2017Q1 results

[luvb2b]

Wow. I really need to meet you in person and hear this story.

lol. you don't need me, you just told the story. and kenliles just told it again. and i bet if you ever found capitalistoppressor or others they would tell it again.

those trades are not "the story." they are like bloody war stories veterans don't like to tell family for fear of the impact it may have - on themselves and others.

"yeah we were holed up together in a network of trenches, running out of munitions as an enemy squad was circling. the battle went on for days. the tide started to turn when we got word that their rearward supply lines had been compromised. we started repositioning our troops with what arms we had left. with coordinated sniper style attacks from different directions we started taking down enemy units. we were so low on ammo we shot each of them in head to save bullets. we murdered every one of those bastards and left their bodies to rot. those were the days!" (not!)​

profitable trades destroy many traders, like losses do, but in many ways they are much worse. recall the jesse livermore story. jesse went from 0, to millions, to bankruptcy... 3 times! he was divorced twice. his wife became an alcoholic, shot his son who was basically a partying drunk himself. he lost the $100 million he made shorting the crash of 1929. he spent a lot of his adult life in depression. he eventually committed suicide shooting himself in the head with a gun. his son became a full fledged alcoholic too and shot his dog and tried to kill a cop before committing suicide himself.

i had similar experiences around the tech bubble although not to that extreme before realizing where i had lost myself. i have seen the stories you have brought out of people @DaveT. they remind me of the story above, which is nothing more than a tragedy.

luv's 2c: the real story is how you live your life daily so you can have fun, enjoy life, keep learning, keep level-headed, and keep taking sensible risk. you need all the other things for the last one to happen. for if you do all that, you will surely find yourself again in a position of great advantage. and you will know how to come out of it and keep having a great life.

 

[30seconds]

Oh man, the web 1.0 bubble. I definitely was up a very big number before going to a very small number. Learned to earn some and walk away in the 2010 - 2012 run-up and then learned to both take some off the table and lever up with TSLA through 2013.

Currently stuck on strategy. Still have a good amount of core TSLA shares, but took most of my LEAPS (bought in 2015) off the table for losses last fall. My big gap is trying to figure out what type of market cap we will get to with successful Model 3 roll out. The market cap comps with other premium automakers (e.g., BMW) put a damper on my estimates since we aren't really seeing a massive roll out of TE.

 

[luvb2b]

made some slight adjustments to my revenue model due to merger accounting.

there is good reason to believe solarcity revenues will be reported lower post-merger vs. pre-merger, all else being equal. it comes down to deferred revenue declining, meaning less "stored" revenue is there to recognize in the future.

previously i felt solarcity revenue should be somewhat comparable to q4 16 because bookings in q3 16 were up another 15% from q2 16's high level. the proforma adjustments (discussed in the general forum) have worked out to about 25m per quarter. so i took my prior solarcity revenue figure down by 25 million. that includes effects of intracompany transactions.

i'm trying to understand if i can do any better than what i have here with questions in the general thread. thought i would put this out in the meantime. scroll back a couple pages to see reasoning behind other various numbers.

revenue (base - conservative model)
auto sales 1,923,550
auto leasing 233,675
1 time autopilot 100,000
reg credits 45,465
total auto 2,302,690
energy storage 66,199
solarcity 200,000
grohmann 34,000
services/other 200,357
total revenue 2,803,247

here's another model slightly less conservative than the above. the difference is lower model has 25% of vehicles leased vs. upper having 27% leased. lower includes slightly more lease revenue. those services revenue may be a bit high.

revenue (base - moderate model)
auto sales 1,976,250
auto leasing 239,274
1 time autopilot 100,000
reg credits 45,465
total auto 2,360,989
energy storage 70,928
solarcity 200,000
grohmann 34,000
services/other 200,357
total revenue 2,866,274

i'm fairly convinced revenue will come in well ahead of the 2.6b analyst consensus. for the most part the determining factors are going to be (1) mix of lease vs. cash/loan for automotive sales, and (2) recognition of zev credit and/or autopilot revenue. i feel like they already guided to #2 the last call when discussing they expect gross margin to match q3 levels gaap/non-gaap. and even with lease mix rising slightly the automotive revenue number should be near 2.3b. storage+grohmann+services+solarcity should be well over 300m in revenue.

