Gigapress
Trying to be less wrong
Update on this: I found @Troy's own production estimates, as of Nov 3rd. I think these are off by at least 40k units, probably 50k, and maybe 60k.Troy's estimation method is garbage. He is looking at lagging indicators and making lots of questionable assumptions, which is inherently flawed when modeling for factories that are rapidly growing production and deliveries. The only realistic way we're getting 415k deliveries in Q4 is either major extended production shutdowns or a huge addition of like 69k vehicles to inventory in transit in an epic single-quarter unwinding of The Wave.
Fremont made about 142k last quarter. Conservatively 146k in Q4 I think. Tesla has reported S&X are still gradually ramping. Troy is estimating 146k too so no conflict there.
The massive problem is the other estimates of deliveries:
Shanghai produced 88k in October, with a holiday week and presumably partial production rate. Also they're probably still making tweaks and fixes after the recent line upgrades. They also have a long-term trend of increasing production in nearly every month that's uninterrupted by major external forces. 90k in Nov and 92k in Dec is likely, if not a bit more. 88+90+92 = 270k for Q4. How the heck would they only deliver 214k? What is Troy thinking there? He keeps harping on demand collapsing in China, but really? ~56k cars from that factory not being delivered in Q4?
- 214k Shanghai
- 29k Berlin
- 26k Austin.
Berlin and Austin already are making 2k+ per week and are ramping faster than ever. Conservatively assume 3k per week quarterly average and only 12 weeks worth of Q4 production between the winter holidays and other downtime, and thus we get 36k each. Remember, Tesla guidance of 5k per week run rate by Dec 31st has not changed.
146+270+2*36 = 488k of production for Q4, with plenty of headroom for an extra 10-30k depending mainly on Ber&Aus ramp speed.
Sanity check:
Tesla again has guided for hitting 50% YoY production growth2022 production was 930k as of Q32021 production was also 930k50% growth --> 1.40M1.40M - 0.93M = 0.47M0.488M is slightly greater than 0.47M, aligning with confident expectation of slightly exceeding 50% targetSo barring any disruptions, 488k production is in the bag based on known production rates. If so, Troy is predicting an incredible 488 - 415 = 73 THOUSAND cars being made in Q4 but not sold. Tesla is selling about 5k cars per day right now. 73k/5k = ~15 days of sales. Troy's estimate implies roughly a tripling of vehicle inventory QoQ from 8 days to 8+15 = 23 days, or roughly 115k cars.
Will production likely exceed deliveries as Tesla continues unwinding the wave? Yes. Will inventory triple? Very unlikely, especially in light of Tesla's firm comments on strong global demand and aggressive expansion of production output.
View attachment 875183
(Source: Tesla quarterly reports)
Source
Troy actually published estimates that imply literally zero growth at Shanghai and growth screeching to a halt at Berlin and Texas in late October. This wouldn't really bother me except that 1) he charged people money for this analysis on the basis of ostensibly having expertise in this area and 2) it is influencing the market.
- When Troy published 246k estimated Shanghai production for Q4, we already knew per the Chinese Passenger Car Association that Shanghai had produced 82k in September. 82k * 3 = 246k.
- As of Nov 3rd we also knew that Berlin had surpassed 2k per week in the last week of September and was, in Elon's words on the Q3 call, "ramping rapidly". Despite this, Troy's estimate for the Q4 average was 30.0k / 13 = 2.3k / week. The official target is still 5k/week by the end of the year.
- Texas had reached 2k in the 3rd week of October according to Elon on the Q3 call. Compare that with Troy's estimates: Texas: 27.2k / 13 = 2.07k / week Q4 average. Again, still officially targeted for 5k/wk EoY.
Last edited: