Near-future quarterly financial projections

[luvb2b]

what i did is take your provision for warranty line and divide by delivered units. the provisioning changes as they expect more or less expenses, or as the reserve gets too high (low). i thought using the provision is better as it incorporates their expectations for the future vs. what they have already accrued.

you'll see the provision per vehicle stable for 4 quarters 17q2-18q1 and suddenly jump 18q2. do a little math assuming s/x provisioning was stable and you'll see the ~6% provision for each model 3.

Here's what Tesla said about their warranty expenses in the Q2 financial report:

"Warranty expense is recorded as a component of cost of revenues. Accrued warranty activity consisted of the following (in thousands):

…

Accrued warranty—beginning of period

Warranty costs incurred

Net changes in liability for pre-existing

warranties, including expirations and

foreign exchange impact

Additional warranty accrued from adoption

of the new revenue standard

Provision for warranty

Accrued warranty—end of period

[TD2] 2018/Q2 [/TD2][TD2] 2017/Q2 [/TD2][TD2] 2018/H1 [/TD2][TD2] 2017/H1 [/TD2] [TD2] $465,866 [/TD2][TD2] $306,951 [/TD2][TD2] $401,790 [/TD2][TD2] $266,655[/TD2] [TD2] -$49,604 [/TD2][TD2] -$25,384 [/TD2][TD2] -$94,285 [/TD2][TD2] -$48,400[/TD2] [TD2] -$10,917 [/TD2][TD2] +$8,915 [/TD2][TD2] -$10,416 [/TD2][TD2] +$2,653[/TD2] [TD2] — [/TD2][TD2] — [/TD2][TD2] $37,139 [/TD2][TD2] —[/TD2] [TD2] $118,664 [/TD2][TD2] $52,797 [/TD2][TD2] $189,781 [/TD2][TD2] $122,371[/TD2] [TD2] $524,009 [/TD2][TD2] $343,279 [/TD2][TD2] $524,009 [/TD2][TD2] $343,279 [/TD2]

For the three and six months ended June 30, 2018, warranty costs incurred for vehicles accounted for as operating leases or collateralized debt arrangements were $7.0 million and $12.8 million, respectively, and for the three and six months ended June 30, 2017, such costs were $7.4 million and $13.5 million, respectively."​

I believe the most important disclosure for our purposes here is the "Warranty costs incurred" line, which must include all warranty expenses, including labor, material and any directly billed services utilized, regardless of where they end up being assigned to in the end: automotive, services or SG&A.

Here's the history of "warranty costs incurred" for the last six quarters and the last 3 years looks like this, with a column for Model 3 deliveries:

Quarter

2015/all

2016/all

2017/all

—

2017/Q1

2017/Q2

2017/Q3

2017/Q4

2018/Q1

2018/Q2

[TD2] Warranty costs incurred [/TD2][TD2] Model S+X delivered [/TD2][TD2] Model 3 delivered [/TD2] [TD2] -$52,760 [/TD2][TD2] — [/TD2][TD2] — [/TD2] [TD2] -$79,147 [/TD2][TD2] — [/TD2][TD2] — [/TD2] [TD2] -$122,510 [/TD2][TD2] — [/TD2][TD2] — [/TD2] [TD2] — [/TD2][TD2] — [/TD2][TD2] — [/TD2] [TD2] -$23,016 [/TD2][TD2] 25,000 [/TD2][TD2] — [/TD2] [TD2] -$25,384 [/TD2][TD2] 22,000 [/TD2][TD2] — [/TD2] [TD2] -$39,481 [/TD2][TD2] 25,930 [/TD2][TD2] 220 [/TD2] [TD2] -$34,629 [/TD2][TD2] 28,320 [/TD2][TD2] 1,550 [/TD2] [TD2] -$44,681 [/TD2][TD2] 21,800 [/TD2][TD2] 8,180 [/TD2] [TD2] -$49,604 [/TD2][TD2] 22,300 [/TD2][TD2] 18,440 [/TD2]

So firstly, while Model S/X deliveries were pretty flat over 2017 (except a bit of a glut early summer and then an end of year burst), there was a spike of warranty costs at Q3 - a big recall perhaps?

Secondly, beyond recalls, most warranty repairs happen in the first couple of months of a vehicle - a typical front-loaded distribution of component faults and assembly quality faults.

Based on that I think we can say that the Model S+X "baseline" warranty costs for 25k deliveries are around $1,000-$1,500/vehicle, or about 1.0-1.5% of ASP - with the real number probably closer to 1%.