to get revenues to under 2.6b requires:
1. take leased vehicles % of total up over 30% where it was in q3 2016.
2. no autopilot revenue
3. 15m in zev credits.
4. drop leasing revenue back to q3 16 levels
5. have solar city go down 30% sequentially.

do all this and you'll get to around 2.5b. i don't think you get this low, for the following reasons:

1. seems unlikely because the resale value guarantees are gone, because tax credits in the usa will decline in 1 year, and because i think they are trying to shift towards more loan and less lease.
2. autopilot revenue could be zero. it's possible.
3. low zev credit sales would surprise me. they sold 138m 2 quarters back and right now they should have a similar number of cars sold that qualify for credits. also they guided to higher gaap and non-gaap margins - back to q3 2016 levels. that would require some decent number of zev credits.
4. hard to see leasing revenue dropping with more cars being leased than there were 3 years ago (even though mix is dropping absolute number rises).
5. 30% drop quarter over quarter would be one of the biggest seen in solarcity's history.

 

[neroden]

Your revenue model seems decent to me, though I think no autopilot revenue in Q1. The tricky part is actually the cost model. We have a decent handle on cost-of-product for the automotive side, but the other costs seem to be a lot less clear.

I don't think analysts will pay any attention to the revenue (I think they'll either look at earnings or at cash flow) so I think the question of costs is key to determining whether there will be a positive reaction.

 

[Zhelko Dimic]

3. low zev credit sales would surprise me. they sold 138m 2 quarters back and right now they should have a similar number of cars sold that qualify for credits. also they guided to higher gaap and non-gaap margins - back to q3 2016 levels. that would require some decent number of zev credits.

Have you accounted for ZEV credits at 50% price or less? 50% is what they got last time, and I think they didn't recognize much of ZEV credits in the 3 preceding quarters. Demand was limited.
It's a buyers market, so are you sure they have buyers? Situation is becoming dire enough for OEMs, that I wouldn't be surprised they pay penalty at 100%, rather than give Tesla 50% of the money, that Tesla will use to slaughter them later.

 

[luvb2b]

to be honest i don't know a whole lot about zev credits' internals. i'm open to hearing another explanation or estimate of what to expect.

here's what i do know. the average dollar value of zev credits sold over the last 12 months has ranged between $1,800 and $2,800.
q4-16 q3 q2 q1 q4-15
2.8 2.8 1.8 2.1 2.2

the number jumps from 1.8 to 2.8 on the big credit sale in q3. the average of the last 4 quarters is around $2.36k credits per vehicle.

so my zev estimate is like this:

add last 4 quarters units sales: 25000 + 22200 + 24821 + 14402 = 86,423
86,423 x $2.36k per vehicle avg. = 203,958
total q2 16-q4 16 zev credits: 19840 + 138541 +51 = 158,432
that means you need 203,958 - 158,432 = approx 45,500 in zev credit sales this quarter.

fine, but here's the problem. they very clearly guided automotive non-gaap and gaap revenue back to q3 16 levels, when the difference between gaap and non-gaap gross margins (the zev credits) was 5% of revenue.

you can't get there with only 45m in zev sales. the number is too low, or the guidance was too high. i lean towards they must have a good handle on the guidance into the quarter, and especially with deliveries strong it's hard to believe margins come in squishy.

assuming 2.2b in auto revenue, somewhere you need around 100-120m in things added to auto revenue that get counted as non-gaap. that could be credits alone, or maybe a combination of deferred autopilot and zev credits.

 

[Rarity]

revenue (base - conservative model)
...
grohmann 34,000

I understand that Grohmann's business has been roughly 50% Tesla, so 50% of the 34,000 would be an intracompany transaction.

 

[luvb2b]

excellent, thank you will make that change.

I understand that Grohmann's business has been roughly 50% Tesla, so 50% of the 34,000 would be an intracompany transaction.

 

[brian45011]

to be honest i don't know a whole lot about zev credits' internals. i'm open to hearing another explanation or estimate of what to expect..

