I'm not ignoring reality -
A) The start of production is subjective in your chart. Austin started deliveries in Q2 2022. Pretty sure Tesla made more than 30 Y's for all of Q2 2022. Your second quarter of production for Austin is actually it's 1st quarter.
B) The numbers in your graph for the most recent quarter do not add up. Austin did not make more Y's than Berlin. The numbers add up even less when you take into account that your 1st month of Austin production is actually the 2nd month. I don't know where you're sourcing these numbers from.
C) I already acknowledged that they're way behind Shanghai. That's obvious and really has no relevancy here. They were never going to recreate Shanghai speed.
Also your graph for battery production. I don't even know where to begin with that. Tesla is sourcing batteries from more than just Panasonic now. 4680 packs are only going into a % of Austin production. We don't know the %. All of Berlin production isn't predicated on 4680 development. Then add in the fact that Tesla has been consistently changing the battery types into which trim on both the 3 and Y over the past 6 months which alleviates bottlenecks on 2170 supply. I'm not saying there's no 2170 shortage or that at times pack supply hasn't been tight. But I have no idea where you're getting that graph and what numbers and factors are going into it.
A. That data is my best understanding of production data (not deliveries). It is production from the factory that is most relevant in understanding where production constraints are. If one looks at delivery data then that adds in a lot of other variables that not only confound the overall picture, but which are also different for each factory due to the different delivery chains, and which change with time. So production data is what I use. The data is shown in the graph, in the table above. The origin point is clearly shown in the table. If you have better data by all means show it, but that is the best data that I have (so far) been able to accumulate. (Often, but not always, I adopt Troy's data - and a thank you is owed to
@Troy for sharing so generously with the TMC community and being one of the few people to be systematic).
B. For the most recent Austin/Berlin quarter (i.e. Q1-2023) the production data is courtesy of Troy. Once again, thank you Troy. If you have better production data then please show it. I too am surprised by the implication that Berlin has fallen behind Austin during Q1, but I can think of a number of reasons why that might be (primarily different down times for New Year holidays; and being given different amounts of battery love; and the length of the delivery chain for battery supply to Berlin vs the overall shortness of the battery supply chain to Austin). But those are after-the-event rationalisations and may not be correct, and indeed the underlying data may be erroneous. However it is the best data that I have. If you have better data please post it. And no, burst data is not better data - it is overall quarterly production data that is required. As you yourself note the VIN data is somewhat ambiguous.
C. Au contraire. The Shanghai ramp rate is the best we have so far for an upper bound, and so should be respected. Yes of course there were particular factors, but nonetheless to suggest that it cannot be repeated is to suggest that Tesla will always be worse at volume manufacturing than once they were.
D. The 3/Y battery production graph is what it says on the tin - a quarterly CHANGE in battery production graph. Since the average pack size is assumed to be constant it might as well be a quarterly change in 3/Y production. What it tells us is how good Tesla is (or is not) at getting cells into the vehicle assembly lines as the lines go through ramps, both in absolute terms (which ought to grow over time) and in relative % terms (which ought to be somewhat constant). What you can see in the available data is that the growth rates in the last three quarters are not as great as one would have hoped for. Whether the issues are in LFP supply, or in 4680 supply, or something else I do not know. I suspect that the LFP is having to do a lot of heavy lifting (almost all of Berlin) to make up for a relatively weak ramp in 4680. But that is only my suspicion as there is not sufficient evidence in the data to be completely confident regarding that conclusion. (If you were to go to the Near future quarterly thread you will find enough data posted by me that you can yourself reproduce that graph - see
Near-future quarterly financial projections ). If you have better data please post it.
While that might be true, there is plenty that we don't know.
1) Is this unplanned and unexpected? Tesla guided for 1.8M to 2M vehicles when they already had cells for 1.8M
2) What role are the other buildings being constructed at Austin and elsewhere in Texas intended to play? Are those construction projects on time?
3) Where are the raw materials coming from? Have their been in delays in raw materials supply chains?
4) How is staff recruitment / training going?
5) Are they waiting for new equipment, including improved equipment?
6) Are there any technical / yield issues Tesla still needs to resolve?
The ramp may not be where we would like it to be, but we don't know if it is broadly in line with Tesla expectations, or if it is falling short of Tesla expectations. I'm saying Tesla expectations, not Elon expectations.
Even if it is falling short of expectations, we don't know why, or how long it will take to resolve the issues.
Tesla also intended to source 4680s from other suppliers, those other suppliers seem to be progressing at an even slower rate.
Absolutely. All I am doing is pointing out the logical implications of the best data we have so far (or at least the best data I've been able to pull together so far). It may of course be that the plan is to fix it all in the last quarter, who knows. But one thing we do know is that both Semi and Cybertruck consume a large number of cells/vehicle so to the extent that we are focussed on the vehicles/year metric (and I am not overly fixated on that, but many will be) then towards the very end of 2023, and definitely during 2024 and 2025, it will be even harder going to ramp the cell supply at the required rate.
Berlin is set up to use LFP, 2760, and 4680 cells, although there is no production of 4680's to speak of. I'm not 100% if they are doing both CATL prismatic LFP and BYD's blade LFP cells, but both were in the works. Their ramp has been anticipated for years, and while it is possible that the low 4680 yield has hampered things it should be offset significantly by the delay in production ramp.
Austin appears to be building 30-40% 4680 cells based on the castings outside, and 4680 volume is supposed to be in the 1,500-2,000/week equivalent volume. AFAIK for the 2760's they are using both LG and Panasonic, and I thought a third vendor was qualified.
...and then there is the Powerwall/Megapack. Those cells (capacity not physical) could easily be used for vehicle production and the storage products shifted to LFP if batteries were in fact an issue.
If the 4680's are not ramped up within the next 12 months then I can see a production barrier being hit in 2025 based on other OEM's ramp schedules.
Bottom line is there were a lot of committed production capacity for cells that the other OEMs have not been able to use to-date, which should increase available cells on the market meaningfully. That short-term benefit will erode over time, but it should not impact Tesla significantly if all goes according to plan.
Thank you, some good info there.
Some minor quibbles.
- I'm not so sure that Megapack LFP cells can be repurposed in any meaningful way to vehicles, as that would require raw material / early stage stuff to be processed in a different set of lines at the cell supplier. But I've not seen a compare-and-contrast of the LFP cells in the Megapack vs the ones going into the vehicles so as to understand to what extent there is commonality. Maybe it is a late-stage-customisation process in which case it might be an easy switch. My suspicion is otherwise and that the implication is that one would have very unbalanced lines and very unhappy supply chains and vendors. It would be good to know more.
- Powerwall had (has ?) yet (?) to switch to LFP and is in any case trivial volumes. (We know that from some recent Powerwall milestone data). A switch to LFP is supposed to be occurring and indeed it would be good if they could engineer commonality with vehicles at that moment, for these very reasons.
- Regarding other OEM capacity becoming available that of course would be great in some respects. (not from that of overall transition mission, but more so from a Tesla-specific TSLA shareholder perspective). However to what extent would Tesla be wiling to adopt other cells that were tailored for other OEMs in a timeframe that is relevant for the next 18-months ? As
@GhostSkater says the BYD and CATL LFP packs are vendor-built. To what extent is there spare capacity at BYD and CATL that is currently allocated to other OEMs but under utilised ? And is it product-specific or common with the Tesla product ? And what about the other (non CATL/BYD) suppliers, especially of LFP ?