Near-future quarterly financial projections

[The Accountant]

I love your work. I think you might be around the correct ballpark. But I also feel that there is chance that Austin and Berlin will do a lot more in Q4 than 24+30k compared to shanghai 255k. I would not be suprised if Tesla aims at getting to say 50+50k in Q4, heck even 100k+100k feels reasonable in comparison to Shanghai 255k. They have had bottle necks, but once these clears it is not guaranteeed that the next bottle neck will be in the same ballpark, it might a lot higher. And by Q4 they should be ready with 24/7 shifts right? It’s not like Berlin and Austin are small plants and they use even more gigapress and other stuff.

So what are the current bottle necks? Batteries, chip and some random other parts such as seats/harness etc?! Batteries we have plenty of new capacity coming online and some backlog from Q2. Chips they have tried to solve for so long and may just have over-orded and get around Q3-Q4. Everything else? Maybe, maybe not?!

I was super aggressive on my quarterly estimates for Austin and Berlin when I first started calculating 2022 numbers only to have to take the forecast down each time I updated my forecast. I think it has clouded my judgment here. I know that Tesla can "turn it on" when all the pieces come together.
As Q3 progresses, let's see if we've got evidence to up the forecast with confidence.

 

[transpondster]

Someone already corrected that you are missing VAT, but I'll add one more fact: there is no cheaper model Y's in eastern Europe. You are most likely referring to countries that Tesla does not sell directly. In those countries, the tesla.com price does not include VAT and buyer has to pick up the car from central Europe and take care of VAT payment manually. So it might look like Estonia or Poland has cheaper Model Y than Finland according to Tesla.com, but that's not the case.

Tesla.com price include VAT/sales tax:

NA: No
Europe: Yes
Eastern Europe: No

Poland has Tesla store and price is with VAT. For some reason in Q3 Poland and France had one of the cheapest 3 SR+ in EU and I bought in December there, I still waiting for VAT refund..., so currently my SR+ is quite price with double VAT .

New cars are special VAT items in EU for private persons, if someone buys in another country he/she have to pay full price with with local VAT and pay VAT in country when registering and get VAT refund from purchase place.

 

[unk45]

When they started leasing in April 2019, it was all US from what I am reading in articles written at that time.
It looks like leases began in Europe in Sept 2019.

Not exactly. They've supported leasing in Europe since 2011:
tesla-motors-and-athlon-car-lease-announce-electric-vehicle-leasing-program-eur
Until 2021 there probably were no measurable Tesla direct risks, but they've had third party Tesla-authorized leasing all along. I am unsure what type of subvention, recourse or lease termination agreements may have been included in those leases.

Surely those would probably not have been material. In the UK manufacturer buyback at lease termination agreement do exist, but usually they're structured to protect the lessor. The only way this could be material if Tesla did have the buyback, in which case there could be significant gains.

I just asked by UK auto lease contact prior to posting this. He said his Tesla leases since he was involved in 2017 all had "residual value guarantees". I find not Tesla disclosures in 2018 and 2019 on that issue so they were immaterial or my contact misstated the fact. I do not know which.

Sorry for my lack of clarity.

 

[mongo]

Can someone explain this a little more. Surely for some vehicles Tesla will own them outright at the end of the lease and have already charged the COGS. If they sell to a new customer for $40k isn't it all profit minus cost of sales? - ie. ~$38k.

They are not paid off at end of lease (that would require loan type terms). So Tesla would then need to record a loss since total lease payments are less than CoGS. So they get tracked as a depreciating asset. At lease end, they still carry unrealized value/ basis which offsets sale price.

$47,000 for car of which $33,000 is CoGS. Three year lease $4,500 down 520/ month = $23k. Still $10k net negative, and $24k under a direct sale.
For example (not sure if this is how it works): 5 year straight depreciation is $28k over 3 years, so Tesla would take a $5k loss on the expense sheet during the lease, and need to use the basis of $19k when disposing.

 

[mongo]

Not exactly. They've supported leasing in Europe since 2011:
tesla-motors-and-athlon-car-lease-announce-electric-vehicle-leasing-program-eur
Until 2021 there probably were no measurable Tesla direct risks, but they've had third party Tesla-authorized leasing all along. I am unsure what type of subvention, recourse or lease termination agreements may have been included in those leases.

Surely those would probably not have been material. In the UK manufacturer buyback at lease termination agreement do exist, but usually they're structured to protect the lessor. The only way this could be material if Tesla did have the buyback, in which case there could be significant gains.