We can also say that the about 9,950 early Model 3's, all delivered by March 31, probably already had most of their warranty work done by end of Q2 (June 30).

So if we conservatively estimate early Model 3 warranty overhead by including the Q2 increases in warranty costs but not include any Model 3 delivered in Q2 in the unit count, and take the very lowest of the Model S+X warranty baseline cost ($25m), then we get a per early Model 3 absolute maximum warranty cost of:

((39m+34m+44m+49m)-4*25m)/9,950 = $6,600 (11% of ASP)
​

Note that this is close to the most pessimistic way to read these numbers and includes very early production batches that are not representative of Q3 production at all - and still it's not that bad for early production batches.

If we use the higher $35m estimate for the S+X warranty baseline (1.5% of ASP), then we get:

((39m+34m+44m+49m)-4*35m)/9,950 = $2,600 (4.3% of ASP)​

Also, if we assume that half of the 18,440 Model 3's delivered in Q2 already had much of their warranty work done and expensed by end of Q2 (there was an end of Q1 peak that was delivered in early Q2), then the unit count increases from 9,950 to 19,170 and per unit warranty costs decrease further:

((39m+34m+44m+49m)-4*35m)/19,170 = $1,350 (2.2% of ASP)​

Furthermore, if we consider that the first two quarters of Model 3 production went to employees, who wouldn't request 'paint speck' and other light defect warranty repairs, then we get the following for the about 19,170 Model 3's delivered in Q1 and the first half of Q2:

((44m+49m)-2*35m)/19,170 = $1,200 (2.0% of ASP)
​

Note that the $1,200 estimate still includes Q1 vehicles with ~40% higher panel gap defects, which are pretty expensive to repair.

So if we only look at the Q2 increase in warranty costs of +$5m (note that Model S+X deliveries were flat at 22k in Q1 and Q2, there was no big recall, so their warranty costt baseline should be similar), and attribute that $5m cost increase to the about 9,950 Model 3's delivered in the second half of Q1 and first half of Q2 (about 10,000 vehicles), then we get this cost estimate for the 'newest' Model 3's:

5m/10,000 = $500
​

Conclusion: the newer Model 3's we consider, the lower the warranty costs are expected to be, and the estimates seem to support this. I believe the current per unit incremental warranty costs for newly produced Model 3's, barring an expensive fleet-wide recalls, should be below $1,000 already in Q3 - maybe even lower than that. I.e. below 2% of ASP, closing in on the Model S/X levels.

This might sound surprising for a ramp-up that is only 12 months old and skipped the 'soft tooling' step, but it's not necessarily unexpected:

• The Model 3 was designed for mass manufacturing, the component count is much lower and the technologies are more automated, which reduces production quality noise once the lines are fine-tuned.
• The components are also much cheaper due to higher unit count sourcing, which reduces warranty material costs.
• Warranty labor costs might also be lower due to serviceability related re-design of certain aspects of the Model 3 that I assume they performed based on experience with the Model S/X.
• There's an incredible amount of scrutiny on early deliveries, due to the 'panel gap' and other Model 3 quality complaints publicity, there's a very detailed, very rigorous "Model 3 delivery check list" that other carmakers generally don't have to deal with. This too will fade out: later buyers won't complain as much about every paint speck or slight panel misalignment, plus the earlier scrutiny has put pressure on Tesla to improve quality to or beyond the quality levels of the competition. I.e. those who used early production quality complaints (which complaints were justified) as a FUD vehicle against Tesla (which is not justified) kind of did them a favor, by forcing improvements that improved Tesla's competitive position. The law of unintended consequences.

In any case your 6% estimate of Model 3 ASP looks conservative, so I wouldn't expect a negative surprise from that.

 

[Fact Checking]

what i did is take your provision for warranty line and divide by delivered units. the provisioning changes as they expect more or less expenses, or as the reserve gets too high (low). i thought using the provision is better as it incorporates their expectations for the future vs. what they have already accrued.

you'll see the provision per vehicle stable for 4 quarters 17q2-18q1 and suddenly jump 18q2. do a little math assuming s/x provisioning was stable and you'll see the ~6% provision for each model 3.

Just curious: have you attempted to add up all the historical S+X warranty reserves and compared them with expenses? If we knew that ratio we'd have a reasonable guess for Q3 warranty outflow.

Warranty reserves would be a balance sheet item - with no cash set aside, right? So while they matter to net income and EPS they won't directly impact cash flow, correct?

