200,000th US Delivery

[Waiting4M3]

Not quite your meaning, but the dial can be turned: from the previous max rate to the new max rate after the slowest step/cell is improved.
You would not want to speed up entire line at once since that might make a downstream bottle neck operate at too fast a rate once the pinch point is resolved. So each cell is individually sped up.The line will clear out downstream of the bottleneck, so it is easy to find one problem spot or they can measure the cycle time of each cell to find the slowest cells and address them in parallel.

Example:
Slowest step allows 200/day
Engineers make a change which should allow a faster rate.
Production rate at that cell is slowly ramped up to validate the change. Meanwhile, things stack up at a new point upstream or downstream.
New bottleneck is found
Repeat.

The step change of one cell improvement is not necessarily the step change of the line.
If 3 steps have rates of 200, 225, and 250 cars per day, and you update the slowest cell to do 300 per day, you only increase the line rate to 225. Then you update the 2nd cell to 300/day, line goes to 250, finally last cell is updated to 350/day, line changes to 300. With each line rate change being driven by the speed at which they increase the cell rate.

So you can have large step changes at specific point, but effectively, a slow ramp on the whole (unless one spot is wayyy slower, then watch out Bloomberg!).

Poor Bloomberg, I have a feeling they're going to be TMC's M3 ramp whipping boy for a long time to come.

 

[jeewee3000]

That's getting kinda close, it depends how you are comparing and if you consider this or that "enough" and set the metric there. SR is a great example of this, is that "enough" for the particulars of your situation? Is that the metric to use? Tesla also does make it pretty fuzzy to compare directly in other ways, because of the way they group options. A lot less groupings, compared to more a la carte you'll see in most other manufacturers. Comparisons shift a lot if you define "enough" as having a number of tech options, which Tesla has a LOT of in base but then AP is a very sizable step. (FSB is....well there's no comparison and not just because there's not yet an a functioning FSB to compare with ).

But it isn't yet clear "this is price parity", you have to squint still to call it that. Maybe there never will be a clear case of it because of how out-of-the-box Teslas are packaged, and inherently are. Which probably isn't entirely accidental/coincidental. "You can't beat them if you just try to copy them" is a pretty good rule-of-thumb for 'industry disruptors'.

I more-so expect that one day we'll wake up, Tesla Generation 4 will ship, and it'll just be so damn clear that ICE vehicles cost more to build and buy.

I disagree you have to squint to call it price parity.

Just configured a standard BMW 3-series with an engine that has the same performance specs as the Long Range Model 3, and it's 51.390,00 EUR. [yes, I'm located in Europe but the comparison holds up]

Model 3 is 49.000,00 EUR with premium options package.

Model 3 is VERY competitive with BMW 3+ series right now. The whole discussion of ICE vs BEV (range anxiety) could imho be the only reason not yet to rule out the 3-series, but this too shall pass.

 

[schonelucht]

Meanwhile, things stack up at a new point upstream or downstream.

Again no. You can't just arbitrarily stack things up. Staging buffers must be designed. To hold production for 5 minutes or so. Maybe. Not for whatever timeframe is necessary to solve the next bottleneck. We are talking about potentially thousands of car frames coming through per day. You can't hold that kind of volume in a buffer. If you need to slow down, everything slows down.

 

[ValueAnalyst]

Again. I need to rant on this.

Production doesn't work the way you saw in the Charlin Chaplin movie. It's not like they turn a dial and up goes the rate. Elon even spelled this out many times. The large movements on the production rate are STEP CHANGE. That means the rate is plus or minus the same (more on that later) untill engineers have fully solved the blocking bottleneck at which time the rate steps up. Then it's a continuous rate again etc. When Tesla said they did 2000 in the trailing 7 days and expect to do 2000 in the next week that's essentially the same. See? Nearly no production movement rate in 2 weeks.

