Near-future quarterly financial projections

[ReflexFunds]

It's not true and very misleading. The resale value guarantee liabilities are more than off set by Operating Lease Vehicles on the asset side of the Balance Sheet.

When a Resale/Residual Value Guarantee transaction was originated, ALL the cash for the full purchase price was paid up-front when the vehicle was first titled by the individual buyer or Banking Affiliate lessor. Over the term of the guarantee, the deferred revenue is recognized ratebly and the cost of the vehicle is depreciated so that at end of the guarantee period, both the liability and the depreciated asset value are equal at the guaranteed value. If the beneficiary of the guarantee keeps rather than returning the vehicle at the end of the guarantee period, both the remaining liability and the vehicle's remaining net asset value are removed from the Balance Sheet with no effect on the Income Statement (other than a small amount of warranty reserve is booked since warranty costs are expensed during the guarantee period.)

If the guarantee beneficiary returns the vehicle at the end of the period, cash is used to pay the beneficiary the contractually agreed guaranteed amount, and the vehicle becomes part of Finished Goods Inventory at the lower of cost (remaining un-depreciated ledger value in Operating Leases) or market.

The most recent Cash Flow from Operations section shows Tesla used $2.8 million for Operating Lease vehicles in 3Q18 and $188.9 million YTD

This is correct, but it's worth noting new operating leases are still costing Tesla c.$150m per quarter. Cars sold under Tesla's inhouse leasing program were 3% of deliveries in Q3 or 2,513 S/X in Q3. Assuming average cash costs of these at c.$62k means Tesla funded $157m new leases.

The reason operating leases in cash flow statement consumed $3m rather than $157m cash in Q3 was likely because old lease cars returning to Tesla were reclassified from operating lease vehicles to finished goods inventory.

On the cash flow statement this would have corresponded to c.$154m cash generated by operating leases and $154m cash consumed by inventory. These cars would later be sold second hand and recorded as service revenue, so the inventory cash burn is then cancelled out by booking it as revenue.

So net net, Tesla is currently funding direct new leases by selling second hand cars once they are returned to Tesla from leases started 2 years ago.

Most leases are now actually sold through third parties rather than Tesla's direct lease program, so Tesla collects cash upfront from leasing partners and books revenue immediately as if it was a customer sale.

 

[brian45011]

This is correct, but it's worth noting new operating leases are still costing Tesla c.$150m per quarter. Cars sold under Tesla's inhouse leasing program were 3% of deliveries in Q3 or 2,513 S/X in Q3. Assuming average cash costs of these at c.$62k means Tesla funded $157m new leases.

The reason operating leases in cash flow statement consumed $3m rather than $157m cash in Q3 was likely because old lease cars returning to Tesla were reclassified from operating lease vehicles to finished goods inventory.

On the cash flow statement this would have corresponded to c.$154m cash generated by operating leases and $154m cash consumed by inventory. These cars would later be sold second hand and recorded as service revenue, so the inventory cash burn is then cancelled out by booking it as revenue.

So net net, Tesla is currently funding direct new leases by selling second hand cars once they are returned to Tesla from leases started 2 years ago.

Most leases are now actually sold through third parties rather than Tesla's direct lease program, so Tesla collects cash upfront from leasing partners and books revenue immediately as if it was a customer sale.

"Services and other revenue consists of repair and maintenance services, service plans, merchandise, sales of used Tesla vehicles, sales of electric vehicle components to other manufacturers and sales of non-Tesla vehicle trade-ins." Sales of used Tesla vehicles include those returned from resale and residual value guaranteed transactions.

Do you have any thoughts on what is causing the not inconsequential (though slightly declining) losses in Services and Other?

1Q18 -$121.6 million Gross Profit --- -46% Gross Margin
2Q18 -$116.2 million Gross Profit --- -43% Gross Margin
3Q18 -$114.7 million Gross Profit --- -35% Gross Margin

Also, Nevada's latest Economic Impact Study stated:

The SB1 incentive package included $195 million in Transferrable Tax Credits. However, the SB1 incentive package did not create any new Transferrable Tax Credits...

As of December 1, 2018 Tesla, has received Transferable Tax Credits in the amount of $173,087,499 based on job creation and capital investment.
​

Do you know where these credits appear in the financial statements Tesla files?

 

[ReflexFunds]

"Services and other revenue consists of repair and maintenance services, service plans, merchandise, sales of used Tesla vehicles, sales of electric vehicle components to other manufacturers and sales of non-Tesla vehicle trade-ins." Sales of used Tesla vehicles include those returned from resale and residual value guaranteed transactions.