You might get a better correlation to ZEV credits earned (but not necessarily sold) by using California deliveries rather than world-wide deliveries:

http://www.cncda.org/CMS/Pubs/CA Auto Outlook 4Q 2016.pdf
http://www.cncda.org/CMS/Pubs/CA Auto Outlook 3Q 2016.pdf
http://www.cncda.org/CMS/Pubs/CA Auto Outlook 2Q 2016.pdf
http://www.cncda.org/CMS/Pubs/CA Auto Outlook 1Q 2016.pdf

At 8/31/16, Tesla had a balance of 3,530 ZEV credits in California. They earn them in about a dozen, other states, but the collective total pales compared to California. I think they still earn 4 ZEV credits per eligible delivery in jurisdictions that follow California's regime.

Zero Emission Vehicle Credits

I show for GHG/CAFE credits (aka "mouse nuts") sold ($MM):
1Q16 15.0
2Q16 20.1
3Q16 30.7
4Q16 21.0
In prior years when deliveries were lower, they averaged around $16 million/quarter

For mouse nuts, I think the best correlation might be to total US deliveries; there used to be a summary for NA and/or USA on Hobbes'/Troy's EU Market Outlook Wiki but I don't see it now; you may have to use InsideEVs' estimated tally.

FWIW, even though 1Q17 deliveries matched 3Q16, to the extent Tesla's management has optionality on when to report AP2 and regulatory credit revenues/profits, I'm not sure May is the optimal timing for another "pie" quarter.

Have fun!

 

[luvb2b]

i've made a few changes based on input, thinking about the ap2 revenues in the general thread, making changes so gross margin matches guidance, and shifting the lease vs. loan/sale mix back to q4 levels.

i'm also introducing my cost of revenue and gross margin estimates for further discussion.

item ... revenue ... cost of revenue ... gross margin%
auto sales 1,976,250 1,515,186 23.3%
auto leasing 239,274 157,535 34.2%
1 time autopilot 65,000 0 100.0%
reg credits 83,552 0 100.0%
total auto 2,364,077 1,672,721 29.2%
energy storage 70,928 74,474 -5.0%
solarcity 200,000 140,000 30.0%
grohmann 17,000 16,150 5.0%
services/other 200,357 206,368 -3.0%
total revenue 2,852,361 2,109,712 26.0%

some commentary -

the autopilot / reg credits need to be that high to drive auto only gaap gross margin out to 29%+ and still keep non-gaap auto gross margin at 24.5%. this could be off if autopilot is not considered non-gaap, but the general idea is 1 time auto revenue of 100+m is needed to reconcile margin guidance.

the solar city gross margin may be high at 30% because of adjustments to revenue / cost of sales. i've made adjustments assuming revenue will come down 25m vs. q4 to account for post-merger adjustments. i've also taken another 5% or so out of solarcitiy gross margin. this could end up being overly conservative as the proforma for back half of 16 implies post-merger accounting adds more profits (where as i am taking haircuts).

energy storage assumed to continue at negative margin with improvement over prior quarter's -29%

grohmann previously i assumed 10-15% margin but since they lose half their business but still have all the cost, cut that more than in half down to 5%

services run flat/negative, slight improvement over prior q but less than q3 16.

Your revenue model seems decent to me, though I think no autopilot revenue in Q1. The tricky part is actually the cost model. We have a decent handle on cost-of-product for the automotive side, but the other costs seem to be a lot less clear.

I don't think analysts will pay any attention to the revenue (I think they'll either look at earnings or at cash flow) so I think the question of costs is key to determining whether there will be a positive reaction.

 

[luvb2b]

where did you find the disclosure/information on mouse nuts (ghg/cafe etc)? they seem to be pretty clear only showing zev credits separately these days. maybe they're combining those other credits into auto revenue?

You might get a better correlation to ZEV credits earned (but not necessarily sold) by using California deliveries rather than world-wide deliveries:

http://www.cncda.org/CMS/Pubs/CA Auto Outlook 4Q 2016.pdf
http://www.cncda.org/CMS/Pubs/CA Auto Outlook 3Q 2016.pdf
http://www.cncda.org/CMS/Pubs/CA Auto Outlook 2Q 2016.pdf
http://www.cncda.org/CMS/Pubs/CA Auto Outlook 1Q 2016.pdf

At 8/31/16, Tesla had a balance of 3,530 ZEV credits in California. They earn them in about a dozen, other states, but the collective total pales compared to California. I think they still earn 4 ZEV credits per eligible delivery in jurisdictions that follow California's regime.