I just asked by UK auto lease contact prior to posting this. He said his Tesla leases since he was involved in 2017 all had "residual value guarantees". I find not Tesla disclosures in 2018 and 2019 on that issue so they were immaterial or my contact misstated the fact. I do not know which.

Sorry for my lack of clarity.

@The Accountant is referring specifically to the Model 3. The US never had lease end purchase options on those. Like you said, since UK (and another region) did, those % would need to be backed out.

 

[Gigapress]

At the Giga Berlin Grunheide County Fair event in October, Elon stated at 21:25 in the video:

“We are aiming to have the [Berlin 4680] cell factory to reach volume production by the end of next year [2022] as well as to reach volume production in general for this whole factory by the end of next year. So, volume production would be in excess of 5,000 cars per week and hopefully closer to 10,000.”​

30,000 units for Q4 from Berlin would be an average of 2,300 per week, well below the target for minimum run rate by the end of the quarter. Supposedly they’re already at 1,000 per week now.

Elon’s low end guidance is roughly for 5k * 13 weeks = 65k in Q4, assuming they start the quarter at less than 5k per week and end higher than that as projected, for an average of 5k per week across the quarter. From what he said, even something like 100k is hopefully targeted for Q4 on the optimistic side of the range.

So there is very significant upside from @The Accountant ‘s forecast for Berlin, and presumably this applies to Austin as well because it has almost identical industrial design and manufacturing processes as far as we’ve been able to see from the factory tours.

 

[Usain]

NoA is running on the old software topology.
AoCS is using the new design. They already support two stacks (plus old HW support)
Single stack is when NoA is transitioned to the new software design which would not occur until after AoCS has been highly vetted and shown to be better than current NoA (likely implying wide release).

Are you thinking of unified vector space which they are working on moving everything in AoCS to? That would likely be done before wide release.

AoCS goes to unified vector space (UVS)
AoCS wide release, FSD fulfilled
AoCS validated
NoA moved to UVS, FSD now single stack
L4+ FSD

However, in October last year, Elon has Tweeted that V11 was going to be single stack, so it may be that full roll out will not occur until after single is intergrated. Or, Tesla may do a full access 10.x version.

This may be moving into something more applicable to the Enigeering thread.

More this thread: is the impact of end of lease FSD option revenue recognition known/ factored in? (Guessing low # and minimal data)

Yeah, this is getting too far in the weeds, so this will be my last post on the subject.

I believe unified vector space and single stack are requirements for further recognition of FSD revenue because I don't think FSD is going to work well enough without both.

And I'm going by what Elon said about V11, as you quoted above. I expect UVS in V11 as well, but maybe that's wishful thinking.

I believe V11 is coming soon (Q3 or Q4). But it will need to be in beta awhile before wide release.

That pushes further FSD recognition into 2023.

 

[heltok]

I believe unified vector space and single stack are requirements for further recognition of FSD revenue because I don't think FSD is going to work well enough without both.

I think FSD recognition is a function of mostly:
1. Percent of users in the program
2. Functionality(how many percent of drives no intervention, hands off or not, robotaxi etc)

Single stack and unified vector space are useful for
1. Saving compute
2. Removing some errors with different vector spaces
3. Making smother transitions between highway and city driving
4. Maybe increasing performance of highway by using city driving knowledge

At some point improvement in functionality will be harder without unified vector space. Dual stacks seem to do transitions from city to highway and back pretty well, it is seldom done in very complex situations anyway. But I think recognition is pretty agnostic to how the improvement in functionality is achieved.

 

[mongo]

Yeah, this is getting too far in the weeds, so this will be my last post on the subject.

I believe unified vector space and single stack are requirements for further recognition of FSD revenue because I don't think FSD is going to work well enough without both.

And I'm going by what Elon said about V11, as you quoted above. I expect UVS in V11 as well, but maybe that's wishful thinking.

I believe V11 is coming soon (Q3 or Q4). But it will need to be in beta awhile before wide release.

That pushes further FSD recognition into 2023.

Good discussion. My last comment is that the level to which NoA is currently recognized does not require single stack. If that is <100%, it will likely need a single stack UVS upgrade for full FSD recognition. If NoA is fully recognized now, dual stack is inefficient but sufficient.

 

[Thekiwi]

I was super aggressive on my quarterly estimates for Austin and Berlin when I first started calculating 2022 numbers only to have to take the forecast down each time I updated my forecast. I think it has clouded my judgment here. I know that Tesla can "turn it on" when all the pieces come together.
As Q3 progresses, let's see if we've got evidence to up the forecast with confidence.

