Fact Checking
Well-Known Member
First, you lost me at skabooshka. I tuned out right after that.
Second, ignores competely 130k VIN registration Pace and tax credit phase outs.
Note that a closer look at skabooshka's daily Tesla production data gives support to your thesis that the 100k annual production for Model S+X is a soft limit that could be significantly exceeded for a quarter or two.
The first clue is in the distribution of Model S+X daily production records. Here's the best 13 days from 38 days of data, sorted by S+X production:
date |
07/11: |
07/25: |
07/26: |
07/28: |
07/14: |
07/20: |
07/13: |
08/01: |
07/12: |
08/02: |
07/19: |
07/18: |
Note that:
- 34% of all days observed were able to hit 320 or better daily production rates, which corresponds to a 2,240 weekly, 29,120 quarterly and 116,480 annual S+X production rate
- There's 6 days at 350+ rates, where 350 daily production corresponds to a 2,450 weekly, 31,850 quarterly and 127,400 annual S+X production rate
- The record daily production is 400 S+X's, which corresponds to a 2,800 weekly, 36,400 quarterly and 145,600 annual S+X production rate
- The 'best' days are Wednesday and Thursday. This suggests that the limit is assembly line staffing related, not supply chain related: it's unlikely for the supply chain to have significantly weekly fluctuations so late in the product's life cycle.
- It's unclear to what extent the options mix of the Model S/X's made influences production rate: Sunroofs and Falcon Wing Doors take more time to assemble, but it's unclear to what extent that influences total throughput.
- There do not seem to be any significant throughput interactions between Model 3 and S+X assembly throughput: "record" S+X days also feature pretty good Model 3 production days. I.e. the paint shop or absolute levels of staffing is not a visible throughput limit in this data set.
- Note that some of the best S+X production data was in August, shortly before skabooshka shut down the data feed. This is one of the strongest hints yet I believe: statistically it's unlikely that the 3rd and 5th best result in the data set would be on August 1 and 2, just 5 days before skabooshka shut down the data. This supports your thesis that S+X production is ramping up, and maybe skabooshka's decision to discontinue this data was done because later data points showed even better production that did not support his agenda.
Note that while the supply chain might be set to a 25k/quarter rate, the main limitation is Panasonic 18650 cell manufacturing supply according to Tesla:
- which is unlikely to be a 100% hard limit on Panasonic's side, and they'd want to maximize 18650 production throughput with existing facilities too, before S+X migrates to the 2170 cells.
- while Panasonic is unlikely to expand their 18650 capacity without Tesla entering into a long term contract (which they probably won't do, they'll migrate all their products to a 2170 base), there's no reason for Panasonic to play games with 18650 supply: it's in their interest to maximize that output for the next year or so until the 2170 migration, as long as there's no significant capex required
- even if there's a hard limit, that limit can be exceeded for a quarter through warehousing, by storing cells in Q1 and Q2 and then using the excess supply in Q3 and Q4. Cells can be stored for a couple of months (as long as their minimum charge levels are maintained), without any significant impact to their life time. This does consume cash flow and increases inventory overhead though.
Another data point in the skabooshka data series is the July 4th week shutdown:
date |
07/01: |
07/02: |
07/03: |
07/04: |
07/05: |
07/06: |
Note how S+X assembly was at ~66% capacity, while the Model 3 lines were down for 3 days and were on low throughput for another 2 days, but the S+X assembly lines were at 70%+ capacity on average. This suggests that Tesla didn't want to lose the production output from the S+X even during the July 4th weekend, and probably paid top hourly wages to keep the lines staffed.
On the flip side, there's a couple of additional factors that count against your thesis, I think:
- While Tesla reiterated it twice this year that new orders for Model S+X is at record levels, we don't know whether this is true for the U.S., or mainly driven by global demand. In particular the Model 3 Performance might already be earning Tesla more margin per unit (~30%-40%) than a low trim Model S or X which probably has around 20% of margin. It makes sense for Tesla to promote low end S/X sales to M3P sales: if an owner has an upper limit of paying 80k for a Tesla car, it's better they buy a M3P, which:
- only uses up 80kWh of 2170 production and not 100kWh of 18650 production (which has a supplier limit) - so it utilizes the Gigafactory better
- has 'full trim', which have much higher margins in excess of 30% according to Munro
- uses Autopilot and other software options that have near 100% margin
- customers who want a Model X and can pay 100k+ for it are unlikely to be lured by even a 80k full trim M3P - so I see no negative pressure on high-trim S+X sales
- BTW., note a nice side-effect of this dynamic: as long as there's record orders for S/X, converting low trim S/X sales to high-trim M3P sales improves the margins of both the Model 3 and the Model S/X product families - win-win.
- only uses up 80kWh of 2170 production and not 100kWh of 18650 production (which has a supplier limit) - so it utilizes the Gigafactory better
- I believe this explains why Tesla has discontinued the 75 kWh Model S/X model and is promoting the Model 3 Performance aggressively: beyond the frequent promotion by Elon, the M3P test-drives there also a cash-back or upgrade-to-any-other-Tesla guarantee (!) which the Model S/X does not have (they are build-to-order). All of these are pushing U.S. demand towards the higher margin M3P.
- If we count the U.S. as about 25% of the S+X market, that's a regular order flow of 6,250 U.S. orders per quarter. While the reduction of the U.S. tax credit from $7,500 to $3,750 and then to zero by the end of 2019 will increase demand in the U.S., I think the biggest effect will be in Q4 - especially as people know that S+X gets delivered much faster. I.e. I'm not 100% sure the new U.S. orders are necessarily there to tick up to 30k or higher quarterly S+X production.
- July was clearly a slower month with about 1-2k of S+X baseline production shortfall - which makes it more difficult to reach new records of production. Also, the best way to help Q3 financials is to increase deliveries, and S+X deliveries take much longer on average due to the global delivery chain. While Tesla could re-order to prioritize U.S. deliveries near the end of quarter, I'm not certain there's extra 4k-5k U.S. orders on top of the regular baseline that would allow them to do so to reach 30k and higher Q3 production.
(I'm wondering whether @luvb2b concurs with this analysis.)
It would be nice to have 1-2 more leaks to Fred and more S+X VIN registrations to be able to measure their Q3 S+X production with higher accuracy. Because patiently waiting 3.5 weeks for the production and delivery report is so last century.
Last edited: