Near-future quarterly financial projections

[EVNow]

Is that quarterly?
​

R&D has been pretty steady at around a billion a year, and will probably stay there. I think it may actually drop a bit from the recent $350 million/quarter as there is only so much R&D you can do at any time, and Models 3/Y/Semi/Pickup don't have high R&D requirements right now.

SG&A, however, is currently $2.2 billion/year, and I expect it to go up to $2.5 - $3 billion as the service centers are deployed geographically.

That would be a total of about $4 billion/year or $1/billion a quarter for R&D + SG&A. Is that the same as your estimate?​

Yes, all quarterly. Last 5 quarters R&D + SG&A has been around $1B/quarter. Probably longer - I've not checked.

Shouldn't most of the increase in service go to Service Cost ? That is definitely one item that has been increasing and I think will continue to increase. I don't know how they treat rent/lease on facilities for service center - whether that goes to SG&A or Service Cost or if its shared some kind of allocation.

R&D - Most of that cost is probably people. I expect Tesla to continue to hire more engineers for FSD and software - even as may be few are let go on the manufacturing side. Even so, generally Tesla wouldn't let a lot of manufacturing R&D engineers to go - may be only the ones they don't like - because it would be difficult to hire back good talent. They will probably get a few relaxed months after years of continuous hard work.

 

[Spacemanspliff]

Shouldn't most of the increase in service go to Service Cost ? That is definitely one item that has been increasing and I think will continue to increase. I don't know how they treat rent/lease on facilities for service center - whether that goes to SG&A or Service Cost or if its shared some kind of allocation.

Was going to mention this in the future when it came up but seems like a good time now. In the US alone, service centers have increased from low 78-82ish at the end of December/Jan to about 97 now. Seems like most of those 15 new service centers will be operational by the end of May and some running in a “Lite” capacity.

 

[neroden]

Yes, all quarterly. Last 5 quarters R&D + SG&A has been around $1B/quarter. Probably longer - I've not checked.

Shouldn't most of the increase in service go to Service Cost ?

No. Tesla is opening additional service centers to have geographic coverage. This is fixed cost overhead -- leasing land (or paying property taxes), paying utilities.

The added centers will actually CUT service costs (of the sort allocated to service / warranty on the balance sheet). People like me in areas remote from service centers, who have prepaid mobile service or whose cars have actually broken, have been driving service costs up substantially due to the extensive expense of transporting the cars back and forth. In order to retain *any* goodwill, Tesla has had to cover a lot of these costs even when they haven't officially promised to, because nobody wants to face a $2000 bill to get their car to and from the service center. With service centers a reasonable distance away, people will drive or use their own towing services and Tesla won't have to deal with hotel stays, flights, etc.

That is definitely one item that has been increasing and I think will continue to increase. I don't know how they treat rent/lease on facilities for service center - whether that goes to SG&A or Service Cost or if its shared some kind of allocation.

The "keeping the lights on" overhead should be SG&A if they're doing the accounting right. I have no idea whether they're doing the accounting right; GAAP has prescribed incorrect accounting in the past.

R&D - Most of that cost is probably people. I expect Tesla to continue to hire more engineers for FSD and software - even as may be few are let go on the manufacturing side. Even so, generally Tesla wouldn't let a lot of manufacturing R&D engineers to go - may be only the ones they don't like - because it would be difficult to hire back good talent. They will probably get a few relaxed months after years of continuous hard work.

While I don't expect Tesla to lay off that many R&D people, I don't think they need to hire more people for hardware R&D. For Autopilot, there's a "too many cooks" problem; I doubt they could get much benefit by spending more than, say, $20 million a year extra. For the rest of their software they definitely need more people but again I don't think it would amount to more than $40 million a year to get it right-sized, probably less. Meanwhile, the costs of experimental tooling and prototypes (which are totally R&D) should be dropping.

By contrast, they could easily raise their overhead by $500 million/year by opening a suitable number of service centers. At that point I'd guess they'd be built out.

 

[moneyburner1]

That is possible - but difficult to tell since there are many moving parts, esp. this quarter.

As I said, every $5k drop in ASP brings down the profit by about $15M.

If the ASP drops by

That is possible - but difficult to tell since there are many moving parts, esp. this quarter.

As I said, every $5k drop in ASP brings down the profit by about $15M.

If the ASP drops due to price reductions from $105k to $100k and Tesla sells 100k X/S Tesla will loose $500M, or?

 

[EVNow]

If the ASP drops due to price reductions from $105k to $100k and Tesla sells 100k X/S Tesla will loose $500M, or?

Assuming a margin of 20%, that 5k difference in ASP turns to 1k difference in margin. So, over a quarter (with 15k s+x), the difference is $15M. Over a year with 100k s+x it would be $100M.

 

[mongo]

Assuming a margin of 20%, that 5k difference in ASP turns to 1k difference in margin. So, over a quarter (with 15k s+x), the difference is $15M. Over a year with 100k s+x it would be $100M.

Fixed margin assumes they pulled out 4k of cost (via volume, part cost, or labor) along with the 5k of sales price. If they only drop the price, the margin % shifts also.
80k cost on 105k car = 24%
80k cost on 100k car = 20%
80k cost on 95k car = 16%

 

[EVNow]

Fixed margin assumes they pulled out 4k of cost (via volume, part cost, or labor) along with the 5k of sales price. If they only drop the price, the margin % shifts also.
80k cost on 105k car = 24%
80k cost on 100k car = 20%
80k cost on 95k car = 16%

Yes - I'm assuming they are keeping that 20% margin. Used to be 25% in Q3/Q4.

The ASP could drop because of discounting pre-Raven cars (in this quarter) or because the mix of trims has shifted downwards. I'd expect them to keep a minimum of 20% margin, though. Obviously if they can't keep margin of 20%, there will be more of a hit to p&l.

 

[Buckminster]

Here's a quick update to the Model 3 inventory levels, with Q1 data added:

  ========|============|============|============|===========|================|=============
  quarter | production | deliveries | Δinventory | inventory | transit        | fleet
  ========|============|============|============|===========|================|=============
  2017/Q2 |          0 |          0 |          0 |         0 |       0        |      0
  2017/Q3 |        260 |        220 |        +60 |        60 |      40        |     20
  2017/Q4 |      2,425 |      1,550 |       +875 |       935 |     860        |     75
  2018/Q1 |      9,766 |      8,180 |     +1,586 |     2,521 |   2,040        |    480
  2018/Q2 |     28,578 |     18,440 |    +10,138 |    12,659 |  11,166        |  1,493
  2018/Q3 |     53,239 |     55,840 |     -2,601 |    10,058 |   8,048        |  2,010
  2018/Q4 |     61,394 |     63,150 |     -1,756 |     8,302 |   1,010        |  7,292
  2019/Q1 |     62,950 |     50,900 |    +12,050 |    20,352 |   8,350 [est.] | 12,002 [est.]

Note that I marked two fields with an estimation disclaimer: Tesla didn't disclose the exact Model 3 in-transit numbers in Q1, so I used the S/X in-transit percentage of 18.6% from Q1'18 to estimate 2,250 in-transit S/X units and 8,350 in-transit Model 3 units. The error bar should be less than ±500 units, maybe more if my assumptions are wrong.

This table is mostly telling the story we know: in-transit Model 3 units primarily increased due to not all of them arriving at customers in China and Europe, the "fleet" inventory increased moderately by 4,700 units, this is a sum of the following categories of inventory cars:

• Unsold inventory of 1-2 weeks of batch production: they started this in Q4, i.e. no Fremont slowdown but near full production rate at the end of the quarter.
• There might have been some build-up of pre-Raven S/X units despite the production cut, as customers were waiting for the Raven refresh. Tells you why ICE OEMs are so paranoid about test mules, camouflaging them ridiculously and punishing design and media leaks with an iron fist. You can bet the media outlet that leaks the details of a new car model that the carmakers don't want to see leaked yet won't receive a single VW, Daimler and BMW advertising dollar for a long time - so the media probably isn't leaking them even if employees are leaking it. Tesla doesn't have that kind of leverage.
  
• U.S. showroom and test drive units which they probably depleted in Q4 but which they had to refill in Q1 due to the retail showroom turnaround and seasonal patterns.
• Increased showroom, test drive and service units in the rest of the world that is now selling Teslas.

I.e. I can see nothing unusual in the data: bad S/X numbers, average-to-good Model 3 numbers if we accept the cell supply limitation assumption.

 

[Doggydogworld]

These numbers are a little off because they use deliveries from the P&D report instead of the updated and slightly higher numbers they publish in the earnings letter. For example, Q4 2018 deliveries were 63,359 vs. 63,150. Actual inventory at end of Q1 was 19,890.

 

[Fact Checking]

New These numbers are a little off because they use deliveries from the P&D report instead of the updated and slightly higher numbers they publish in the earnings letter. For example, Q4 2018 deliveries were 63,359 vs. 63,150. Actual inventory at end of Q1 was 19,890.

Good point - I'll try to update the table with the earnings report numbers, because AFAICS this small error I've introduced accumulates with every quarter and makes the real inventory seem several hundred units higher than it really is, correct?

 

[Fact Checking]

New These numbers are a little off because they use deliveries from the P&D report instead of the updated and slightly higher numbers they publish in the earnings letter. For example, Q4 2018 deliveries were 63,359 vs. 63,150. Actual inventory at end of Q1 was 19,890.

I updated my table with the incomplete info that I was able to find in the 10-Q's and 10-K's, and arrived at Model 3 inventory of 19,996 units (in-transit plus fleet), and I'm wondering how you arrived at the 19,890 figure:

  =========|============|============|============|===========|================|=============
  quarter  | production | deliveries | Δinventory | inventory | transit        | fleet
  =========|============|============|============|===========|================|=============
  2017/Q2  |          0 |          0 |          0 |         0 |       0        |      0
  2017/Q3  |        260 |        220 |        +40 |        40 |      40        |      0
  2017/Q4  |      2,425 |      1,550 |       +875 |       915 |     860        |     55
  2017/10K |            |         -6 |         +6 |       921 |     866        |     55
  2018/Q1  |      9,766 |      8,180 |     +1,586 |     2,507 |   2,040        |    467
  2018/Q2  |     28,578 |     18,440 |    +10,138 |    12,645 |  11,166        |  1,479
  2018/Q3  |     53,239 |     55,840 |     -2,601 |    10,044 |   8,048        |  1,996
  2018/Q4  |     61,394 |     63,150 |     -1,756 |     8,288 |   1,010        |  7,278
  2018/10K |            |       +312 |       -312 |     7,976 |     698        |  7,278
  2019/Q1  |     62,950 |     50,932 |    +12,018 |    19,996 |   8,350 [est.] | 11,644 [est.]

Notes, methodology:

• Quarterly Model 3 sales were not reliably reported, for example in Q3'18 they only reported a figure of "over 56,000". End of year sales were more reliably reported in the 10-K's, and I noted those correction factors as an extra '10K' line for every completed year.
• For these end of year correction lines I assumed that any changes in deliveries increase/reduce the in-transit figure accordingly, so "fleet" remained unchanged.
  
• Some of the data was obscured, such as: "Automotive sales revenue increased $9.10 billion, or 107%, in the year ended December 31, 2018 as compared to the year ended December 31, 2017, primarily due to an increase of approximately 144,330 Model 3 deliveries from our significant production ramp in the year ended December 31, 2018, delivered at average selling prices that remained relatively consistent year-over-year."
• I.e. the real 2018 sales figure isn't 144,330, but 144,330 plus the 1,764 in 2017. This obfuscation is annoying.
  
• For Q1'19 they did report 50,932 deliveries in the 10-Q, which is slightly up from the 50,900 in the delivery report - I updated the Q1 figure accordingly.
• I assumed that the production figures were final in the P&D report.
• I estimated Q1'19 Model 3 in-transit level, which wasn't reported directly, from the reported total in-transit figure of 10,600, combining it with the Q1'18 ~18.6% in-transit percentage of the Model S/X, scaled to the lower Q1'19 delivery volume which gives 2,250 units S/X in-transit, and 8,350 units Model 3 in-transit. This estimate only impacts the 'fleet' column and the error should disappear next quarter.
  
• I did not add scrapped units because I found no information about them. I.e. customer returning a lemon and it being declared a total loss and written off and scrapped should reduce production by one. I assume that this is a minor factor though.
  
• Rules:
  • production = the quarterly Model 3 production figures from the P&D reports
  • deliveries = the delivery figures from the P&D reports, corrected with the 10-K at year end
    
  • Δinventory = production-deliveries
  • inventory = Δinventory summed up forward
  • transit = the in-transit figures from the P&D reports
  • fleet = inventory-transit

My inventory figure is now 106 units higher than yours so I guess it's close enough, but still, I either calculated it incorrectly somewhere or missed some data somewhere.

 

[EVNow]

My inventory figure is now 106 units higher than yours so I guess it's close enough, but still, I either calculated it incorrectly somewhere or missed some data somewhere.

I've Q3 deliveries as 56065. You have it as 55840. Other differences are smaller.