Zero Emission Vehicle Credits

I show for GHG/CAFE credits (aka "mouse nuts") sold ($MM):
1Q16 15.0
2Q16 20.1
3Q16 30.7
4Q16 21.0
In prior years when deliveries were lower, they averaged around $16 million/quarter

For mouse nuts, I think the best correlation might be to total US deliveries; there used to be a summary for NA and/or USA on Hobbes'/Troy's EU Market Outlook Wiki but I don't see it now; you may have to use InsideEVs' estimated tally.

FWIW, even though 1Q17 deliveries matched 3Q16, to the extent Tesla's management has optionality on when to report AP2 and regulatory credit revenues/profits, I'm not sure May is the optimal timing for another "pie" quarter.

Have fun!

 

[Darren12]

@luvb2b I dug into the EAP revenues and based on my initial research and calculations, I think they will recognize about $67.5 million in Q12017 (this is my base case). My bull case puts them at $90 million and the bear case at $45 million. I’ll break down my assumptions over the take-rate of the options first and then the percentage of revenue that they will recognize after.

Base-case assumptions:

Q42016 deliveries with AP2 hardware: 40% of deliveries or 9,000 vehicles (they delivered 22,252 in Q4). They started producing AP2 vehicles on October 19th All Tesla Cars Being Produced Now Have Full Self-Driving Hardware which is about 3 weeks into the quarter. I’m assuming all vehicles produced in the first 3 weeks were delivered during Q4 (2,200 per week times 3 weeks is 6,600 vehicles). They had a delivery overhang from Q3 of 5,500 vehicles. The total non-AP2 cars delivered in Q4 is about 12,000 vehicles. I’ll take another 1,000 off of this for non-AP2 inventory cars.

EAP and FSD take-rates: 80% EAP, 20% FSD. This seems reasonable based on the discussion in the general thread and how popular the autopilot features seem to be in the media. My analysis done below of the deferred revenues reported in the Q4 10-K supports take-rates that are close to this.

With these take-rates and 9,000 AP2 vehicles delivered in Q42016, I get $41.4 million in EAP and FSD revenues, most of which cannot be recognized.

Percentage of EAP revenue recognizable in Q42016: 15%. The first roll-out of AP2 went to 1,000 cars on December 31, 2016. It had very basic functionality, but something was delivered nonetheless. With 1,000 cars recognizing 15% of EAP, $750,000 was recognized in Q42016, leaving $40.65 million as deferred revenue. This isn’t significant, and the actual amount recognized could be 0%. But to be conservative, recognizing a portion in Q42016 means less revenues in Q12017.

This leaves total deferred AP2 revenues of $40.65 million at the end of 2016. Looking at the Summary of Significant Accounting Policies on the 10-K, Tesla discloses the following for deferred revenues: As of December 31, 2016, and 2015 we had deferred $291.2 million and $138.2 million related relating to the purchase of vehicle maintenance and service plans, access to our Supercharger network, internet connectivity, autopilot and over-the-air software updates. In the previous quarterly filings in 2016, Tesla does not mention autopilot in the deferred revenue disclosure. Therefore, the deferred autopilot revenue is 100% related to EAP and FSD. They also break out how much of the deferred revenue amount relates to each deliverable. I made a quick table summarizing this for Q42015 all the way to Q32016. From here, I looked at the QoQ changes in each category to estimate Q42016’s breakdown. I projected Q4 using the growth rates from Q2 to Q3 as it is the largest since they had record deliveries. Obviously they delivered less in Q4 than in Q3, but this gives a conservative estimate of how much of the deferred revenue can be attributed to autopilot. Using these conservative assumptions, I end up with $40.1 million in deferred autopilot revenue, which is quite close to my deferred revenue from the take-rate calculations.

Q12017 deliveries with AP2 hardware: 100% of deliveries or 25,000 vehicles. I doubt there would be many, if any, non-AP2 deliveries in Q12017.

