Near-future quarterly financial projections

[mongo]

How do you reconcile your 1.8B CapEx in 2018, vs Tesla's 2.35B CapEx forecast from their Q1 letter?

CapEx isn't something Tesla can just "turn off." The massive majority of that expenditure isn't optional. Heavy machinery is purchased 12-18 months in advance before it becomes operational and the expenditure recorded. Similar with construction jobs - you can't just bail on the construction company halfway through without incurring a significant penalty.

So how do you see Tesla reducing 2018 CapEx by 550M?

"At this stage, we are expecting total 2018 capex to be slightly below $3 billion".
1.8 Billion plus 655k (2.455 Billion) is also less that 3 Billion.
(for certain values of slightly, this would be 18% below 3 Billion).

The previous post's spend of 2.4 Billion would be over the 3 Billion mark for the year.

 

[luvb2b]

i asked @Reality to review this post offline (he is on a temporary ban i guess). there was some good discussion which i am sharing below.

@Reality (i added bold to what i felt was the most relevant part):

Your math looks right. Altho the statement "GM will improve more than i thought" is correct for Q2 but as with everything there is always a downside, as it will hurt Q3.

I think the "gross margin impact %" number is a bit irrelevant, I mean the real number is shaving 10k off each car in depreciation, which is enormous.

I agree with your implicit assumption of 110m in dep spread over 18k cars, the only caveat is that it's just one variable and we dont know what the other inputs are, but for this variable i agree that shaving 10k off each car in depreciation is enormous.

The only thing I would say might be off, is given how much tooling there is, and they tripled production, just a 15m increase in depreciation might be low. it implies that you think only 7.5m of the 130m in Q1 was tooling and that definitely does not sound correct to me. That's unit based so should scale directly obv.​

me:

yes yes yes... this is the input i was looking for. you are correct it will require some modifications/thought.​

very grateful for @Reality's input.

i went back to review the tooling data. ballpark 500m of tooling has been added since they started scaling model 3 production. supposedly these tools have a ~1m unit life, so it's $500 per vehicle of tooling depreciation, regardless of production level.

a ~20k increase in unit production would automatically increase depreciation by $10m. i assumed another 10m of depreciation was added as they added various types of manufacturing equipment during the quarter.

from the 18q1 comment which approximates to ~130m quarterly depreciation at 5k/weekly, it implies $1,500 in production-related manufacturing equipment depreciation. so at 65k units it would be around $100m in fixed production related depreciation based on the 18q1 comment. for 18q2 i assume fixed production related depreciation increases by 10m/quarter, so $110m of that plus ~$15m in tooling depreciation to reach $125m.

so perhaps this is a better evolution of production related depreciation?

17q3: 80m
17q4: 95m
18q1: 105m (100m in fixed depreciation + 5m in tooling depreciation)
18q2: 125m

now what's interesting, if i use a higher or lower number for 18q2 (say within a range of 115m-135m), the math is roughly ~9k in depreciation gets clipped which is around a 16% gross margin improvement from 18q1 to 18q2.

ok. i think i agree with your interpretation. i checked in with my accounting expert too and he gave me this link, which describes what you described.
https://www.aipb.org/pdf/DEPRECIA.pdf

so.... going with your interpretation, i think it means margins are going to improve more than i expected. i haven't yet posted an update because i wanted to make sure i understood how to evolve model 3 gross margin to this quarter.

let me present the following view for discussion purposes. the idea here is based on the depreciation insights provided above and some assumptions of the level of production related depreciation. i think we should see a 19% improvement in gross margin just from depreciation math and not including any other efficiency benefits.

presented for discussion purposes and very interested to hear comments related to the methodology, assumptions, and conclusion. note that the blended depreciation calculation is using units in-transit from the prior quarter and assigning them the depreciation per vehicle from the prior quarter's production. that means some vehicles sold get the current depreciation "per model 3" and some get the prior quarter's "per model 3" line. the gross margin impact is based on my asp assumptions, dollar wise you can see it's around 10.7k per vehicle improvement from 18q1 to 18q2.

depreciation assumptions based on this comment from abuja on last conference call (excerpted from seeking alpha)

... if we look at our depreciation costs on a per unit basis at steady run rate of 5,000 or so cars per week, we are in my mind well below most of our competitors – well below $2,000 per unit depreciation cost.​

so 5,000 model 3's x 13 weeks x $2,000 = $130,000,000 in production related depreciation per quarter (or perhaps less).