 

[Reality]

thanks. i think the items you highlight are legitimate risk points. were they really building out service capacity as they said? or are they shifting items that maybe should be in cogs to service? only the next report can answer that question. i have seen them take up warranty accrual meaningfully, i estimate to about 6% of the asp of a model 3. so it is possible they are taking post-delivery repairs against warranty reserves.

Thanks @luvb2b

I am glad you are taking a break. Managing yourself is as important as managing your portfolio. It is good to have balance. Just don't be like a Tesla exec and make that leave permanent! (i kid i kid, please dont ban me again)



Only response i have, is this is the logical construct of what under reserving may result in. Clearly there is a schedule that is detailing results vs estimates, if your results show a differences in your warranty estimate you have to 'true up' for the whole fleet. So even a minor adjustment during the period could look like a huge adjustment because it all hits in the current period. So your likely warranty assumption per car for that period would be too high because they are taking the load of the entire 'true up'.


***
Example:

I have delivered 100k cars with a 2k warranty reserve for each (200m reserve).


I deliver 10k cars during the period, but I realize that I should have a 2.1k warranty reserve per car.

(let's assume no warranty costs for simplicity, but you get it im sure)


now i have 110k cars out there and a warranty reserve that needs to be (110k * 2.1k = 231m)

So I take a 31m warranty hit that period on only 10k cars, or $3.1k per car!

So what was really just a mere 5% increase in warranty reserve, looks like a massive 55% increase in warranty reserve on cars in the current period! So what looks "bad" in the current period might not really be that "bad" in the big picture (5% increase in warranty reserve is nothing).


***

I do realize Tesla separates this out in the rollforward (pre vs new). I am just stating that the effect may look exaggerated by even minor assumption changes.


With that said, the bad part is that it shows how much even a small true-up can mean when extrapolating across an entire fleet when it hits in a single period. Warranty reserves are a heavy audit item. Not a lot of gaming outside of what is and isn't classified as falling under warranty.

 

[Reality]

just saw all the back and forth from you and @Fact Checking


You guys are way ahead of me, i was just reviewing the basics.

 

[Zhelko Dimic]

just saw all the back and forth from you and @Fact Checking


You guys are way ahead of me, i was just reviewing the basics.

I am a Bear. I come in peace. Please don't ban me.

I LOVE your tagline

 

[Reality]

should really say "Please stop banning me"

 

[Esme Es Mejor]

As someone said on Twitter

 

[brian45011]

<rant>
i guess this especially bugs me in the context of opex, which i still don't understand. i mentioned it before: we scraped the barnacles. we scrutinized all the expense items. we laid of 9% of the workforce. we took out one-time charges due to layoffs. and we even excluded the stock-based comp jump from the calculations. and after all that, expenses went up about 3% for the quarter. there's definitely more upside opex pressure than i had expected.
</rant>

My apologies if previously posted (little wee fee in Iceland/Greenland):

"Stock-based compensation expense associated with the 2018 CEO Performance Award is recognized over the longer of the expected achievement period for each pair of market capitalization or operational milestones, beginning at the point in time when the relevant operational milestone is considered probable of being met. If additional operational milestones become probable, stock-based compensation expense will be recorded in the period it becomes probable including cumulative catch-up expense for the service provided since the grant date. The market capitalization milestone period and the valuation of each tranche are determined using a Monte Carlo simulation and is used as the basis for determining the expected achievement period. The probability of meeting an operational milestone is based on a subjective assessment of our future financial projections. Even though no tranches of the 2018 CEO Performance Award vest unless a market capitalization and a matching operational milestone are both achieved, stock-based compensation expense is recognized only when an operational milestone is considered probable of achievement regardless of how much additional market capitalization must be achieved in order for a tranche to vest. At our current market capitalization, even the first tranche of the 2018 CEO Performance Award will not vest unless our market capitalization were to approximately double from the current level and stay at that increased level for a sustained period of time. Additionally, stock-based compensation represents a non-cash expense and is recorded as a selling, general, and administrative operating expense on our consolidated statement of operations. As of June 30, 2018, we had $710.6 million of total unrecognized stock-based compensation expense for the operational milestones that were considered probable of achievement, which will be recognized over a weighted-average period of 3.5 years. As of June 30, 2018, we had unrecognized stock based compensation expense of $1.51 billion for the operational milestones that were considered not probable of achievement. From March 21, 2018, when the grant was approved by our stockholders, through June 30, 2018, we recorded stock-based compensation expense of $62.4 million related to the 2018 CEO Performance Award. For the three months ended June 30, 2018, we recorded stock-based compensation expense of $55.7 million related to this award. "
​

All non-cash and added back in Cash Flow Statement.