Now I promised to say more on plus or minus. Every production shift has issues that operators and line chefs have to solve. Some of these issues are recurring. A tool may blunt. Some gripper may fade, Some transport belt tears or jams. Whatever. As production learns to understand what can and does go wrong on a line, it is quicker in solving those issues. Meaning that instead of taking 20 minutes to find out that it's a camera that's gone dirty and they need to hit it with a spot of compressed air to solve the alignment fault, it takes them 20 seconds. Those are small production rate changes, the difference between 95% and 98% running time. It's slow efficiency creep in a way. But that kind of process takes months and months (a LOT can go wrong on complicated production lines).

There is no 'S' curve that is smooth and nice. It's an abstraction. The reality is a set of step changes. The fact that a step change came in between 23'rd and now to bring us from 200 to 300 says nothing about getting another step change from 300 to 400 this week. In fact, the report from Tesla explicitly told us that they DO NO EXPECT SUCH A STEP CHANGE TO HAPPEN THIS WEEK.

Speculation that does not even match up with official communication by Tesla, why do it????? On seeking Alpha maybe, there are plenty of people there who claim that Tesla lies in official investor communications. But here???????

I appreciate the difficulty in ramping production, as you well explained it in your rant and as Elon also has many times. Having said that, I don't know if Tesla said anything in the P&D release that would contradict 4,000+ weekly rate by the end of April or early May (by the earnings release maybe?), which would align with this morning's large NHTSA batch. I'm not saying that this will happen, but it's not impossible either. Let's see the pace of invites and VIN assignments in the next few weeks. I know one thing: I was not expecting this latest registration for another few weeks.

 

[Waiting4M3]

I appreciate the difficulty in ramping production, as you well explained it in your rant and as Elon also has many times. Having said that, I don't know if Tesla said anything in the P&D release that would contradict 4,000+ weekly rate by the end of April or early May (by the earnings release maybe?), which would align with this morning's large NHTSA batch. I'm not saying that this will happen, but it's not impossible either. Let's see the pace of invites and VIN assignments in the next few weeks. I know one thing: I was not expecting this latest registration for another few weeks.

TBH the new batch of ~4800 VINs today was a bit of a surprise to me as well on 1st thought. Here is some more slow digestion of the data:

The previous registered VIN range ends at 20581. We already see VINs close to 16000 being assigned, probably mean they are close to coming out of the production line, if not already out. At 2k/wk, and WAG 2wk WIP time for a car to go through the line, it's possible that a lot of the VINs from 15886 to 20581 (~4.5k) have entered production. With that in mind, it's not a big surprise that Tesla registered more VIN today. The main surprise, to me, and I suspect to most, is the size of the batch, 4.8K. I think 4k/wk by Q1ER is too optimistic. A very rough WAG based on the jump from 200/day-300/day in ~1-2 weeks, I think it's possible that ramp could go to 2.5k/wk by mid Apr, and 3k/wk by end of Apr.

I think we will have a couple of ways to check this in a couple of weeks:

1) VIN assignment, we'll see how quickly the 16k-20k VINs come out
2) new VIN registration, at current rate of 2k/wk this 4.8k VIN should last 2.5 weeks, any new registration sooner than that could point towards new step up happening

In the mean time I'm trying to keep calm, keep an open mind and just see what data comes. It's not easy as the production numbers are starting to get eye-wateringly large.

 

[mongo]

Again no. You can't just arbitrarily stack things up. Staging buffers must be designed. To hold production for 5 minutes or so. Maybe. Not for whatever timeframe is necessary to solve the next bottleneck. We are talking about potentially thousands of car frames coming through per day. You can't hold that kind of volume in a buffer. If you need to slow down, everything slows down.

Darn you English language and your multiple definitions for the same word/ meanings for a phrase (and my inability to use a clearer one)!
Stack up as in log jamb: things clear out downstream of slowest operation, and things eventually stop upstream and only move when slowest operation cycles. Everything upstream of slowest op moves at rate of slowest op, things downstream of slowest op develop extra of inter part space.

As opposed to: things keep running and buffer forever upstream.

Edit: missed a letter...

 

[ℬête Noire]

I disagree you have to squint to call it price parity.

Just configured a standard BMW 3-series with an engine that has the same performance specs as the Long Range Model 3, and it's 51.390,00 EUR. [yes, I'm located in Europe but the comparison holds up]

Model 3 is 49.000,00 EUR with premium options package.

Model 3 is VERY competitive with BMW 3+ series right now. The whole discussion of ICE vs BEV (range anxiety) could imho be the only reason not yet to rule out the 3-series, but this too shall pass.

That's exactly what I was talking about, it depends where you decide to set the mark. The options list matters, and if you let one or the other dictate the grounds upon which those happen then that's the way things tilt.

If you let Tesla dictate all the tech package stuff, then it tilts towards parity. Because they effectively MUST include all the underlying stuff anyway, making those other things on top far cheaper to include. It's the nature of the drivetrain.

Similarly, if you let the BMW option list dictate that you need a 12 paint colours to choose from or something, then Tesla is out in the weeds because you're talking custom colour jobs in the mix that's into thousands of dollars. Because Tesla doesn't do that at all, it's not even really an "option" outside of 3rd party.

 

[Waiting4M3]

That's exactly what I was talking about, it depends where you decide to set the mark. The options list matters, and if you let one or the other dictate the grounds upon which those happen then that's the way things tilt.

If you let Tesla dictate all the tech package stuff, then it tilts towards parity. Because they effectively MUST include all the underlying stuff anyway, making those other things on top far cheaper to include. It's the nature of the drivetrain.

Similarly, if you let the BMW option list dictate that you need a 12 paint colours to choose from or something, then Tesla is out in the weeds because you're talking custom colour jobs in the mix that's into thousands of dollars. Because Tesla doesn't do that at all, it's not even really an "option" outside of 3rd party.

But that's true for all automakers offerings, everyone offers different packages/option/colors, engine/transmission combination, etc. My impression is that when someone shop for a 3-series, C-class or A4, they're looking to spend $40-60K, which is right around where Tesla Model 3 is aimed at, and what Tesla offer is competitive against what people get from BMW/MB/Audi for that price range. If you have any data suggesting lots of people buying stripped down version of 3-series, I'd like to see it.

 

[ℬête Noire]

But that's true for all automakers offerings, everyone offers different packages/option/colors, engine/transmission combination, etc.

Not even close to the extent it is for the Model 3, for reasons explained.

My impression is that when someone shop for a 3-series, C-class or A4, they're looking to spend $40-60K, which is right around where Tesla Model 3 is aimed at, and what Tesla offer is competitive against what people get from BMW/MB/Audi for that price range. If you have any data suggesting lots of people buying stripped down version of 3-series, I'd like to see it.

That's begging the question.

"It's price parity because you're looking to spend the same amount of money." Well then the Model 3 is price parity with a 90HP 4x4 tractor and a mid-range ski boat, too.

 

[Reciprocity]

Unless you are the type who believes that Tesla is lying in investor communications, the current real output is 2000/week, well below capacity of 2500 for the semi-automated line. Stop speculating about additional Grohman lines without basis, it helps no one.

If your going to be snippy, at least quote my whole post:

I like to look at this over a bit longer period of time. Its around 13,000 VINs in just under a month and is throwing out the first number. That is well above the 2,500/w pace and closer to 3,000. I dont know why they would need so many VINs going back to March 10th and again, throwing out the number registered on March 10. It would be different if they just registered large chunks because you could assume it was for some future production, but the consistent drum beat tells me a different story. It could be that the Grohmann machine is finally in place and actually contributing to the 2500+ number for packs. It could also mean that the rest of the line can scale up quicker then I had assumed. At least to a point. The important thing for us here is how to we estimate what the real output is?

I agree that you should believe what Tesla puts out officially. But they could also be holding some cards closer to the vest. Its not like they are forth coming with details when they dont have to be. I agree 100% that 2020 was what they did and 2000/w wast he pace for this week. It does not mean that is what they anticipate is going to be the throughput next week.

To use your logic, I think we should believe Tesla when they Register 15,000 VINs in 25 days. Its meaningful. More meaningful then 4700 this week, or 4700 last. It shows that they have a need for a lot more VINs then what they say they are able to produce this week. Either they are randomly registering thousands of VINs they dont need or they need them.

 

[Waiting4M3]

Not even close to the extent it is for the Model 3, for reasons explained.