Do you have any thoughts on what is causing the not inconsequential (though slightly declining) losses in Services and Other?

1Q18 -$121.6 million Gross Profit --- -46% Gross Margin
2Q18 -$116.2 million Gross Profit --- -43% Gross Margin
3Q18 -$114.7 million Gross Profit --- -35% Gross Margin

Also, Nevada's latest Economic Impact Study stated:

The SB1 incentive package included $195 million in Transferrable Tax Credits. However, the SB1 incentive package did not create any new Transferrable Tax Credits...

As of December 1, 2018 Tesla, has received Transferable Tax Credits in the amount of $173,087,499 based on job creation and capital investment.
​

Do you know where these credits appear in the financial statements Tesla files?

I assume most revenue in the services business is trade in cars, end of lease Tesla car sales and some other end of quarter clearance of Tesla inventory (loaner cars, test drive cars etc). I'd assume this second hand car business is close to 0% margin on average.

I guess most of the losses likely relate to the service business and is mainly due to lack of operating leverage. They need a global service infrastructure (for both warranty related service which are expensed via auto COGs, and other service business) but don't yet have a large enough global fleet to leverage the costs and increase utilisation of infrastructure and staff.

I haven't done a lot of work on this line though and i don't know for sure.

 

[neroden]

Do you have any thoughts on what is causing the not inconsequential (though slightly declining) losses in Services and Other?

I'd lay bets a lot of it is Service. They write a LOT off to goodwill, and they have a LOT of service demands every time a new (buggy) model comes out, though it seems to settle down.

I would definitely prefer it if the used car sales were separated out from the service costs, but they aren't

 

[ReflexFunds]

I'd lay bets a lot of it is Service. They write a LOT off to goodwill, and they have a LOT of service demands every time a new (buggy) model comes out, though it seems to settle down.

I would definitely prefer it if the used car sales were separated out from the service costs, but they aren't

"Services and other revenue consists of repair and maintenance services, service plans, merchandise, sales of used Tesla vehicles, sales of electric vehicle components to other manufacturers and sales of non-Tesla vehicle trade-ins." Sales of used Tesla vehicles include those returned from resale and residual value guaranteed transactions.

Do you have any thoughts on what is causing the not inconsequential (though slightly declining) losses in Services and Other?

1Q18 -$121.6 million Gross Profit --- -46% Gross Margin
2Q18 -$116.2 million Gross Profit --- -43% Gross Margin
3Q18 -$114.7 million Gross Profit --- -35% Gross Margin

Also, Nevada's latest Economic Impact Study stated:

The SB1 incentive package included $195 million in Transferrable Tax Credits. However, the SB1 incentive package did not create any new Transferrable Tax Credits...

As of December 1, 2018 Tesla, has received Transferable Tax Credits in the amount of $173,087,499 based on job creation and capital investment.
​

Do you know where these credits appear in the financial statements Tesla files?

I have looked a bit more at the service & other business now.

Average Tesla fleet size was 410k in Q3, assuming maintenance plan revenues are c.$600 per car and non warranty service/repair c.$500 per car on average, then we get $246m annual maintenance plan revenue and $205m other service revenue.

Total annualised service & other revenue was $1,305m in Q3, so this leaves $854m revenue for the used car business (assuming Grohmann now at 0).

Annualised service & other COGs was $1,780m in Q3. If we assume used car business is 0% average margin, then the service business COGs was $926m.

So this leaves service business revenue at $451m and service COGs at $926m.

The actual (non accounting) cost of Tesla’s overall service business infrastructure also includes the warranty costs incurred ($59m in Q3, $234m annualised or $572 per car) even though this is expensed to the warranty provision built out of the auto COGS.

Therefore Tesla’s total service infrastructure costs c.$1,160m per year. Assuming 250 service centres with 15 average staff and $100k average staff cost, the staff costs account for $500m of this. Service spare parts stood at $301m on the balance sheet at Q3. Assuming 9 months inventory turnover (it is going to be slow because there are a huge number of spare parts needed in stock at each service centre for each vehicle), then parts costs is c.$401m. This leaves $260m other costs, i’d guess a large component of this may be loaner car costs.

I imagine there are also some service business costs in SG&A, but probably not much other than rent costs.

These are all very rough numbers, let me know if anything looks way out but the numbers seem to make sense to me.

So overall, the problem looks like a lack of operating leverage in the service infrastructure and underutilisation of staff, infrastructure and spare parts. This will be solved as Tesla increases its fleet size and learns to operate the business more efficiently.

 

[neroden]

Tesla actually has to expand its service infrastructure, just for geographic coverage.