I used the same EAP and FSD take-rates in Q12017. With 25,000 AP2 vehicles delivered, I get $115 million in EAP and FSD revenues, some of which can now be recognized. At the end of Q1, there are 34,000 AP2 vehicles (9,000 from Q4 plus 25,000 from Q1). Using the 80% take-rate out EAP, 27,200 of these vehicles have some EAP revenue that can be recognized in Q1. This post is getting fairly long, so I’ll make another post to discuss the percentage of revenue that can be recognized in Q1. My current assumption is 50% of the EAP revenue, which at $5,000 per car and 27,200 cars (less the $750,000 recognized in Q42016), puts autopilot revenues in Q12017 of $67.25 million. For the 50%, I’m looking at the EAP option description on the order page and comparing it to the features currently available, and also looking at some PwC software revenue recognition guidance: https://www.pwc.com/us/en/cfodirect...guides/pwc-revenue-recognition-march-2009.pdf


Summary:

Base-case assumptions: 40% of Q4 deliveries have AP2 hardware (or 10,000 vehicles). EAP and FSD take-rates of 80% and 20%. 50% of EAP revenue recognized in Q1. $67.25 million revenue recognized in Q1.

Bull-case assumptions: 45% of Q4 deliveries have AP2 hardware (or 10,000 vehicles). EAP and FSD take-rates of 85% and 25%. 60% of EAP revenue recognized in Q1. $88.5 million revenue recognized in Q1.

Bear-case assumptions: 35% of Q4 deliveries have AP2 hardware (or 8,000 vehicles). EAP and FSD take-rates of 70% and 10%. 40% of EAP revenue recognized in Q1. $43.7 million revenue recognized in Q1.

That's what I have so far but I'm still tweaking it. The base case of $67.25 million lines up quite nicely with your assumed $65 million. Thanks for sharing your model with us it's very helpful!

 

[Bgarret]

i've made a few changes based on input, thinking about the ap2 revenues in the general thread, making changes so gross margin matches guidance, and shifting the lease vs. loan/sale mix back to q4 levels.

i'm also introducing my cost of revenue and gross margin estimates for further discussion.

item ... revenue ... cost of revenue ... gross margin%
auto sales 1,976,250 1,515,186 23.3%
auto leasing 239,274 157,535 34.2%
1 time autopilot 65,000 0 100.0%
reg credits 83,552 0 100.0%
total auto 2,364,077 1,672,721 29.2%
energy storage 70,928 74,474 -5.0%
solarcity 200,000 140,000 30.0%
grohmann 17,000 16,150 5.0%
services/other 200,357 206,368 -3.0%
total revenue 2,852,361 2,109,712 26.0%

some commentary -

the autopilot / reg credits need to be that high to drive auto only gaap gross margin out to 29%+ and still keep non-gaap auto gross margin at 24.5%. this could be off if autopilot is not considered non-gaap, but the general idea is 1 time auto revenue of 100+m is needed to reconcile margin guidance.

the solar city gross margin may be high at 30% because of adjustments to revenue / cost of sales. i've made adjustments assuming revenue will come down 25m vs. q4 to account for post-merger adjustments. i've also taken another 5% or so out of solarcitiy gross margin. this could end up being overly conservative as the proforma for back half of 16 implies post-merger accounting adds more profits (where as i am taking haircuts).

energy storage assumed to continue at negative margin with improvement over prior quarter's -29%

grohmann previously i assumed 10-15% margin but since they lose half their business but still have all the cost, cut that more than in half down to 5%

services run flat/negative, slight improvement over prior q but less than q3 16.

@luvb2b wanted to ask on this thread in addition to market thread, are you including AP revenue for Q1 in your ASP? The 1 time autopilot is approx. $65 million for Q4, but Q1 should be about $95 million based on 25k deliveries. Just wanted to make sure that is included in your Q1 revenue.

 

[brian45011]

where did you find the disclosure/information on mouse nuts (ghg/cafe etc)? they seem to be pretty clear only showing zev credits separately these days. maybe they're combining those other credits into auto revenue?

You have to back into mouse nuts by going back and forth between the SH letters and the 10Qs/10ks, including prior period comparisons. ZEV credits are still regularly reported because they are part of the GAAP/non-GAAP reconciliation in the SH letters.

For instance, the 10k states: "In addition, there is an increase of $133.4 million to $302.3 million of the sale of regulatory credits from the year ended December 31, 2015 to the corresponding period in 2016. " For ZEV credits I show ($k):
1Q16 56,987
2Q16 .......64
3Q16 138,541
4Q16 19,840
Total 215,432

$302.3 million minus $215.4 million = $86.9 million in mouse nuts

 

[luvb2b]

@Darren12 - awesome work... thank you so much. have you discovered something else here?

seems some portion of the purchase price is amortized for supercharger access? i didn't realize this before.
since they discontinued free supercharging does that now mean what was once getting booked as deferred revenue now gets booked as sales instead? that's going to raise the average vehicle price in my model if so.

i like your estimate of deferred ap revenue a lot. i think from your work i agree ~$40m is the right number for q4.

i'll have to study that note about revenue recognition, as i don't know a lot about exactly how revenues are recognized for software features.