​

 

[Reality]

My apologies if previously posted (little wee fee in Iceland/Greenland):

"Stock-based compensation expense associated with the 2018 CEO Performance Award is recognized over the longer of the expected achievement period for each pair of market capitalization or operational milestones, beginning at the point in time when the relevant operational milestone is considered probable of being met. If additional operational milestones become probable, stock-based compensation expense will be recorded in the period it becomes probable including cumulative catch-up expense for the service provided since the grant date. The market capitalization milestone period and the valuation of each tranche are determined using a Monte Carlo simulation and is used as the basis for determining the expected achievement period. The probability of meeting an operational milestone is based on a subjective assessment of our future financial projections. Even though no tranches of the 2018 CEO Performance Award vest unless a market capitalization and a matching operational milestone are both achieved, stock-based compensation expense is recognized only when an operational milestone is considered probable of achievement regardless of how much additional market capitalization must be achieved in order for a tranche to vest. At our current market capitalization, even the first tranche of the 2018 CEO Performance Award will not vest unless our market capitalization were to approximately double from the current level and stay at that increased level for a sustained period of time. Additionally, stock-based compensation represents a non-cash expense and is recorded as a selling, general, and administrative operating expense on our consolidated statement of operations. As of June 30, 2018, we had $710.6 million of total unrecognized stock-based compensation expense for the operational milestones that were considered probable of achievement, which will be recognized over a weighted-average period of 3.5 years. As of June 30, 2018, we had unrecognized stock based compensation expense of $1.51 billion for the operational milestones that were considered not probable of achievement. From March 21, 2018, when the grant was approved by our stockholders, through June 30, 2018, we recorded stock-based compensation expense of $62.4 million related to the 2018 CEO Performance Award. For the three months ended June 30, 2018, we recorded stock-based compensation expense of $55.7 million related to this award. "
​

All non-cash and added back in Cash Flow Statement.

​

thanks for this


How about this part:

"As of June 30, 2018, we had $710.6 million of total unrecognized stock-based compensation expense for the operational milestones that were considered probable of achievement, which will be recognized over a weighted-average period of 3.5 years. "



Any idea how they get that 710.6m number?

Musk gets 1b (1%) if they hit 20b Annual Rev (likely in 2019) AND 100b market cap (um....well)


I am not well versed in this calc, but my question is what happens when this shifts from 'proable' to 'remote' (i believe those are the technical terms).

Are they going to take all that benefit to OPEX? Kind of feels like a cashless honeypot to dip into when needed. Or am I misunderstanding?

 

[ValueAnalyst]

Just to follow up on the S/X figure, in 2017 they sold roughly 35% of their annual US Model X/S sales in Q3. If that pattern is the same for global sales, that would be about 37k S/X sales in Q3 based on last year's production. If inside evs estimates are accurate, they're at 1.5k more US Model S/X sales than the same time last year, and I wouldn't be surprised to see global S/X sales at 40k+in Q3. That also bodes well for them moving into the black in Q3, since the S/X gross margin was nearly 30% IIRC.

Tesla sold 25.6% of its 2017 S/X in 3Q17.

 

[omgwtfbyobbq]

Tesla sold 25.6% of its 2017 S/X in 3Q17.

Doh! I probably should have just looked at their quarterly reports. At the same time, I still think there's a decent chance S/X sales will be significantly higher because the full US tax credit will only last until 1Q18.

 

[luvb2b]

yes. i roughly understood this to be ~50m.

but here's what bugged me:

18q1 opex exc. one time items 1,053,400
18q1 total stock-based comp 141,639
18q1 sbc in cost of goods 15,078
18q1 opex exc one time items & sbc: 926,939

18q2 opex exc. one time items 1,136,888
18q2 total stock-based comp 197,344
18q2 sbc in cost of goods 13,198
18q2 opex exc one time items & sb: 952,742
q2 to q1 change 2.8%

meaning: excluding one-time shifts, excluding stock based comp, and including all the cost cutting, opex went up by about 3%.