Existing service centers are already overworked, too. So basically the only way to get operating leverage from the service network is to make the cars break down less often, so that the fleet can expand without a proportional expansion of the service staff levels. I think this is happening, but I can't be sure. There is a known "new model breaks down more often" phenomenon, of course.

 

[generalenthu]

Troy's tracker now showing lower q4 production for q3 than q4. That's quite weird. Sampling rate is a bit lower, but that should not impact the estimates too much.

Edit: just for model 3

 

[schonelucht]

Well at some point, pent up demand from the reservation list should quieten down and Model 3 deliveries in the USA move to a more gradual selling trend. Possibly that point is now. But even then it's still not a big deal. China/EU production can start up a little sooner and with the second leasing bond, Model 3 USA leasing is a step closer for another round of releasing pent up demand. I am not so worried about a slight slide in deliveries. A fall in production however would need some good explanation.

 

[neroden]

I'm suspicious of sample bias in Troy's tracker; US LR cars have now completely run through the pent-up reservations demand, which means subsequent buyers are highly unlikely to even know about his tracker let alone report data. VIN scrambling increases the odds of not being able to interpret the production from the "highest VIN seen".

Tesla had low production in the early part of Q3, and I just don't believe that they've dropped enough to fall significantly *below* that production level in Q4, though I expect it'll be only slightly higher. Should be somewhat lower ASP (because of the MR) but significantly lower COGS (as they got their act together). Reports are also that battery sales are up significantly. Based on this, I'm basically predicting the same operating cash flow as last quarter, or somewhat higher.

 

[Zhelko Dimic]

Troy's tracker now showing lower q4 production for q3 than q4. That's quite weird. Sampling rate is a bit lower, but that should not impact the estimates too much.

Edit: just for model 3

One possible way to think about this is that Tesla can borrow only 85% of the cost, i.e. 70% of the price from the warehouse facility. So that increases interest costs, but also ties in capital.
Once Tesla moves from delivering quickly in the USA, they need significant capital tied into inventory until car is delivered in the Europe/China.

I don't know that my numbers are reasonable, but here is a primer: say delivery overseas is 50 days from production instead of 10 days in the USA, at average price of $55K, production of 6000/week, i.e. 857/day
Cash flow impact compared to sale in USA: $55K*(50-10days)*(100%-70%)*857 = $565 million cash tied in (difference between what warehouse line will pay vs. value of the sales, actual inventory is $1.8B in transit)

Of course, this is simplistic analysis and this much capital wouldn't be tied until 40 production days were dedicated to overseas, and also cash flow impact I've shown is compared to sales in USA, not to costs.

However, finally my point: If Tesla is trying to show great cash flow, and to squeeze every last dollar from the pipeline, I guess they would rather drop production to the level of the existing sales opportunity by the end of Dec 31st, rather than start producing for Europe/China. Come Jan 1st, they can restart production full speed, as it's only cash snapshot and cashflow at/by Dec 31st that is important. Before anyone say this is fake, I'd say it's smart, prudent cash and book management. This is one of those things they wouldn't need to worry about if they were private

 

[Zhelko Dimic]

One possible way to think about this is that Tesla can borrow only 85% of the cost, i.e. 70% of the price from the warehouse facility. So that increases interest costs, but also ties in capital.
Once Tesla moves from delivering quickly in the USA, they need significant capital tied into inventory until car is delivered in the Europe/China.

I don't know that my numbers are reasonable, but here is a primer: say delivery overseas is 50 days from production instead of 10 days in the USA, at average price of $55K, production of 6000/week, i.e. 857/day
Cash flow impact compared to sale in USA: $55K*(50-10days)*(100%-70%)*857 = $565 million cash tied in (difference between what warehouse line will pay vs. value of the sales, actual inventory is $1.8B in transit)

Of course, this is simplistic analysis and this much capital wouldn't be tied until 40 production days were dedicated to overseas, and also cash flow impact I've shown is compared to sales in USA, not to costs.

However, finally my point: If Tesla is trying to show great cash flow, and to squeeze every last dollar from the pipeline, I guess they would rather drop production to the level of the existing sales opportunity by the end of Dec 31st, rather than start producing for Europe/China. Come Jan 1st, they can restart production full speed, as it's only cash snapshot and cashflow at/by Dec 31st that is important. Before anyone say this is fake, I'd say it's smart, prudent cash and book management. This is one of those things they wouldn't need to worry about if they were private

BTW, one reason to do this is to simplify the story telling. Tesla has a big story covered, but it's been unable to tell financial story, except to enlightened financial analysts, and enthusiasts like us.
By removing complications of shipping overseas, situation is much simpler, no need to add footnotes and they they get to say: "see, this is how well we are doing". Then even semi-blind analysts can understand why cash flow is a bit worse next quarter, as they can compare to the one clean quarter.