@Bgarret i thought i had blended the ap revenue into an asp increase for my q1 model. it's possible i didn't go far enough, but i ended up with asp similar to what it was in q3 16, when ap1.0 was being fully recognized. perhaps that's too low, as there is a 1% mix shift effect and perhaps another 1% on forex / price adjustments. i really have to think about it but perhaps asp's could rise 1-2% over q3 2016?

btw my calculation of average selling price (asp) is a bit different than others posted.
i take the disclosed mix of lease vs. sale (only disclosed last 2 quarters i think). for example q3 16 = 32%, q4 16 = 25%.
i take the auto sales revenue number and divide by ((1 - %leased) x total units delivered). that's auto sales only, excluding all the one time stuff and leasing revenue.
the denominator is the number of units sold on cash/loan (i.e. not leased).
that gets me revenue per vehicle which i assume is average selling price - maybe i should call it average revenue per vehicle instead.

using this i get 105.4k for q3 2016 and 103.3k for q4 2016.
i adjusted that q4 number up 1% for mix shift & 1% for foreign prices as mentioned. but perhaps i need to (a) either adjust another 2% or so for autopilot or (b) explicitly put the recognized autopilot from q1 17 on a separate line.

thanks everyone will think about it.

@luvb2b wanted to ask on this thread in addition to market thread, are you including AP revenue for Q1 in your ASP? The 1 time autopilot is approx. $65 million for Q4, but Q1 should be about $95 million based on 25k deliveries. Just wanted to make sure that is included in your Q1 revenue.

@luvb2b I dug into the EAP revenues and based on my initial research and calculations, I think they will recognize about $67.5 million in Q12017 (this is my base case). My bull case puts them at $90 million and the bear case at $45 million. I’ll break down my assumptions over the take-rate of the options first and then the percentage of revenue that they will recognize after.

Base-case assumptions:

Q42016 deliveries with AP2 hardware: 40% of deliveries or 9,000 vehicles (they delivered 22,252 in Q4). They started producing AP2 vehicles on October 19th All Tesla Cars Being Produced Now Have Full Self-Driving Hardware which is about 3 weeks into the quarter. I’m assuming all vehicles produced in the first 3 weeks were delivered during Q4 (2,200 per week times 3 weeks is 6,600 vehicles). They had a delivery overhang from Q3 of 5,500 vehicles. The total non-AP2 cars delivered in Q4 is about 12,000 vehicles. I’ll take another 1,000 off of this for non-AP2 inventory cars.

EAP and FSD take-rates: 80% EAP, 20% FSD. This seems reasonable based on the discussion in the general thread and how popular the autopilot features seem to be in the media. My analysis done below of the deferred revenues reported in the Q4 10-K supports take-rates that are close to this.

With these take-rates and 9,000 AP2 vehicles delivered in Q42016, I get $41.4 million in EAP and FSD revenues, most of which cannot be recognized.

Percentage of EAP revenue recognizable in Q42016: 15%. The first roll-out of AP2 went to 1,000 cars on December 31, 2016. It had very basic functionality, but something was delivered nonetheless. With 1,000 cars recognizing 15% of EAP, $750,000 was recognized in Q42016, leaving $40.65 million as deferred revenue. This isn’t significant, and the actual amount recognized could be 0%. But to be conservative, recognizing a portion in Q42016 means less revenues in Q12017.

