I think we have to face that fact that we are not able to predict the production in this quarter with any accuracy. I have tried to look back through this thread to see what I could find of original data points or analysis, rather than just echo:
VIN assignment
This method is actually very clever. It assumes that although Tesla have been intentionally skipping back and forth in the VINs to throw off any analysis, they will still over time be assigning VINs at a pace that corresponds to production. So rather than tracking the VINs themselves, it tracks the number of people who get VINs assigned. Having said that, Q3 assignment rates are clearly higher than production, because many EU customers got VINs assigned for production way into the future, while that was not the case in Q2. The method may become more accurate in Q4. As an indication of the lack of accuracy, the VIN assignment rate was actually
falling recently. At best the rate could be used as a ceiling for Q3. The rate has been 728 VINs per week over the past two months. (Source:
http://www.teslamotors.com/forum/forums/new-vin-thread-please-post-your-newly-assigned-vins-here)
Max VINs produced/delivered
Several contributors have posted analyses based on tracking VINs that have been produced or delivered. I am not including these, due to my complete conviction that this is an inherently unreliable method. I am waiting for delivery of my car in early December, and higher VINs than mine have been delivered long ago. Many others are in the same situation. q.e.d.
Aug 20 video
This:
http://www.youtube.com/watch?v=TMYmSuicTXE
Elon says "Now we are at around 500 to 550 a week". Assuming this is the average for the quarter, this means a production of
5,900-6,490 cars. Taking into account that they are generally sandbagging in good times, maybe go with the higher number of ~6,500?
Note that although the video was posted on Aug 20, it is unclear when it was recorded.
Employee chatter and rumors
2pearls: "
I Have heard from an informed source that they are presently manufacturing 700+ cars per week. 500 for US and 200 for Europe. Has anyone else heard this or taken a tour to see what they are manufacturing?"
brianstorms: "
I went on a 1300-mile drive up to Northern CA and along the way spoke to various Tesla employees who I asked about the 700+ week figure and they pretty much confirmed that that is what the factory is up to"
Sleepyhead: "I have my own ways of determining how many cars were produced/delivered but I have not had enough time to go through the process. But I can already tell that at a maximum Tesla produced 7,000 cars in Q3. When I say maximum, I mean that it could be a number anywhere from 5,500 to 7,000"
New analysis from me:
Exponential extrapolation
Just a simple exercise of mine. Q1 shareholder letter says they built "more than" 5000 cars in that quarter. A report on this forum cites a congratulatory sign at the factory reporting 5454 cars in Q2. There were 60 working days in Q1, which means 83.3 cars/working day assuming the 5000 figure. In Q2 there were 58 work days, taking into account the week they closed the factory. This gives 94.0 cars per working day, an increase of 12.8%.
Assuming they were able to do 12.8% again in Q3, their daily production would be 106.1 cars/working day. Based on 59 working days (taking into account one week off for the factory), this would give a Q3 production of 6,258 cars.
One could wonder whether the recall depressed production. However, seeing that they reported the cost at only $150,000 and that the cars were fixed at service centers I don't think there was any significant impact.
Conclusion: Taking the above together, I feel that we are going to be in the 6,500-6,900 range, going for a middle estimate of 6,700. It is tempting to put a higher figure based on the employee chatter/rumors, but this data feels like an outlier that must be disregarded. At the same time, the video gives me confidence of at least 6,500.
A summary of the firm predictions of others that I could find:
DaveT: 6600
Sleepy: 5500-7000 (said 6000 in a post in early September)
Vgrinshpun: >7080 (>600/week)
Julian Cox: >6600
Convert: 7340-7930
DrDave: 6490-7080 "at best" (550-600 at best)
Clprenz: 7,750
bonnie, Kevin99 and others: No firm prediction, but expressing a feeling that others are too optimistic
To discuss the other items that go into the analysis of deliveries:
Increase in inventory of loaners and showroom cars: I cannot see how this would not be higher than zero. The word at end of Q2 was that all loaners had been sold off, and they have opened new showrooms. On the other hand, they seem to be pushing hard to maximize Q3 numbers and there are rumors of low stock of loaners also now. I will guesstimate an increase in the inventory of some 150 cars.
Increase in cars-in-transit to delivery: Some (Julian Cox I think) have claimed that cars-in-transit was high at the start of the quarter. I do not have hard data, but here in the biggest EU market there were few reports of deliveries prior to Aug 15. In the Netherlands and Switzerland, the total for July and August was 51, and many of those cars would have been produced in July. I will guesstimate the ingoing inventory at 100. Outgoing is going to be lower than expected, based on a data point of an 80% drop in the rate of new registrations of Model S in Norway from September to October. Guesstimate: 500-700 cars outgoing, will assume 600. The increase would then be 500.
Unless the train crash killed a significant number of cars, Q3 deliveries would then be 6,700-150-500 = 6,050.
----
EDIT:
The more I think about it, the more I think this estimate should not be considered pessimistic in any way. They key question that is going through my head is: "Why would they drop EU deliveries almost to zero in October, and then ramp them up wildly afterwards?". The effect of doing so is boosting Q3 at the expense of Q4 (and some logistics disruptions). If they were beating Q3 massively, there would be no reason to do so.
EDIT2: Felt that a sensitivity analysis would be in order. So here goes:
Pessimistic: 6,200-150-600 = 5,450
Optimistic: 6,900-50-400 = 6,450
Assumptions for pessimistic scenario: 525 cars/week (the middle of the 500-550 cited by Elon in the Aug 20 video). 150 net new non-delivery cars. 700 cars in transit to Europe at end-of-quarter, 100 at beginning.
Assumptions for optimistic scenario: 585 cars/week (an increase of 26.5% on previous quarter, as compared to the 12.8% they managed in Q2). Only 50 non-delivery cars (incl. train wreck). 500 cars in transit to EU after quarter, 100 before.
My gut feeling is that the optimistic scenario is really a stretch.
The pessimistic scenario, on the other hand, still implies a similar quarter on quarter increase in production than they managed in Q2 (of whice they were justly proud).