My apologies if previously posted (little wee fee in Iceland/Greenland):

"Stock-based compensation expense associated with the 2018 CEO Performance Award is recognized over the longer of the expected achievement period for each pair of market capitalization or operational milestones, beginning at the point in time when the relevant operational milestone is considered probable of being met. If additional operational milestones become probable, stock-based compensation expense will be recorded in the period it becomes probable including cumulative catch-up expense for the service provided since the grant date. The market capitalization milestone period and the valuation of each tranche are determined using a Monte Carlo simulation and is used as the basis for determining the expected achievement period. The probability of meeting an operational milestone is based on a subjective assessment of our future financial projections. Even though no tranches of the 2018 CEO Performance Award vest unless a market capitalization and a matching operational milestone are both achieved, stock-based compensation expense is recognized only when an operational milestone is considered probable of achievement regardless of how much additional market capitalization must be achieved in order for a tranche to vest. At our current market capitalization, even the first tranche of the 2018 CEO Performance Award will not vest unless our market capitalization were to approximately double from the current level and stay at that increased level for a sustained period of time. Additionally, stock-based compensation represents a non-cash expense and is recorded as a selling, general, and administrative operating expense on our consolidated statement of operations. As of June 30, 2018, we had $710.6 million of total unrecognized stock-based compensation expense for the operational milestones that were considered probable of achievement, which will be recognized over a weighted-average period of 3.5 years. As of June 30, 2018, we had unrecognized stock based compensation expense of $1.51 billion for the operational milestones that were considered not probable of achievement. From March 21, 2018, when the grant was approved by our stockholders, through June 30, 2018, we recorded stock-based compensation expense of $62.4 million related to the 2018 CEO Performance Award. For the three months ended June 30, 2018, we recorded stock-based compensation expense of $55.7 million related to this award. "
​

All non-cash and added back in Cash Flow Statement.

​

 

[Fact Checking]

yes. i roughly understood this to be ~50m.

but here's what bugged me:

18q1 opex exc. one time items 1,053,400
18q1 total stock-based comp 141,639
18q1 sbc in cost of goods 15,078
18q1 opex exc one time items & sbc: 926,939

18q2 opex exc. one time items 1,136,888
18q2 total stock-based comp 197,344
18q2 sbc in cost of goods 13,198
18q2 opex exc one time items & sb: 952,742
q2 to q1 change 2.8%

meaning: excluding one-time shifts, excluding stock based comp, and including all the cost cutting, opex went up by about 3%.

(Firstly, an inconsequential detail: I believe the fourth line is 926,839, not 926,939. It doesn't affect the 2.8% figure nor your conclusions.)

What opex cost cutting effects could we have expected in Q2? The 9% layoffs were announced on June 11th, with a "this week" qualifier, but it probably took the rest of the week to do it. So that saved about 2 weeks worth of payroll from opex, max - or about 15% of the quarterly savings - and maybe less.

Most of the 'barnacle scrubbing' was done in late May and June I think, so if it has an opex effect we'd see at best 50% of it. But I think a fair chunk of the contractor costs was about discretionary capex: I think this is supported by the fact that Tesla made a lot of contractors super unhappy and has over a dozen claims against them in court in form of Mechanic's Liens. It's small sums of less than $10m total, but those filings suggest that these were various contractors that were installing & building bits of the Fremont factory - and those expenses would mostly be capex.

The Q1=>Q2 opex increases (excl. one-time) came from the following sources:

• Tesla SG&A, from $551m to $610m (+$59m, +10.7%)
• R&D rose from $322m to $341m (+$19m, +5.9%)
• Solar City SG&A from $140m to $145m (+$5m,+3.5%)
• Solar City R&D stayed flat at $45m
  
• A total increase from $1,058m to $1,136m, (+$78m, +7.3%)

The biggest part of that was the increase in stock based comp, $141m => $197m (+$56m, +39.7%):

• The stock was depressed in Q1 and had a big spike in June in Q2, which I suspect caused many employees to sell shares or exercise options.
• Another factor is that many of the laid off employees were white collar, many of whom were reportedly with Tesla for a long time. They might have exercised ancient stock options that vested already, which had to be accounted not as a one time expense related to the layoff, but as stock based compensation?

$2m of this SBC increase went into cost of goods, the rest ($54m) into opex.

Anyway, if we take out the SBC component (+$54m) from the +$78m increase then there's a +$24m opex increase left.

I believe the 9% layoffs will have an about ~$100m opex reduction effect in Q3, of which savings only about $15m was realized in Q2, so that alone should be enough to counter the +$24m increase.

The question is, how much of the opex increase scaled with revenue, and how much of it was a one-time increase or some ramp-up related expense which might have been lower in Q3?

 

[mrdoubleb]

(Firstly, an inconsequential detail: I believe the fourth line is 926,839, not 926,939. It doesn't affect the 2.8% figure nor your conclusions.)

