Thread by @TaylorOgan

1/ Thread following up on our $TSLA Q3 production & delivery estimates from last week: our estimates were high for three main reasons, which we will go into.

@snowbullcapital

2/ Our S/X production estimates were off by only 149 vehicles, & S/X delivery estimates by 700. We thought there would be a bit more inventory this quarter, but our accuracy here is pretty consistent with previous quarters, as we used the same methods. 3/Y is where it was hard.

3/ S/X production lines (GA1 & GA2) have changed very little since Fremont reopened, and we expect similar production capacity to remain through the next 3+ quarters on these lines.

4/ As for 3/Y, the first aspect we overshot was not thinking Tesla had started VIN binning Model Y yet. We think Tesla is assigning actual Model Ys to ~82-85% of VINs in the VIN batch they registered for with NHTSA.

5/ We saw Tesla do this in 2018 when it began allocating production VINs over large ranges, thereby no longer assigning VINs sequentially, most likely out of spite for day traders who, according to Elon, were trading based on tracking Model 3 weekly production rates.

6/ That’s really what put a stop to Bloomberg’s Model 3 Tracker. We’ve spoken with @TroyTeslike extensively about Tesla’s history of doing this. He has great insight on this. Here's what he said on it back in 2018 when Tesla started VIN binning Model 3.

https://twitter.com/TroyTeslike/status/1078417406379417600?s=20

7/ There’s a really interesting story to be told here about Tesla VIN binning, but that is for another thread. Let’s just say the “German Tank Problem” is still relevant, à la Wright’s Law. Seriously though, an investigative journalist should write about this. It’s crazy.

8/ We didn’t anticipate Tesla VIN binning so drastically this early into Model Y production, which we think overshot our Model Y production estimates by 6,186. The next part where we were overly ambitious was Giga Shanghai.

9/ Given what we know about Fremont production heads visiting GF3 at the end of August to prep for new general assembly lines at GF3 (potentially coming online this week now that the factory is reopen), following the shutdown, we thought GA3 was more efficient than it was in Q3.

10/ GA3 has been Tesla’s general assembly blueprint upon which future assembly lines will be based. Initially, GA3 was hyper-automated for Model 3 production. Amid “production hell” in 2018, Elon said, “Excessive automation [on GA3] was a mistake.” So they built the tent.

11/ We expected GA3 to gradually reintroduce robots, thereby increasing Model 3 production @ Fremont. Given the large robot orders arriving at Shanghai throughout Q3 we mentioned last week, we thought GA3 had become more automated, since it was supposed to be the template for GF3

12/ We now think the new 3/Y lines at Giga Shanghai will be the template for highly automated batch production, upon which GA5 at Fremont will likely be based. Tesla has been posting a lot of job openings for new general assembly production at Shanghai over the past few months.

13/ Most of these job postings are in Mandarin (which is new) and list extensive experience on highly automated automotive production lines among the requirements.

14/ The most automated automotive assembly lines in the world are in China. Tesla is trying to poach many Chinese engineers from other highly automated lines. In some cases, even, Tesla has reached out directly to the lead engineers of these lines with attractive offers.

15/ Some of the job requirements we have seen even list experience with specific robot models, which align with the recent Kuka & Fanuc robots Tesla received this past quarter.

16/ What we discern from this very clearly shows GA3 is nowhere near full capacity, or at least the capacity Tesla wished it could reproduce at GF3. It seems Giga Shanghai is being built from a near-clean slate relative to GA3. This is a big deal.

17/ The current Model 3 lines at GF3 are fine, and we can assume running smoothly. The NEW lines, especially the dedicated Model Y line, are where it gets interesting.

18/ It may take GF3 a bit longer to ramp the newest Model Y line, but when it is firing on all cylinders, it will be the most efficient and impressive assembly line in automotive history.

19/ That being said, GA3 may be ~8,000 units short (in Q3) of what we expected it to be by this time; these production bottlenecks will not be passed onto GF3. GA3 is no longer the pinnacle of Tesla’s manufacturing prowess.

20/ The 10-day shutdown of GF3 at the end of the quarter was supposed to be retooling of existing Model 3 lines, and we anticipated new GAs for 3/Y to be the focus. While the September numbers aren’t yet out, it seems a factory-wide slowdown/halt did occur.

21/ We expected GF3’s August retooling to outweigh the 10-day shutdown at the end of the Q by some 5,300 units; however, we no longer think the retooling efficiency benefits have been realized. This is good news for Q4.

22/ From this quarter on, it will be increasingly more difficult to quantifiably project even a quarter out where Tesla will be, given new factories, new GAs with different approaches, new models, the state of FSD, 4680 cell production, increased competition in China/Europe, etc

23/ Given that the stock clearly doesn’t mirror its financials, and how wholly different Tesla is, we believe traditional investment methods should not be applied to $TSLA, and investors should instead value the company based on the impact they think it will have on the future.

If this doesn’t become the new model for investing, I’ll have to find a new industry.