That's begging the question.

"It's price parity because you're looking to spend the same amount of money." Well then the Model 3 is price parity with a 90HP 4x4 tractor and a mid-range ski boat, too.

You ignored my premise that the discussion is aimed at "someone shop for a 3-series, C-class or A4"

 

[SBenson]

Again. I need to rant on this.

Production doesn't work the way you saw in the Charlin Chaplin movie. It's not like they turn a dial and up goes the rate. Elon even spelled this out many times. The large movements on the production rate are STEP CHANGE. That means the rate is plus or minus the same (more on that later) untill engineers have fully solved the blocking bottleneck at which time the rate steps up. Then it's a continuous rate again etc. When Tesla said they did 2000 in the trailing 7 days and expect to do 2000 in the next week that's essentially the same. See? Nearly no production movement rate in 2 weeks.

Now I promised to say more on plus or minus. Every production shift has issues that operators and line chefs have to solve. Some of these issues are recurring. A tool may blunt. Some gripper may fade, Some transport belt tears or jams. Whatever. As production learns to understand what can and does go wrong on a line, it is quicker in solving those issues. Meaning that instead of taking 20 minutes to find out that it's a camera that's gone dirty and they need to hit it with a spot of compressed air to solve the alignment fault, it takes them 20 seconds. Those are small production rate changes, the difference between 95% and 98% running time. It's slow efficiency creep in a way. But that kind of process takes months and months (a LOT can go wrong on complicated production lines).

There is no 'S' curve that is smooth and nice. It's an abstraction. The reality is a set of step changes. The fact that a step change came in between 23'rd and now to bring us from 200 to 300 says nothing about getting another step change from 300 to 400 this week. In fact, the report from Tesla explicitly told us that they DO NO EXPECT SUCH A STEP CHANGE TO HAPPEN THIS WEEK.

Speculation that does not even match up with official communication by Tesla, why do it????? On seeking Alpha maybe, there are plenty of people there who claim that Tesla lies in official investor communications. But here???????

Very insightful post. Thank you very much!

Have a question for you.

Adam Jonas still believes that 5K/wk may not be attained until tail end of the year. I see that he is an auto analyst. I suspect that he would have toured and learnt plenty about auto making plants over his career. So to me it feels like that is an informed opinion.

But I am looking for your opinion of his opinion

What do you think of his projection? Sure, he may very well turn out to be wrong. But is it a reasonable prediction? Do you feel that it is a blind WAG or is he basing it on some amount of insight?

 

[ggr]

Very insightful post. Thank you very much!

Have a question for you.

Adam Jonas still believes that 5K/wk may not be attained until tail end of the year. I see that he is an auto analyst. I suspect that he would have toured and learnt plenty about auto making plants over his career. So to me it feels like that is an informed opinion.

But I am looking for your opinion of his opinion

What do you think of his projection? Sure, he may very well turn out to be wrong. But is it a reasonable prediction? Do you feel that it is a blind WAG or is he basing it on some amount of insight?

Adam Jonas tends to swing wildly, sometimes on the money, sometimes well below it, but on average often below the subsequent reality. Andrea James, on the other hand, while being often dismissed as wildly optimistic, was by far the most accurate. We miss you, Andrea.

 

[ℬête Noire]

You ignored my premise that the discussion is aimed at "someone shop for a 3-series, C-class or A4"

I didn't ignore that. What I did do was point out how your argument, your premise, was begging the question. I tossed in a little lighthearted reductio ad absurdum to help you see your way free of that. I guess you're on your own to figure it out now.

Take care.

 

[elasalle]

My notes:

Assumptions:
19K deliveries to hit 200K in US

2Q production:
24K S/X - same as 1Q 18
26K M3 - assuming 1Q exit rate, shutdown, etc.

so while production might be bit higher on M3, due to final 2-3 weeks, lets assume 24 + 26 = 50K max production (S3X)

How to stay under 19K?