…

3 deliveries

3 production

3's in-transit

est model 3 production-related depreciation

per model 3

blended production related depreciation per 3

gross margin impact

[TD2] luv q4-18e [/TD2][TD2] luv q3-18e [/TD2][TD2] luv q2-18e [/TD2][TD2] Mar-18 [/TD2][TD2] Dec-17 [/TD2] [TD2] 60,000 [/TD2][TD2] 50,000 [/TD2][TD2] 18,440 [/TD2][TD2] 8,182 [/TD2][TD2] 1,542 [/TD2] [TD2]n/a[/TD2][TD2]n/a[/TD2][TD2]28,578[/TD2][TD2]9,766[/TD2][TD2]2,425[/TD2] [TD2]n/a[/TD2][TD2]n/a[/TD2][TD2]11,166[/TD2][TD2]2,040[/TD2][TD2]860[/TD2] [TD2]n/a[/TD2][TD2]n/a[/TD2][TD2]145,000[/TD2][TD2]130,000[/TD2][TD2]80,000[/TD2] [TD2] n/a [/TD2][TD2] n/a [/TD2][TD2] 5.07 [/TD2][TD2] 13.31 [/TD2][TD2] 45.36 [/TD2] [TD2] n/a [/TD2][TD2] n/a [/TD2][TD2] 5.99 [/TD2][TD2] 16.68 [/TD2][TD2] 45.36 [/TD2] [TD2] n/a [/TD2][TD2] n/a [/TD2][TD2] -10.7% [/TD2][TD2] -29.8% [/TD2][TD2] -85.6% [/TD2]

 

[ValueAnalyst]

How do you reconcile your 1.8B CapEx in 2018, vs Tesla's 2.35B CapEx forecast from their Q1 letter?

CapEx isn't something Tesla can just "turn off." The massive majority of that expenditure isn't optional. Heavy machinery is purchased 12-18 months in advance before it becomes operational and the expenditure recorded. Similar with construction jobs - you can't just bail on the construction company halfway through without incurring a significant penalty.

So how do you see Tesla reducing 2018 CapEx by 550M?

Tesla always guides high and ambitious on growth and CapEx, and as growth targets come down, investment requirements also do. Tesla's *maintenance* CapEx is only a few hundred million a quarter, and roughly $1.5B per year as of 3Q18. The rest can easily be "turned off," as Tesla is doing today.

 

[FirebirdAlpha]

Tesla always guides high and ambitious on growth and CapEx, and as growth targets come down, investment requirements also do.

Source? I've never seen Tesla's annual CapEx forecasts to be more than 15% off.

Can you point to a year where they guided more than 15% high?

You're saying they're guiding 33% high from Q1 and 42% high from Q4?

 

[ValueAnalyst]

Source? I've never seen Tesla's annual CapEx forecasts to be more than 15% off.

Can you point to a year where they guided more than 15% high?

You're saying they're guiding 33% high from Q1 and 42% high from Q4?

Sorry, I can't do your homework for you. Good luck with SBOTC.

 

[Reality]

Tesla always guides high and ambitious on growth and CapEx, and as growth targets come down, investment requirements also do. Tesla's *maintenance* CapEx is only a few hundred million a quarter, and roughly $1.5B per year as of 3Q18. The rest can easily be "turned off," as Tesla is doing today.

I dont believe that the rest can be 'turned off' as easily as you are stating, but i dont have enough information to challenge the 600m assumptions so probably will just leave it be.