 

[Doggydogworld]

I updated my table with the incomplete info that I was able to find in the 10-Q's and 10-K's, and arrived at Model 3 inventory of 19,996 units (in-transit plus fleet), and I'm wondering how you arrived at the 19,890 figure:

  =========|============|============|============|===========|================|=============
  quarter  | production | deliveries | Δinventory | inventory | transit        | fleet
  =========|============|============|============|===========|================|=============
  2017/Q2  |          0 |          0 |          0 |         0 |       0        |      0
  2017/Q3  |        260 |        220 |        +40 |        40 |      40        |      0
  2017/Q4  |      2,425 |      1,550 |       +875 |       915 |     860        |     55
  2017/10K |            |         -6 |         +6 |       921 |     866        |     55
  2018/Q1  |      9,766 |      8,180 |     +1,586 |     2,507 |   2,040        |    467
  2018/Q2  |     28,578 |     18,440 |    +10,138 |    12,645 |  11,166        |  1,479
  2018/Q3  |     53,239 |     55,840 |     -2,601 |    10,044 |   8,048        |  1,996
  2018/Q4  |     61,394 |     63,150 |     -1,756 |     8,288 |   1,010        |  7,278
  2018/10K |            |       +312 |       -312 |     7,976 |     698        |  7,278
  2019/Q1  |     62,950 |     50,932 |    +12,018 |    19,996 |   8,350 [est.] | 11,644 [est.]

Notes, methodology:

• Quarterly Model 3 sales were not reliably reported, for example in Q3'18 they only reported a figure of "over 56,000". End of year sales were more reliably reported in the 10-K's, and I noted those correction factors as an extra '10K' line for every completed year.
• For these end of year correction lines I assumed that any changes in deliveries increase/reduce the in-transit figure accordingly, so "fleet" remained unchanged.
  
• Some of the data was obscured, such as: "Automotive sales revenue increased $9.10 billion, or 107%, in the year ended December 31, 2018 as compared to the year ended December 31, 2017, primarily due to an increase of approximately 144,330 Model 3 deliveries from our significant production ramp in the year ended December 31, 2018, delivered at average selling prices that remained relatively consistent year-over-year."
• I.e. the real 2018 sales figure isn't 144,330, but 144,330 plus the 1,764 in 2017. This obfuscation is annoying.
  
• For Q1'19 they did report 50,932 deliveries in the 10-Q, which is slightly up from the 50,900 in the delivery report - I updated the Q1 figure accordingly.
• I assumed that the production figures were final in the P&D report.
• I estimated Q1'19 Model 3 in-transit level, which wasn't reported directly, from the reported total in-transit figure of 10,600, combining it with the Q1'18 ~18.6% in-transit percentage of the Model S/X, scaled to the lower Q1'19 delivery volume which gives 2,250 units S/X in-transit, and 8,350 units Model 3 in-transit. This estimate only impacts the 'fleet' column and the error should disappear next quarter.
  
• I did not add scrapped units because I found no information about them. I.e. customer returning a lemon and it being declared a total loss and written off and scrapped should reduce production by one. I assume that this is a minor factor though.
  
• Rules:
  • production = the quarterly Model 3 production figures from the P&D reports
  • deliveries = the delivery figures from the P&D reports, corrected with the 10-K at year end
    
  • Δinventory = production-deliveries
  • inventory = Δinventory summed up forward
  • transit = the in-transit figures from the P&D reports
  • fleet = inventory-transit

My inventory figure is now 106 units higher than yours so I guess it's close enough, but still, I either calculated it incorrectly somewhere or missed some data somewhere.

I use deliveries from the earnings letters - as you note they don't consistently report deliveries in the Form 10s. Here's what I have for production, deliveries and total inventory:
Q3 2017 - 260 222 38
Q4 2017 - 2,425 1,542 921
Q1 2018 - 9,766 8,182 2,505
Q2 2018 - 28,578 18,449 12,634

Q3 2018 - 53,239 56,065 9,808
Q4 2018 - 61,394 63,359 7,843
Q1 2019 - 62,975 50,928 19,890

I don't know why the "increase of approximately 144,330" language in the 10-K leads to 146,094 2018 deliveries vs. 146,055 from the earnings letters. It's a 0.03% difference. I'll stick with the earnings letters where they report consistently, to the last digit and without the word 'approximately'. Also because I'm lazy ​

 

[EVNow]