Usual thumb rule is that 60% of S/X is US sales, that would be 14K, if they want to meet 200K, this number will have to be lower, so say 40%, 9.6K

US - 9.6K S/X ~ 8K (adjust)
- 10K M3 (same as Q1)

- total 18K

Rest of world
- S/X usually 40%, increase slightly higher 50%
- up from usual 9.6K to 12K

M3.
Canada - 6K (wild ass assumption)

total = 18K

So 18K + 18K = 36 K accounted for out of 50K max.

Remaining
S/X = 24-8(US)-12(World) = 4K
M3 = 26 - 10(US)-6(Canada) = 10K

Notes.
.If AWD comes along Canada intake might be higher
.Germany numbers for 60K cars might be higher this quarter now that rebate issue has been resolved
. Norway S/X deliveries looking good, and this might be due to transport issues, or working as planned
.M3 delivered to some more countries
.Keep rest in inventory
.China Tariffs might throw a curveball

Questions:
How will market react if say Tesla says 50K produced, only 36K delivered. Market should know why this happened, to meet 200K ?

thoughts?

 

[mongo]

Questions:
How will market react if say Tesla says 50K produced, only 36K delivered. Market should know why this happened, to meet 200K ?

thoughts?

If they hit 5k/ wk on 3 by the end of Q2, that puts them at 3.5x the previous S/X only build rate (7k vs 2k). It would follow that the in transit number could also scale by the same factor. Q2 2017 had 3,500 in transit, so 12,250 would be a straight extrapolation. That compares nicely with your 14k overhang (50k - 36k, (but does this number need to increase to include the 6,100 S/X/3 Q1 overhang?)).

 

[elasalle]

If they hit 5k/ wk on 3 by the end of Q2, that puts them at 3.5x the previous S/X only build rate (7k vs 2k). It would follow that the in transit number could also scale by the same factor. Q2 2017 had 3,500 in transit, so 12,250 would be a straight extrapolation. That compares nicely with your 14k overhang (50k - 36k, (but does this number need to increase to include the 6,100 S/X/3 Q1 overhang?)).

we can just negate vehicles in transit each quarter, as they are definitely not delivered right? Simplifies it. But yes 2Q should have slightly higher vehicles in transit (or inventory)

the 5K/week at end of quarter, can be factored out by assuming that there will be some scheduled down time this quarter too. I just calculated for 13 weeks, but it could be 12 weeks.

Also what has worked so far is to take the exit rate of prior quarter as rate for the current quarter, and I think that approach will work for this quarter as well.

 

[mongo]

we can just negate vehicles in transit each quarter, as they are definitely not delivered right? Simplifies it. But yes 2Q should have slightly higher vehicles in transit (or inventory)

the 5K/week at end of quarter, can be factored out by assuming that there will be some scheduled down time this quarter too. I just calculated for 13 weeks, but it could be 12 weeks.

Also what has worked so far is to take the exit rate of prior quarter as rate for the current quarter, and I think that approach will work for this quarter as well.

In terms of market reaction, the number it would likely look at is the deviation of in-transit vehicles (inventory overhead). If Q2 needs to hold back 14k cars from the 50k production along with the 6,100 from the Q1 overhang, that results in 20k vehicles in-transit. That is 6k higher than the straight extrapolation based on production rate increase. However, these are all destined to buyers so there is no fiscal risk other than revenue being in Q3 vs Q2, so I don't think market will care much.

The FUDdy-duddys may try to say demand has disappeared, at which point we all chuckle and move on with our lives..

 

[elasalle]

In terms of market reaction, the number it would likely look at is the deviation of in-transit vehicles (inventory overhead). If Q2 needs to hold back 14k cars from the 50k production along with the 6,100 from the Q1 overhang, that results in 20k vehicles in-transit. That is 6k higher than the straight extrapolation based on production rate increase. However, these are all destined to buyers so there is no fiscal risk other than revenue being in Q3 vs Q2, so I don't think market will care much.

The FUDdy-duddys may try to say demand has disappeared, at which point we all chuckle and move on with our lives..

Also, the inventory held back in 2Q, would very likely guarantee +ve earnings in 3Q ..

 

[RobStark]

36k Q2 deliveries would be +20% QoQ.

I don't think The Street will care very much.