Where are you getting the quarterly maintenance number from, EC or filing?

 

[ValueAnalyst]

I dont believe that the rest can be 'turned off' as easily as you are stating, but i dont have enough information to challenge the 600m assumptions so probably will just leave it be.

Where are you getting the quarterly maintenance number from, EC or filing?

I'll let the people who wanted you back answer your questions/FUD.

 

[Reality]

I'll let the people who wanted you back answer your questions/FUD.

fwiw, im not sure anyone said they "wanted" me back

 

[Reality]

@luvb2b

I have a question about your Model 3 ASP for Q3 (too early to talk Q3?)


You have 54k deliveries, so im going to use that number (seems reasonable with the 11k in transit)

and an ASP of 60k


Are you somewhat winging it here or do you model in expected take rate of options and P3D mix?

 

[NannerAirCraft]

My predictions for Q2-Q4. Some of these numbers are wild guesses.

 

[luvb2b]

i am only somewhat winging it to get a 18q3 60k asp.

my modeling estimates a 56k model 3 asp for 18q1. error bars are probably +/- 1.5k there.

from there you have two new powertrain options: awd is an extra 4k, and performance an extra 15k.

every 10% take rate on awd increases asp by around 0.4k, all else equal.
every 10% take rate on performance increases asp by around 1.5k, all else equal.

so take your mix. and note performance adds things like white interior as well.

i probably will bring my delivery estimate back down to 50k, the last delivery report didn't convince me 5k / weekly is sustainable (yet).

@luvb2b

I have a question about your Model 3 ASP for Q3 (too early to talk Q3?)


You have 54k deliveries, so im going to use that number (seems reasonable with the 11k in transit)

and an ASP of 60k


Are you somewhat winging it here or do you model in expected take rate of options and P3D mix?

 

[luvb2b]

thanks for sharing that. seems their efforts to cut opex haven't convinced you?

View attachment 316146 View attachment 316147 View attachment 316148 My predictions for Q2-Q4. Some of these numbers are wild guesses.

 

[NannerAirCraft]

I would assume the people let off were still being paid for almost 2 months of the quarter right? And will be getting severance so I don't know if there will be much difference in Q2. Also aren't they still hiring hundreds of new people for the Model 3 production lines and with all this new revenue I would assume they would be building more stores and hiring more people and that the layoff was just cutting out the ineffective employees. I'm erring on the side of caution with my opex numbers but if they do succeed in lowering those costs or even just keeping them steady then profitability will be much easier to attain in the second half.

 

[luvb2b]

yeah i get the skepticism.

i remember once ahuja guided some number of opex for the next quarter.

by the time the quarter ended like, oh, 8 weeks later, the number was $150m higher.

I would assume the people let off were still being paid for almost 2 months of the quarter right? And will be getting severance so I don't know if there will be much difference in Q2. Also aren't they still hiring hundreds of new people for the Model 3 production lines and with all this new revenue I would assume they would be building more stores and hiring more people and that the layoff was just cutting out the ineffective employees. I'm erring on the side of caution with my opex numbers but if they do succeed in lowering those costs or even just keeping them steady then profitability will be much easier to attain in the second half.

 

[Reality]

This isnt a huge number and it is only 1 time, but how do you calculate the 81m one time cost? Are you including more than the layoff in there? Seems high to me.


Reports I saw said Tesla laid off 4k employees.


If we assume each employee made 100k a year (they targeted higher earners but a higher avg non-stock comp than this doesnt feel super realistic, but maybe it is given cali salaries) means you are assuming about 10.5 weeks severance per employee. If I remember right, reports were 2 weeks plus 1 week per year up to 10 weeks max (i think, dont quote that though, cant find it online now).

 

[luvb2b]

they also accelerated some vesting of shares is my understanding.
i may be a bit high on the 1 time cost estimate though, which is fine. i'm sure i've overestimated some other item

This isnt a huge number and it is only 1 time, but how do you calculate the 81m one time cost? Are you including more than the layoff in there? Seems high to me.