This leaves total deferred AP2 revenues of $40.65 million at the end of 2016. Looking at the Summary of Significant Accounting Policies on the 10-K, Tesla discloses the following for deferred revenues: As of December 31, 2016, and 2015 we had deferred $291.2 million and $138.2 million related relating to the purchase of vehicle maintenance and service plans, access to our Supercharger network, internet connectivity, autopilot and over-the-air software updates. In the previous quarterly filings in 2016, Tesla does not mention autopilot in the deferred revenue disclosure. Therefore, the deferred autopilot revenue is 100% related to EAP and FSD. They also break out how much of the deferred revenue amount relates to each deliverable. I made a quick table summarizing this for Q42015 all the way to Q32016. From here, I looked at the QoQ changes in each category to estimate Q42016’s breakdown. I projected Q4 using the growth rates from Q2 to Q3 as it is the largest since they had record deliveries. Obviously they delivered less in Q4 than in Q3, but this gives a conservative estimate of how much of the deferred revenue can be attributed to autopilot. Using these conservative assumptions, I end up with $40.1 million in deferred autopilot revenue, which is quite close to my deferred revenue from the take-rate calculations.

Q12017 deliveries with AP2 hardware: 100% of deliveries or 25,000 vehicles. I doubt there would be many, if any, non-AP2 deliveries in Q12017.

I used the same EAP and FSD take-rates in Q12017. With 25,000 AP2 vehicles delivered, I get $115 million in EAP and FSD revenues, some of which can now be recognized. At the end of Q1, there are 34,000 AP2 vehicles (9,000 from Q4 plus 25,000 from Q1). Using the 80% take-rate out EAP, 27,200 of these vehicles have some EAP revenue that can be recognized in Q1. This post is getting fairly long, so I’ll make another post to discuss the percentage of revenue that can be recognized in Q1. My current assumption is 50% of the EAP revenue, which at $5,000 per car and 27,200 cars (less the $750,000 recognized in Q42016), puts autopilot revenues in Q12017 of $67.25 million. For the 50%, I’m looking at the EAP option description on the order page and comparing it to the features currently available, and also looking at some PwC software revenue recognition guidance: https://www.pwc.com/us/en/cfodirect...guides/pwc-revenue-recognition-march-2009.pdf


Summary:

Base-case assumptions: 40% of Q4 deliveries have AP2 hardware (or 10,000 vehicles). EAP and FSD take-rates of 80% and 20%. 50% of EAP revenue recognized in Q1. $67.25 million revenue recognized in Q1.

Bull-case assumptions: 45% of Q4 deliveries have AP2 hardware (or 10,000 vehicles). EAP and FSD take-rates of 85% and 25%. 60% of EAP revenue recognized in Q1. $88.5 million revenue recognized in Q1.

Bear-case assumptions: 35% of Q4 deliveries have AP2 hardware (or 8,000 vehicles). EAP and FSD take-rates of 70% and 10%. 40% of EAP revenue recognized in Q1. $43.7 million revenue recognized in Q1.

That's what I have so far but I'm still tweaking it. The base case of $67.25 million lines up quite nicely with your assumed $65 million. Thanks for sharing your model with us it's very helpful!

 

[luvb2b]

i had found quarterly regulatory credit disclosure in the past, but i could not find it this time around. where did you find the quarterly numbers? i searched for example q1 16 letter and 10q to no avail.

i had unknowingly included these ghg/cafe credits in to my auto revenues and therefore into my average revenue per vehicle calculation. as they seem small and flattish from your breakdown i may not worry about them because the average revenue per vehicle would approximate them anyway.

You have to back into mouse nuts by going back and forth between the SH letters and the 10Qs/10ks, including prior period comparisons. ZEV credits are still regularly reported because they are part of the GAAP/non-GAAP reconciliation in the SH letters.

For instance, the 10k states: "In addition, there is an increase of $133.4 million to $302.3 million of the sale of regulatory credits from the year ended December 31, 2015 to the corresponding period in 2016. " For ZEV credits I show ($k):
1Q16 56,987
2Q16 .......64
3Q16 138,541
4Q16 19,840
Total 215,432

$302.3 million minus $215.4 million = $86.9 million in mouse nuts

 

[brian45011]

@Darren12 seems some portion of the purchase price is amortized for supercharger access? i didn't realize this before. since they discontinued free supercharging does that now mean what was once getting booked as deferred revenue now gets booked as sales instead? that's going to raise the average vehicle price in my model if so .