 

[ReflexFunds]

I'm suspicious of sample bias in Troy's tracker; US LR cars have now completely run through the pent-up reservations demand, which means subsequent buyers are highly unlikely to even know about his tracker let alone report data. VIN scrambling increases the odds of not being able to interpret the production from the "highest VIN seen".

Tesla had low production in the early part of Q3, and I just don't believe that they've dropped enough to fall significantly *below* that production level in Q4, though I expect it'll be only slightly higher. Should be somewhat lower ASP (because of the MR) but significantly lower COGS (as they got their act together). Reports are also that battery sales are up significantly. Based on this, I'm basically predicting the same operating cash flow as last quarter, or somewhat higher.

Troy bases his delivery forecasts on production estimates using a chart of distribution of assigned VINs. He then applies to this an estimate of inventory cars without VIN assigned (a bit of a fudge in my opinion) to get to number of cars delivered.

I agree that with Teslas increased scrambling of produced VINs, increased non-assigned cars in inventory and lower sample rates for the survey, the forecasts are likely to get less accurate,

I tried using Troy’s data to arrive at a delivery estimate using an alternative method, using a model for delivery date sample rates, rather than assigned VINs distribution.

I estimate Q2 reservation holder sample rate at 8.1% and non reservation holder at 1.4% - 6x seems to make sense for relative fandom/delivery anticipation. From here i reduce sample rate for each by 8.5% per month due to lower wait times, less focus on production problems etc. These numbers reconcile with the known Q2 and Q3 overall sample rates.

This means by Dec reservation holder sample rate is c.4.8% and non res c.0.8% (similar to S).

This leads to a Q4 delivery estimate of 56,652 of which 83% are from non res holders.

This is relatively close to Troy’s alternative production forecasting method, and it also seems to align with Electrek’s report of 81% of US delivery targets achieved on 18th December (aligns with an internal target of c.60-61k model 3s).

If my model is correct, overall in 2H18, 80,152 orders are from non res holders vs 50,634 from res holders. Which could be suggestive of c.160k annual US demand for the current model 3 options without leasing.

 

[EVNow]

This leads to a Q4 delivery estimate of 56,652 of which 83% are from non res holders.

Another data point is that Musk said Insideevs numbers were wrong for October (?).

Insideevs has Q3 numbers as 14k, 18k, 22k. Then October, November as 17.7k and 18.6k. So, if the October number is actually higher (would make sense, because of large number of in transit etc from Sep getting pushed to October) - we can say Oct & Nov were about 19k each. That leaves December to be just 22k to reach total of 60k. This looks possible - even though Troy has a 55k estimate. May be they have produced enough for 65k, but they will be left with a sizeable inventory if they couldn't generate enough last minute sales - this would account for last minute push to get more people to quickly order and take delivery of the cars.

I do think market will like 60k, but not 55k (which would be same as Q3) with a large left over inventory.

 

[defc0n]

I just don’t see a possibility of Tesla failing to reach 60k model 3 deliveries. How many demand levers have been pulled? Next to none in my opinion. If Tesla were worried about posting a bad delivery number, we should be seeing more scrambling. I very much get the vibe that they are in a comfortable position.

Keep in mind the 8000+ model 3 in transit from last quarter. Not as many as the 11000+ in transit from Q2, but still sizable, and production/logistics was surely much smoother throughout Q4.

This should be a near repeat of last quarter, and Europe will fulfill the slack that will happen in North America in Q1 to make it 3 in a row.

I’m looking forward to delivery numbers.

 

[mrdoubleb]

For what it's worth, Norway came in just above the Q3 numbers and Europe in general was doing very well through the end of November (no December data yet except for Norway). I think S/X will not disappoint. It really is just a matter of how many 3s they could deliver, but with no production or weather issues and October + November already 4k above same time last quarter, why would they not hit the target?

 

[Fact Checking]

Come Jan 1st, they can restart production full speed, as it's only cash snapshot and cashflow at/by Dec 31st that is important. Before anyone say this is fake, I'd say it's smart, prudent cash and book management.

Note that other carmakers are doing similar end of quarter inventory management as well, for example BMW sales in the U.S. fluctuate with a 20-30% quarterly cycle:

So the inventory management and working capital optimization Tesla is doing is not only prudent but also done by 100-year old carmakers.

 

[Fact Checking]

There is a known "new model breaks down more often" phenomenon, of course.

Applies to ICE luxury cars with a parts count of over 40,000, thousands of which are moving.

The Model 3 only has 10,000 parts, with less than 100 moving parts.