Reports I saw said Tesla laid off 4k employees.


If we assume each employee made 100k a year (they targeted higher earners but a higher avg non-stock comp than this doesnt feel super realistic, but maybe it is given cali salaries) means you are assuming about 10.5 weeks severance per employee. If I remember right, reports were 2 weeks plus 1 week per year up to 10 weeks max (i think, dont quote that though, cant find it online now).

 

[brian45011]

i may be a bit high on the 1 time cost estimate though, which is fine. i'm sure i've overestimated some other item

Another potential one-time hit on Q2 OpEx in R&D could be:

The in-process research and development (“IPR&D”), [$86.8MM] which we acquired from SolarCity Corporation (“SolarCity”), is accounted for as an indefinite-lived asset until the completion or abandonment of the associated research and development efforts. If the research and development efforts are successfully completed and commercial feasibility is reached, the IPR&D would be amortized over its then estimated useful life. If the research and development efforts are not completed or are abandoned, the IPR&D might be impaired. The fair value of the IPR&D was estimated using the replacement cost method under the cost approach, based on the historical acquisition costs and expenses of the technology adjusted for estimated developer’s profit, opportunity cost and obsolescence factor. We expect to complete the research and development efforts in the second quarter of 2018, but there can be no assurance that the commercial feasibility will be achieved. The nature of the research and development efforts consists principally of planning, designing and testing the technology for viability in manufacturing solar cells and modules. If commercial feasibility is not achieved, we would likely look to other alternative technologies.
​

This probably relates to SCTY's acquisition of Silevo in 2014. There are rumors one of the "alternative technologies" could be Panasonic's for use at Riverbend. Regardless, the 10Q in about three weeks should provide more information useful in projecting Solar Generation volumes and margins.

they also accelerated some vesting of shares is my understanding.

Could RIF-ed workers liquidating the vested options have contributed to the early July down draft in share price?

 

[schonelucht]

Given the volume of trading I think it was probably a drop in the bucket.

 

[brian45011]

Given the volume of trading I think it was probably a drop in the bucket.

Is it Chanos then?

 

[luvb2b]

hmm interesting, so this adds the possibility of 1-time $80m or so charge for silevo. i don't think that technology was used meaningfully, but not sure there.

i don't think it was employee share liquidation that drove the price down. you have a ton of options outstanding on tesla, and this tends to exaggerate all up and down moves. you also had the huge buildup for the "short burn of the century" and of course what we got was the long burn of the century instead. a nicer way to say that would be there was some "buy the rumor & sell the news."

Another potential one-time hit on Q2 OpEx in R&D could be:

The in-process research and development (“IPR&D”), [$86.8MM] which we acquired from SolarCity Corporation (“SolarCity”), is accounted for as an indefinite-lived asset until the completion or abandonment of the associated research and development efforts. If the research and development efforts are successfully completed and commercial feasibility is reached, the IPR&D would be amortized over its then estimated useful life. If the research and development efforts are not completed or are abandoned, the IPR&D might be impaired. The fair value of the IPR&D was estimated using the replacement cost method under the cost approach, based on the historical acquisition costs and expenses of the technology adjusted for estimated developer’s profit, opportunity cost and obsolescence factor. We expect to complete the research and development efforts in the second quarter of 2018, but there can be no assurance that the commercial feasibility will be achieved. The nature of the research and development efforts consists principally of planning, designing and testing the technology for viability in manufacturing solar cells and modules. If commercial feasibility is not achieved, we would likely look to other alternative technologies.
​

This probably relates to SCTY's acquisition of Silevo in 2014. There are rumors one of the "alternative technologies" could be Panasonic's for use at Riverbend. Regardless, the 10Q in about three weeks should provide more information useful in projecting Solar Generation volumes and margins.



Could RIF-ed workers liquidating the vested options have contributed to the early July down draft in share price?