On new deliveries, probably. But there are still "immaterial" operating expenses that the previously deferred revenue has to partially cover as the locations expand:
"As of December 31, 2016 and 2015, the net book value of our Supercharger network was $207.2 million and $166.6 million, respectively, and as of December 31, 2016, our Supercharger network included 790 locations globally. We plan to continue investing in our Supercharger network for the foreseeable future, including in North America, Europe and Asia, and expect such spending to continue to be a minimal portion of total capital spending. During 2017, we expect that this investment will grow our Supercharger network greatly. We allocate Supercharger related expenses to cost of total automotive revenues and selling, general, and administrative expenses. These costs were immaterial for all periods presented."

 

[Darren12]

@Darren12 - awesome work... thank you so much. have you discovered something else here?

seems some portion of the purchase price is amortized for supercharger access? i didn't realize this before.
since they discontinued free supercharging does that now mean what was once getting booked as deferred revenue now gets booked as sales instead? that's going to raise the average vehicle price in my model if so.

i like your estimate of deferred ap revenue a lot. i think from your work i agree ~$40m is the right number for q4.

i'll have to study that note about revenue recognition, as i don't know a lot about exactly how revenues are recognized for software features.

@Bgarret i thought i had blended the ap revenue into an asp increase for my q1 model. it's possible i didn't go far enough, but i ended up with asp similar to what it was in q3 16, when ap1.0 was being fully recognized. perhaps that's too low, as there is a 1% mix shift effect and perhaps another 1% on forex / price adjustments. i really have to think about it but perhaps asp's could rise 1-2% over q3 2016?

btw my calculation of average selling price (asp) is a bit different than others posted.
i take the disclosed mix of lease vs. sale (only disclosed last 2 quarters i think). for example q3 16 = 32%, q4 16 = 25%.
i take the auto sales revenue number and divide by ((1 - %leased) x total units delivered). that's auto sales only, excluding all the one time stuff and leasing revenue.
the denominator is the number of units sold on cash/loan (i.e. not leased).
that gets me revenue per vehicle which i assume is average selling price - maybe i should call it average revenue per vehicle instead.

using this i get 105.4k for q3 2016 and 103.3k for q4 2016.
i adjusted that q4 number up 1% for mix shift & 1% for foreign prices as mentioned. but perhaps i need to (a) either adjust another 2% or so for autopilot or (b) explicitly put the recognized autopilot from q1 17 on a separate line.

thanks everyone will think about it.

Looking at the supercharger access, I wouldn't make any big changes here are they discontinued the free supercharging for all vehicles ordered after January 15th. So with the overhang vehicles and all of the orders made before Jan 15 that ended up being delivered in Feb and Mar I think most of the Q1 deliveries will still have unlimited supercharging and have some deferred revenue here.

And for the software revenue recognition note it's pretty long but the relevant part is section 5 where they talk about specified upgrades. Basically if you sell a software product to a customer but part of the deal is that they get specific upgrades in the future, then these upgrades need to be allocated a defined portion of the revenue. You can then recognize the revenue when these upgrades are delivered. I think EAP would be treated this way since the features that are offered are specifically written on the order page.

The actual revenue recognition comes down to professional judgement here to decide how much of each feature has been delivered to the customers. Tesla will have some flexibility here since it's tough to quantify the exact amount that has been delivered of the EAP package.

 

[brian45011]

i had found quarterly regulatory credit disclosure in the past, but i could not find it this time around. where did you find the quarterly numbers? i searched for example q1 16 letter and 10q to no avail.

i had unknowingly included these ghg/cafe credits in to my auto revenues and therefore into my average revenue per vehicle calculation. as they seem small and flattish from your breakdown i may not worry about them because the average revenue per vehicle would approximate them anyway.

I'd have to go back and reconstruct how I backed into them. As I said, the process involved Qtr over Qtr and Yr over Yr comparisons since the data has been inconsistently reported. If you have it covered and may not worry about them, I won't bother with the reconstruction. The irony is mouse nuts are not an insignificant percentage of ZEV credits. I show:

2016 40.3%
2015 50.6%
2014 42.4%
2013 49.7%

Why try to obscure what he readily admits will not scale as deliveries ramp exponentially?

 

[Reciprocity]

I dont think they will recognize revs from AP2. They dont need it with strong deliveries from Q1 and Q2 delivers could be weaker with a smaller overhang and the ramp of M3 production. I think there are also some issues with the fact that they do not have full parity with AP1 and are not using 4 cameras yet.

Does anyone have an idea how much revs could come from S60 upgrades to S75? Or is it more mouse nutz?