I use deliveries from the earnings letters - as you note they don't consistently report deliveries in the Form 10s. Here's what I have for production, deliveries and total inventory:
Q3 2017 - 260 222 38
Q4 2017 - 2,425 1,542 921
Q1 2018 - 9,766 8,182 2,505
Q2 2018 - 28,578 18,449 12,634

Q3 2018 - 53,239 56,065 9,808
Q4 2018 - 61,394 63,359 7,843
Q1 2019 - 62,975 50,928 19,890

I don't know why the "increase of approximately 144,330" language in the 10-K leads to 146,094 2018 deliveries vs. 146,055 from the earnings letters. It's a 0.03% difference. I'll stick with the earnings letters where they report consistently, to the last digit and without the word 'approximately'. Also because I'm lazy ​

Yes, this is what I use as well. BTW, using S & X info since early days that Troy has, inventory number for S+X is 13,997. Probably some of those have been totaled/lost and written off.

 

[Doggydogworld]

Yes, this is what I use as well. BTW, using S & X info since early days that Troy has, inventory number for S+X is 13,997. Probably some of those have been totaled/lost and written off.

Yeah, I'm a couple percent lower on S/X at 13,735, but I never verified my source for the early year numbers. I always wondered where all those cars are. Some bigger sales/service centers probably have dozens of them, but I figure most locations are like my local store/SC that usually has about 10.

 

[Spacemanspliff]

Nobody has a good estimate of deliveries. Tesla quarters are too back-loaded. My guesses for Model 3 were:

35k - US (10/10/15 April/May/June)
20k - Europe (Q2 has a head start over Q1, but ramp is slower)
10k - China (total guess)​

That's 65k, plus maybe 15k S/X. They seem to have a problem with Raven S/X, but they have plenty of pre-Raven inventory so 15k is still a low bar. With some heroics they could be 70k+ Model 3 and 20k S/X, which would meet their guidance.

It wasn't clear to me why you added 30k shipped cars to your 14k/month of delivered production and 10k+ of Q1 carryover.

So 13,950 model 3's estimated by Insideevs for May, both months did not include Canada and there seems to be strong demand there at least to conservatively say 5k worth over 2 months.

S- 1025
X-1375

Which as Insideevs noted is below May 2018 numbers of S- 1520 and X-1450.

Some points to consider most of these S/X sales were likely Pre-Raven, I'd guess 25% Raven at most given that US deliveries only started happening late in May and many are targeted for June.

Seems like most production for Raven cars was aligned for overseas in May so that gives us about 5 carriers that could be loaded with Raven cars that will for sure be delivered in Q2 abroad.

 

[Spacemanspliff]

S/X specifically per InsideEVs:

Jan-Feb 2019 = 1,675 X
Apr-May 2018 = 2,475 X
Apr-May 2019 = 2,425 X = +45% over the first two months of Q1 2019, on par with first two months of Q2 2018.

Jan-Feb 2019 = 1,350 S
Apr-May 2018 = 2,770 X
Apr-May 2019 = 1,850 S = +37% over the first two months of Q1 2019, -33% vs first two months of Q2 2018.

So, the X has recovered this quarter to approximately match last year's Q2-so-far numbers, despite the Ravens not yet being ramped up. The S is still significantly down vs last year, but has recovered a bunch compared to Q1 of this year.

Good post from @Zaxxon regarding X/S trends

 

[Spacemanspliff]

18-20K S&X seems reasonable at this point.

We will be able to gauge European S/X demand off those country registration tabs prior to end of quarter.

 

[Spacemanspliff]

April: 10.05
May: 13.95
Overseas: 20
Transit Q1: 8
June: 15

Puts us at 67k deliveries + conservatively 2.5k from Canada is 69.5

Possible surprises:
-Canada numbers, BC has been burning through the 10million incentive (I think twice) which is roughly 2000 cars so maybe +1500 on original estimate?
-More cars per carrier per Kirkhorn
-More US deliveries than 15k (possible w/ expiring tax credit)
-Carrier #12 to shanghai delivers any percentage of their load which is possible.
-Negatives carriers have less than originally thought, they don’t hit June USA #’s.

 

[Zhelko Dimic]

So 13,950 model 3's estimated by Insideevs for May, both months did not include Canada and there seems to be strong demand there at least to conservatively say 5k worth over 2 months.

...

Conservatively?
Canada is 10% of US car market for regular cars, and less than that for luxury cars.
So I find 5K for two months highly unlikely, eqv. to maybe 70K units in US in 2 months.
New federal incentive helps, but it's still about the same as US incentive.