So I'm wondering how much this improves the usual bathtub curve of reliability (which applies to car factories just as much as individual cars):

While the S/X only has average reliability historically, they also include R&D risks and were first generation designs.

The Model 3 is a second generation mass manufacturing design, with special focus on simplicity of manufacturing and a drive train designed for a million miles.

While only time can tell whether Tesla is able to execute this part of their Model 3 plans, if they are executing it only half as well as they were executing their Q3 financials then it's going to be a very nice reliability surprise down the road.

 

[mongo]

Applies to ICE luxury cars with a parts count of over 40,000, thousands of which are moving.

The Model 3 only has 10,000 parts, with less than 100 moving parts.

Teslas trade off mechanical complexity for eletrical complexity. The pack alone is 4,416 cells connected together which is 17,664 external wire bonds and 8,832 internal bonds. Each electronics module has thousands of connections and components most of which are critical (as do modern cars to a lesser level). So you trade the bathtub of mechanical (and electrical, still a lot of modules) failures of an ICE for the bathtub curve of electrical (and a little mechanical) of a Tesla.

Electronics failure rate is primarily influenced by temperature*duration and thermal cycles. Tesla seems to manage that well. However, there os nothing like a new model with hundreds of thousands on the road to reveal where the weak points in your build process or design are (XBox Red Ring of Death, ICE oil system failures)

Side note:
I think the Tesla moving part count groups the subassemblies like bearings as one item (verses 10+). The drive unit has the rotor, reduction gear, and final gear/ differential that is 6 bearings for 60+ parts. Then there are 2 half shafts with 11 parts each (22 more), Then the wheel hubs (>>10+ each), brake piston, caliper, and pads (4 each) 28+. Already over 100 parts, without the front wheels, pumps, fans.

Not sure if the ICE part count does the same thing (not enough fingers and toes to estimate). Toyota says 30k parts total per vehicle, that is likely the BOM level to them. With your 40k number and moving part estimate in the multi-thousands I'd guess that total and moving part count does include each sub-component.

 

[Singer3000]

Troy’s work has been great in recent quarters and his estimates for q4 shouldn’t be easily dismissed.

However I would only observe that we haven’t had factory leaks indicating a resumption of production hell, with unverifiable twitter datapoints indicating that production at both Fremont and Giga has been quite smooth. And as noted by @defc0n, if there is truly a demand problem and Tesla have been have been unable to sell +53k this quarter, you’d have expected far more effort at demand creation.

TSLAQ types are getting quite foamy-mouthed at Elon’s repeated recent tweets highlighting the drop in the tax credit, as if this indicates panic amidst completely saturated demand. I think the true answer lies in this discussion from the Q3 call:

“it occurs to me that, even if the only thing - like even if this was the only thing that Tesla did different was to shorten the time from factory to end customer. In any given company that would outcompete all of the companies over time. It would not be a contest”.

In short, Tesla has finally learnt working capital discipline. Almost three years after Jason Wheeler pronounced cash is king and then oversaw a substantial deterioration in Tesla’s cash conversion cycle.

The consequences of this? Gut says a decent beat on units produced vs q3 but especially so units delivered. And given the apparent lack of production chaos this quarter, I think we’ll see further improvement in gross margin despite lower ASP and tariff headwinds.

 

[ReflexFunds]

Troy’s work has been great in recent quarters and his estimates for q4 shouldn’t be easily dismissed.

However I would only observe that we haven’t had factory leaks indicating a resumption of production hell, with unverifiable twitter datapoints indicating that production at both Fremont and Giga has been quite smooth. And as noted by @defc0n, if there is truly a demand problem and Tesla have been have been unable to sell +53k this quarter, you’d have expected far more effort at demand creation.

I think Elon should be happy enough with c.57k deliveries. QoQ delivery growth means they meet their delivery guidance, and together with production cost improvements should be enough to drive QoQ growth in net profit too.

My assumption is, given QoQ profit growth in the bag (even on flattish QoQ volume), Elon preferred to hold US demand levers back for 1H19. With leasing option and lower options availability, Tesla can sustain decent US demand in 1H19 and buy them time to lower production costs enough to release the base model 3. With US demand levers still available, together with Europe and Asia deliveries in 1H19, Tesla should have the demand to scale production to c.7k per week without having to release the base car. A higher production rate will lead to reduced depreciation and staff costs per car, together with likely reduced pricing from suppliers on purchasing scale. Going from c.4.5k to 7k per week should reduce production cost of a base model 3 by c.$3k.

So saving more of the US backlog for next year can help them to get to a position to produce the $35k model 3 profitably by mid 2019.