What opex cost cutting effects could we have expected in Q2? The 9% layoffs were announced on June 11th, with a "this week" qualifier, but it probably took the rest of the week to do it. So that saved about 2 weeks worth of payroll from opex, max - or about 15% of the quarterly savings - and maybe less.

I know the Labor Code in the US is generally pretty lax compared to Europe and may change by state, but isn`t there any notice period? Not even like 2 weeks?

Over here it depends on whether you are an employee or a contractor and if the former, how many years have you served with the company. So in California, even if you are e.g. 8 years with Tesla as an employee and they let you go, it`s like here is your last paycheck as of today and now good luck?

I always thought Tesla did this no later than mid-month to settle all payments by the end of the Q.

 

[MitchJi]

Yeah, and in fact it's even better than that. In their Q2 report they described it the following way:

"We continue to expect Model S and Model X deliveries to be approximately 100,000 in total in 2018, constrained by the total available supply of cells with the 18650 form factor used in those vehicles."​

Note the 'approximately' qualifier, which many believe includes the possibility of 'a bit more than', because it's in the best interest of Panasonic as well to make as many 18650 cells as possible.

I.e. several independently sourced data points that Model S/X demand in Q3 is not only strong, but record strong.

I think that this means that they might be getting close to migrating the MS-MX to 2170’s. Doing that will increase the,pack sizes by at least seven percent and decrease their pack costs by at least 40-50 percent.

 

[MitchJi]

Calm down with the accusations, he most likely meant making and then delivering twice as many cars in Q3. Every car delivered has to be made first.

81k delivered is about 6k cars beyond the lowest end of their guidance - which is nice and makes it virtually certain that they are going to meet guidance comfortably.

I think that Rob does a good analysis on today’s podcast:
Tesla Daily: Tesla News & Analysis by Rob Maurer on Apple Podcasts

 

[brian45011]

I know the Labor Code in the US is generally pretty lax compared to Europe and may change by state, but isn`t there any notice period? Not even like 2 weeks?

Over here it depends on whether you are an employee or a contractor and if the former, how many years have you served with the company. So in California, even if you are e.g. 8 years with Tesla as an employee and they let you go, it`s like here is your last paycheck as of today and now good luck?

I always thought Tesla did this no later than mid-month to settle all payments by the end of the Q.

It's the "WARN" (worker adjustment & retraining notification) legislation, both state and federal.

https://www.edd.ca.gov/jobs_and_training/Layoff_Services_WARN.htm

 

[Fact Checking]

I think that Rob does a good analysis on today’s podcast:
Tesla Daily: Tesla News & Analysis by Rob Maurer on Apple Podcasts

Any notable highlights?

 

[Zhelko Dimic]

yes. i roughly understood this to be ~50m.

but here's what bugged me:

18q1 opex exc. one time items 1,053,400
18q1 total stock-based comp 141,639
18q1 sbc in cost of goods 15,078
18q1 opex exc one time items & sbc: 926,939

18q2 opex exc. one time items 1,136,888
18q2 total stock-based comp 197,344
18q2 sbc in cost of goods 13,198
18q2 opex exc one time items & sb: 952,742
q2 to q1 change 2.8%

meaning: excluding one-time shifts, excluding stock based comp, and including all the cost cutting, opex went up by about 3%.

Lots and lots of overtime in Q2 to hit goals is the only thing that makes sense to me...

 

[Fact Checking]

Lots and lots of overtime in Q2 to hit goals is the only thing that makes sense to me...

Dunno, most of Q2 was underutilized, due to the 200k deliveries limit.

Last 2 weeks was busy - but most of the overtime would be factory workers, whose overtime would be accounted as cost of revenue.

Most of the opex increase comes from SG&A.

A +$22m increase isn't that bad, especially if we didn't see the full effects of cost cutting yet.

But Tesla does tend to find ways to increase opex and I'm curious whether they are going to keep the "flat opex" guidance for Q3.

If yes then good Q3 results should be in the bag. If not then it could be a nail-biter.

Note that @luvb2b's latest Q3 guidance does include about $20m higher opex, excluding one-time charges, and diluted GAAP EPS is somewhat of a nail-biter, with $0.09 - which is still above all Street estimates.

I'm wondering why the Street estimates are so low. That's only possible if they don't buy the deliveries and opex guidance and assume a big miss for one or both of these.

BTW., the comparatively low share price during much of Q3 should reduce stock based compensation costs somewhat.