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.
Q: What drove S&X GM to rise 5% in the quarter? Also makes the mistake of asking about Model 3 margins. Why can't these guys ask one question at a time?

Elon: rambles for 2 minutes, and gives apology for prior call. And talks generalities about challenges of production lines, but doesn't address the S&X GM change.

Deepak: Likewise.

8 minutes on this one Q that wasn't answered. Somehow Elon moved to the story about reservations...

In the last month, more orders for dual motor and performance cars than the rwd version. Also, lots of demand from people who didn't have a reservation.

Now he's given a life story bio about his VP of sales?

Elon asks his own question: What are the five top non-Tesla trade-in cars for the Model 3:

Prius, BMW 3s, Honda Accord, Civic and Leaf.

I think that’s called controlling the narrative. ;)
 
Second Q: China Gigafactory. What are cost and production capability.

Elon: again apologies for last call. Should cost a lot of less than GF1. Closer to $2B cost, at 250K vehicle per year rate. Includes pack production, vehicle, etc. DOES NOT INCLUDE cell production. Still figuring that out for Shanghai.

Elon: So, less than half GF1 cost to get to 250K level (but I thought GF1 level was 500K?).

JB: in GF1 on Model 3 line, has been able to speed up production speed by 20-30% in some areas just by tweaks, no extra capex. Including battery cells. So GF3 should incorporate those efficiencies.

Elon: Jerome led GA4 assembly line.

Now they are shooting the *sugar* reminiscing about the G4 line. I feel like I'm at a bar quaffing beers with Elon and Jerome. They found that they could move cars around by pushing them on a 1% grade. OK. GA4 is actually less labor intensive than GA3. For example, GA3 repackages parts from incoming trucks to boxes for the GA3 line. GA4 has trucks just backing up to the tent at the right place in the line.

12 minutes on that question.
 
Third Q: Capital spending plans. How do you plan on funding without going back to capital markets. And is there a notice preventing you from raising.

Elon: No expectation of raising equity. At all. Expect to borrow funds for China factory (local debt). Could raise money, but don't need to.

Deepak: Reiterates.

Elon: Default plan to actually pay off debts. Not refi, actually pay them off. No such notice from regulator.
 
How are you going to fund growth?? Regulator outstanding.
Elon "not raising any equity"
"china default plan use loan from local banks in china and fund with local debt"
"we could raise money but we don't need too"
"we start paying off debt, for example convertiable due soon 900m we expect to pay it off"

Sounds like they are not going to refinance, no capital raise and pay of debt when it comes due.
 
Elon is touting the Tesla AP chip saying it is much more efficient than the current GPU architecture. Basically, its what they had 2 years ago with Mobileye. Cough, cough.

They said current software runs 250 frames per second on the Nvidia hardware, runs 2000 frames per second on their new bespoke hardware. No telling if that's 250 fps of all cameras in use, or 250 fps shared across them (i.e. if 5 front / front side cameras active, only 50fps per camera), but I doubt that the mobileye hardware they had before could perform the same work with that much headroom. EyeQ1 was 0.0044 "TOPS" (trillion operations per second), EyeQ3 is 0.256, EyeQ4 is finally a big jump to 2.5 and EyeQ5 will be 24 "TOPS". I'm going to guess Tesla's bespoke hardware is closer to EyeQ5 than EyeQ1. It could be closer to EyeQ4 than EyeQ5, but I doubt it - go big or go home.

Making a guess using the information at Drive PX-series - Wikipedia to try and scale the TFLOPS of the existing Tesla hardware to NN "TOPS" ... it looks like TOPS of the existing Tesla Nvidia based hardware is probably 8-16 TOPS. So getting 10x performance improvement (even accounting for improved efficiency of dedicated design versus general purpose) seems that it should at least meet if not exceed EyeQ5 "TOPS"
 
Status
Not open for further replies.