I think right to be conservative. For example I think the odds of another multi-week long covid shutdown in Shanghai are probably greater than 50% this year. Unless everyone starts to use a bear/bull range for deliveries, and accompanying bear/bull EPS estimates, then it is probably wise to go with conservative estimates which will hopefully even out any Shanghai shutdown with better than expected production from Berlin & Austin.

 

[unk45]

They are not paid off at end of lease (that would require loan type terms). So Tesla would then need to record a loss since total lease payments are less than CoGS. So they get tracked as a depreciating asset. At lease end, they still carry unrealized value/ basis which offsets sale price.

$47,000 for car of which $33,000 is CoGS. Three year lease $4,500 down 520/ month = $23k. Still $10k net negative, and $24k under a direct sale.
For example (not sure if this is how it works): 5 year straight depreciation is $28k over 3 years, so Tesla would take a $5k loss on the expense sheet during the lease, and need to use the basis of $19k when disposing.

Whoever makes leases, less an operating lease, establishes a residual value, the contractual lease end value. If the net realized sale price at lease termination is less than the residual value there is a loss. No reputable lessor sets residual value above lease end expected net price, unless someone subsidizes (subvenes) either the money factor or the residual value or both. I believe Tesla has never done that.

COGS does not directly enter into this because the capitalized cost includes the purchase price. If the transaction would, at termination, cause a direct loss if residual value were achieved the loss would be recognized at lease inception.The only relevant basis at normal lease termination is the contractual residual value, unless the lease itself has other terms. Tesla's leases have a stated residual value, normally quite conservative. It would be exceedingly rare for Tesla to lose money on normal lease termination.

Even though Tesla does not allow the lessee to purchase the vehicle at lease end the residua value is still stated. The lessee pays the entire difference between the Capitalized Cost (initial purchase price including anything added to the vehicle as part of the initial sale including all applicable taxes and fees) and the Residual Value. In addition the Money Factor is used to calculate the entire interest due during the lease, which s called the Money Factor rather than interest rate because the entire contractual interest is added to the capitalized cost. Perhaps oddly, where not prohibited the interest due in the event of early termination is usually based on the infamous 'rule of 78's' which dramatically front loads the interest due.

The slightly arcane leasing practices are why US auto dealers are anxious to lease rather than sell outright, because the price can be manipulated with very, very few buyers understanding the contract. The Tesla practices are vastly less deceptive than are those typically written by auto dealers, and Tesla does not mark up nominal rates. That is one reason why Tesla originated leases have the industry lowest default rates. Lease securitization data does reflect this, but is not at all easy to decipher, partly due to collateral substitution clauses.

 

[unk45]

@The Accountant uses standard and logical assumptions for lease accounting.

The more arcane points simply explain why Tesla leases are so much easier to manage than are those fo nearly all other OEM leases. Notable exception: Mercedes Benz Financial Services has similar high quality paper, including the commercial vehicles. Nearly all others have very minimal initial quality standards, so allow egregious 'dealer adds'.

 

[mongo]

Whoever makes leases, less an operating lease, establishes a residual value, the contractual lease end value. If the net realized sale price at lease termination is less than the residual value there is a loss. No reputable lessor sets residual value above lease end expected net price, unless someone subsidizes (subvenes) either the money factor or the residual value or both. I believe Tesla has never done that.

COGS does not directly enter into this because the capitalized cost includes the purchase price. If the transaction would, at termination, cause a direct loss if residual value were achieved the loss would be recognized at lease inception.The only relevant basis at normal lease termination is the contractual residual value, unless the lease itself has other terms. Tesla's leases have a stated residual value, normally quite conservative. It would be exceedingly rare for Tesla to lose money on normal lease termination.

Even though Tesla does not allow the lessee to purchase the vehicle at lease end the residua value is still stated. The lessee pays the entire difference between the Capitalized Cost (initial purchase price including anything added to the vehicle as part of the initial sale including all applicable taxes and fees) and the Residual Value. In addition the Money Factor is used to calculate the entire interest due during the lease, which s called the Money Factor rather than interest rate because the entire contractual interest is added to the capitalized cost. Perhaps oddly, where not prohibited the interest due in the event of early termination is usually based on the infamous 'rule of 78's' which dramatically front loads the interest due.

