes, performance from low-mid range GPUs under CUDA is 10-20x. I was reporting the FLOPS. Hard to argue with a metric that is literally operations per second.
Good point. Yet it is hard to see how a 60D Model S generates more cash than fully loaded P100D. Margins of the X were already sharply up and should be tracking the S quite closely about right now. Maybe the latter takes longer to build and that moves the needle once again?
Maybe @vgrinshpun can shed some light on this, because he is the resident encyclopedia on all things factory for Tesla but I believe they can basically build S/X at will without production of one slowing down the other.
Certainly. Although I rather tend to agree with consensus that the need for a cap raise for Model 3 is quickly going down to zero as time passes. As you note TA cash generation has several convincing positives for this quarter. I also expect them to continue to monetise historical PPAs from SolarCity as well. With rising interest rates they may not make as much on them but it remains a direct cash injection anyway you look at it.
Since it's over 18 countries I rather suspect the number to also include Powerwalls. Just 5000 of those delivered worldwide would fill it up. Several hundreds went to Eneco in the Netherlands for example.
It seems clear to me that they are production constrained. So the question is not whether they can, in theory, make X or S interchangeably, but how to optimize production to generate maximum cashflow. To make up some numbers for purposes of illustration, if the bottleneck is final assembly and they can produce 1.5S for every X in final assembly it could make sense to produce more S than X even if marginal profit on each X were higher. I am assuming they can basically make S or X interchangeably. If that is incorrect, and one of the lines is dedicated to S only and is the bottleneck, a theoretical explanation is that they simply have reached their production limit for Model X but are not as constrained with Model S. I tend to doubt that is correct because I understand that final assembly is the bottleneck, but would be interested in views of @vgrinshpun or anyone else who has looked deeply into this.
Other possible explanations are they need to take X production off line to make improvements for X production or make some design or part change. I think those are less likely but possible they are part of the explanation.
I agree -- look forward to hearing more on the call.
Fair point. Unlikely that total profit for S60 production exceeds profit on XP100D, even assuming they can crank out significantly more S60s. But they may not have liked the optics of keeping orders open for XP100D while delaying S60 or had other reasons for not slicing it that finely. I do think the end result of more S production in Q1 will be higher total production volume and higher overall cashflow than if they had kept S and X delivery dates on the same timeline, and I believe it is likely that is part of the thinking behind not delaying S production to make room for more X production (assuming they have the capability to do that).
This is all speculative so would be interested to hear others' thoughts.
As others have pointed out, your assumptions are all over the place. It is almost like you haven't done your homework at all. Did you view that Gigafactory tour video from last summer I posted a while back? Easy enough to find on Youtube. Plus, there are the reports from the additional tour given in early January.
Clearly, for PowerPacks installed in Q4 and earlier, they could not have used Gigafactory production at all. We know that Powerpack 2 production installed through at least Q4 used 2170 cell production from Panasonic. Therefore, it is hard to look at previous installs in order to estimate the Gigafactory.
Let's put your 3.6 GWh number in context. That is about 50% more than LG Chem's total automotive cell sales in 2016 for all those automakers. It would represent about $1.25 billion in revenue, maybe $200 million in gross profit or $50 million a quarter at that scale after production ramp. The first phase is no longer roughly 7 GWh... the first line of the first phase is likely around 8-10 GWh. If the Gigafactory production is 50-50 auto and stationary storage, then we are looking at more like $2.7 to 3.4 billion in revenue, or $400 to $500 million in gross profit per quarter. Take a 50% discount for the first year ramp. And that is with a fraction of the capacity online.
Thumper
Active Member
NASA
Today's @Spacex launch was scrubbed due to a second stage thrust vector control issue. Updates: nasa.gov/live pic.twitter.com/NSamN1RhEH
2/18/17, 10:08 AM
Next launch window is 9:38am tomorrow (if they decide to use it)
edit: more from Elon:
@elonmusk: All systems go, except the movement trace of an upper stage engine steering hydraulic piston was slightly odd. Standing down to investigate.
Interesting video reportedly from Tesla Santa Monica store stating that GF achieves a "35% reduction in battery costs."
I believe the earlier statements I saw were something along the lines of "more than 30%" cost reduction so this is a bit more specific.
Spy Video: Tesla could be hinting the Model 3 is coming faster than expected
Quality is a little shaky since it is a video of a video ....
Per the article title -- interesting to speculate whether Model 3 in the video is a hint from Tesla that Model 3 is getting close (on time).
Edit: I looked at the video again and focus is on GF with a quick glimpse of Model 3 so I personally don't think the video means much for Model 3.
I believe the earlier statements I saw were something along the lines of "more than 30%" cost reduction so this is a bit more specific.
Spy Video: Tesla could be hinting the Model 3 is coming faster than expected
Quality is a little shaky since it is a video of a video ....
Per the article title -- interesting to speculate whether Model 3 in the video is a hint from Tesla that Model 3 is getting close (on time).
Edit: I looked at the video again and focus is on GF with a quick glimpse of Model 3 so I personally don't think the video means much for Model 3.
Last edited:
- Status
Similar threads
- Sticky
- Locked
