Updated analysis
I start with just two assumptions:
1. The production rate data points are correct (Q2 shareholder letter - "almost 500" = 485-499), Aug 20 video of EM and the Chinese tourist pic)
2. Tesla's production rate has never gone down week-on-week in the period from end-July and Oct 25
Then I can with certainty say that the production has been 5756 and 6480. These are my "absurd low" and "absurd high" numbers. They are absurd because the production rate is not gradual - it assumes the theoretical worst and the theoretical best scenarios between the known data points:
In the scenarios above, I have assumed that when EM said that production is "between 500 and 550" on Aug 20, it could be anywhere in that range. For the further analysis, I am assuming that if it were between 500 and 510 he would say "approx. 500" and that if it were between 540 and 550 he would say "approx. 550". Therefore I will narrow that range to 510-540. Here are the "realistic" scenarios:
As you can see, I have assumed different scenarios for how production increase happened over the intervals as well (exponential, linear and square root, respectively).
This gives the following production estimates:
Low: 5,921
Middle: 6,051
High: 6,225
The great fit of the linear middle scenario gives me strong confidence in it.Now the question is what the difference between production and deliveries is. Based on the extremely low deliveries reported in October in Europe, I would not 100% rule out that there was no increase in cars in transit to Europe. We have to assume some increase in show room cars (if not loaners), so let's say 50 for that.
This means that the most optimistic realistic scenario is that we have to subtract 50 cars from production figures to get deliveries.
On the other hand, there were some deliveries in Europe and could be quite a bit more than the anecdotal evidence gives an impression of. I would say that it would not be unrealistic to assume an increase in cars-in-transit of 300-400. For loaners and show room cars, those could possibly have increased by as much as 100-150.
In total, let's say that the maximum adjustment is 500.
Based on my gut feel, I will say that the "middle" scenario is towards the low end of the range - so 150 cars.
This gives the following scenario matrix for deliveries:
| Low production | Middle production | High production |
Low correction | 5,871 | 6,001 | 6,175 |
Middle correction | 5,771 | 5,901 | 6,075 |
High correction | 5,421 | 5,551 | 5,725 |
The bottom line: I am going to go with an estimate of 5,900 cars delivered. Furthermore, there is more of a downside than an upside to this figure, as the table above shows. 6,200 would be very surprising - I think the only way that could happen is if Elon was sandbagging in the August 20 video. However, that seems unlikely, since the Aug 20 data fits well with the two other data points.
Discussion of conflicting data:
I think VIN based analysis is fatally flawed by some assumptions that are completely baseless (either "proportional" or "the pipepline is filled up"). I think the huge discrepancy between the VIN data and these pretty firm data points, clearly the VIN data must be discarded.
On the other hand, the registration numbers collected by Maekuz still provides significantly higher estimates than the figures above. We have not been able to get any information on the original source of that data, so we cannot know if it is reliable or not. While this data reduces my certainty about my estimates above, I will not disregard solid data based on data of unknown quality.
- - - Updated - - -
Updated Q3 P&L projection:
Cars and CARB credits | 5900 cars x $101k ASP | 595 900 |
ZEV credits | Guesstimate | 25 000 |
Development services | Guesstimate | 1 000 |
Total revenues |
| 621 900 |
|
|
|
Gross profit | 21% GM ex. ZEV | 150 349 |
|
|
|
R&D expenses |
| 55 000 |
Selling, G&A expenses |
| 60 000 |
|
|
|
Net income (non-GAAP) |
| 35 349 |
Net income/share (non-GAAP) |
| 0.29 |
|
|
|
Revenues (non-GAAP) | 621 900 |
|
Deferred due to lease accounting | 137 000 |
|
Revenues (GAAP) | 484 900 |
|
|
|
|
Net income (non-GAAP) | 35 349 |
|
Deferred profit | 27 000 |
|
Stock based comp | 22 000 |
|
Non-cash interest on convertibles) | 1 800 |
|
Net Income (loss) (GAAP) | -15 451 |
|
EPS (GAAP) | -0.13 |
|
Note about the projections: The uncertainty about ZEV credits and R&D expenses makes these projections highly uncertain. My range for non-GAAP EPS is $0.16-0.38.