The slightly arcane leasing practices are why US auto dealers are anxious to lease rather than sell outright, because the price can be manipulated with very, very few buyers understanding the contract. The Tesla practices are vastly less deceptive than are those typically written by auto dealers, and Tesla does not mark up nominal rates. That is one reason why Tesla originated leases have the industry lowest default rates. Lease securitization data does reflect this, but is not at all easy to decipher, partly due to collateral substitution clauses.

Thanks for the more detailed coverage.
What I was attempting to illustrate in response to

Can someone explain this a little more. Surely for some vehicles Tesla will own them outright at the end of the lease and have already charged the COGS. If they sell to a new customer for $40k isn't it all profit minus cost of sales? - ie. ~$38k.

was that, AFAIK, at end of lease, the basis is not zero. So selling a EoL car for $40k is not $38k pure profit. The car ultimately sells for $40k + lease down payment + lease payments.
Though I see you liked that post, so is it?
Or is it that, if Tesla owned it outright, it would be all profit (scenario is true), but no one would agree to those lease terms so it won't happen (senario is implausible)?

 

[unk45]

Thanks for the more detailed coverage.
What I was attempting to illustrate in response to


was that, AFAIK, at end of lease, the basis is not zero. So selling a EoL car for $40k is not $38k pure profit. The car ultimately sells for $40k + lease down payment + lease payments.
Though I see you liked that post, so is it?
Or is it that, if Tesla owned it outright, it would be all profit (scenario is true), but no one would agree to those lease terms so it won't happen (senario is implausible)?

By definition at lease termination Tesla owns vehicle outright. Their cost basis is the Residual Value plus transaction costs. That is it.

In original sales accounting, at lease origination a vehicle sale is recorded and booked. For GAAP purposes the lease is subject to income recognition that is different. Despite hat at normal lease end they still wind up with an asset at the Residual Value.

Any scenario that considers vehicle COGS after the initial sale is not correct. The unit P&L is determined on initial sale. After that gains or losses are due to the lease terms, so the lessor has profit or loss, not the seller.
When the seller and the lessor belong to the same corporate taxpayer the GAAP rules establish accounting recognition.

For more details we'd need to delve into the accounting rules which rival cryptocurrency for overcomplexity and inappropriateness. @The Accountant may have better ideas on how to state this, and with zero doubt he's better qualified than am I. Anyway, I hope this helps a little bit.

NOTE: this discussion is US retail sales specific. Every country has different rules, and different basis for treatment of lease definitions. There's even Hire Purchase in quite a few places, yet another permutation, not to mention Consortia- perhaps the weirdest one of all.

 

[Doggydogworld]

was that, AFAIK, at end of lease, the basis is not zero. So selling a EoL car for $40k is not $38k pure profit.

Correct. Oversimplifying, let's say Tesla leases a 60k Model Y for 36 months at 500/month. The car has 45k COGS and they assume 36k residual value, thus 9k of total depreciation over the 36 months. Each quarter they report:

1500 - Auto Leasing Revenue (3 months * 500/month)
750 - Auto Leasing COGS (9k depreciation / 12 quarters)
---------
750 - Auto Leasing Gross Profit
50% - Auto Leasing Gross Margin

When the lease ends the car is on Tesla's books at the 36k depreciated cost. If they sell it for more than 36k it generates a profit, if less a loss.

This accounting treatment can apply even when Tesla sells cars to 3rd party lessors, depending on the terms of the deal. Back in the day, when Tesla provided Residual Value Guarantees to individual customers, they were supposed to account for those deals as leases and got in trouble with auditors/regulators when they didn't.

 

[unk45]

Correct. Oversimplifying, let's say Tesla leases a 60k Model Y for 36 months at 500/month. The car has 45k COGS and they assume 36k residual value, thus 9k of total depreciation over the 36 months. Each quarter they report:

1500 - Auto Leasing Revenue (3 months * 500/month)
750 - Auto Leasing COGS (9k depreciation / 12 quarters)
---------
750 - Auto Leasing Gross Profit
50% - Auto Leasing Gross Margin

When the lease ends the car is on Tesla's books at the 36k depreciated cost. If they sell it for more than 36k it generates a profit, if less a loss.

This accounting treatment can apply even when Tesla sells cars to 3rd party lessors, depending on the terms of the deal. Back in the day, when Tesla provided Residual Value Guarantees to individual customers, they were supposed to account for those deals as leases and got in trouble with auditors/regulators when they didn't.

As a technical point. Residual value is not assumed. It is established.
If resale value is greater than residual value at lease termination, the lessor can make a gain.

Residual value is a contractual amount. Resale value is an actual market amount.

Residual value can be manipulated at will.
Resale value really is a market determined price.

 

[mongo]

By definition at lease termination Tesla owns vehicle outright.

Ah, good point. Tesla as the manufacturer starts with full ownership. I was thinking in consumer terms where owned outright means fully paid off which, in terms of receivables, an end of lease car is not (lease payments < cash sale price).
From a before resale to post resale perspective, Tesla's position would improve by (sales price - cost of sales), though their before sale position is worse than an initial direct sale (well sort of, asset instead of cash).
Did I get it right this time?

 

[unk45]

Ah, good point. Tesla as the manufacturer starts with full ownership. I was thinking in consumer terms where owned outright means fully paid off which, in terms of receivables, an end of lease car is not (lease payments < cash sale price).
From a before resale to post resale perspective, Tesla's position would improve by (sales price - cost of sales), though their before sale position is worse than an initial direct sale (well sort of, asset instead of cash).
Did I get it right this time?

YES! The only minor quibble might be "worse than..sale". Although Tesla does not do the aggressive tactics of auto dealers there are a couple things that make leases attractive to Tesla despite the accounting rules. The typical purchase price of a leased vehicle is well higher than for finance, and even higher than cash sales. Lessees generally choose higher spec vehicles, tend to care for the vehicle better than owners and tend to negotiate less (even with Tesla the trade in does have some negotiating.

So over time the leases are typically more profitable than are sales. The timing of income recognition is over time rather than outright, but eventually that evens out in stability. Of course we expect consistent growth for Tesla. Further, I expect Tesla insurance to be wrapped into leases at some point soon, that will further generate higher yield because lessors invariably have lower physical damage accidents than do either cash or loan buyers.

At present nobody outside Tesla can model these factors reliably. within a couple of years we should have enough data to work with. My assertions are based primarily on lease securitization vs loan securitization data, I use Mercedes Benz auto only, BMW sometimes, because there are sources for both loans and leases on the two. The cash buyer differences are from proprietary data. All of these are analogues, not direct Tesla data because the Tesla securitizations have change very substantially since they began, making inferences from earlier ones fraught. Lastly, each model has distinct dynamics.

 

[mongo]

YES! The only minor quibble might be "worse than..sale". Although Tesla does not do the aggressive tactics of auto dealers there are a couple things that make leases attractive to Tesla despite the accounting rules. The typical purchase price of a leased vehicle is well higher than for finance, and even higher than cash sales. Lessees generally choose higher spec vehicles, tend to care for the vehicle better than owners and tend to negotiate less (even with Tesla the trade in does have some negotiating.

So over time the leases are typically more profitable than are sales. The timing of income recognition is over time rather than outright, but eventually that evens out in stability. Of course we expect consistent growth for Tesla. Further, I expect Tesla insurance to be wrapped into leases at some point soon, that will further generate higher yield because lessors invariably have lower physical damage accidents than do either cash or loan buyers.

At present nobody outside Tesla can model these factors reliably. within a couple of years we should have enough data to work with. My assertions are based primarily on lease securitization vs loan securitization data, I use Mercedes Benz auto only, BMW sometimes, because there are sources for both loans and leases on the two. The cash buyer differences are from proprietary data. All of these are analogues, not direct Tesla data because the Tesla securitizations have change very substantially since they began, making inferences from earlier ones fraught. Lastly, each model has distinct dynamics.

Awesome Thanks!
Regarding worse than direct, that was only before the end of lease car was resold. Agree that lease plus sale ultimately brings in more (caveat, for a Tesla ;-) )

 

[unk45]

It's time to quantify the average realized gain Tesla might have were they to auction 2019 Model 3 LR coming off a 36 month lease. Obviously the clean ones they'll resell for higher prices than I now will show.
2019 Base Price $45,700
2022 current Mannheim Atlanta average 36,000 miles, clean $48,500
Tesla 2019 Model 3 LR typical residual value: $23,307
for these terms there will be no addition consumer charges because they did not exceed mileage allowance, no damage, no options.
It is almost absurd to realize they now can just auction those cars and take around $25,000 gain on each lease return. This probably cannot last too long but for 2022 if Tesla want to have some easy quick margin helpers all they need to do is sell off lease returns.
My calculations built from Atlanta because my Mannheim access is there.
This is such a large unit gain that I think we should quickly try to establish what lease return volumes are expected the rest of 2022.

I have not attempted to do that yet. Perhaps someone has a better approach than I would use, which is just extracting numbers from securitization pools.