On December 5, 2014, I published an article explaining what I thought was wrong with Tesla Motors' (NASDAQ:TSLA) Supercharger accounting. Basically, dividing the amount of revenue that the company has deferred for the service by the number of cars sold gave you about $500 per car - this amount seems extremely inadequate considering the company states "Supercharging is free for the life of Model S."
On December 18, 2014, the SEC contacted Tesla and asked about largely the same issue. Perhaps the SEC decided to inquire about this based on my article - or perhaps not. In any case, Tesla's response is extremely interesting.
Let's go through the most important quotes in that response letter:
Supercharger equipment is depreciated over an estimated useful life of 12 years and infrastructure improvement costs are depreciated over the shorter of the lease term of the Supercharger site or five years.
Not disclosed before or since. Not in the quarterly reports, or the yearly report, or the letters to shareholders or anywhere else - seriously, google it. Nobody except those who happened to read this letter would have a clue as to the depreciation period for the Supercharger network. And seeing that the quote does not appear in any forum or article, it's safe to say virtually nobody had stumbled on this letter.
The $1.6 million of costs incurred during 2013 for the operation, repair and maintenance of the Supercharger network.
This is the first (and so far only) time Tesla has disclosed operating costs for the Superchargers. They have never disclosed capital costs, i.e. depreciation (or amortization as some people call it), which are not included in the $1.6 million they mention here.
You know how much a kilowatt-hour costs in the US? About 13 cents for homes, and significantly less for industrial customers. How much do you think Tesla spent for each kilowatt-hour delivered in 2013? Read on to find out.
We have identified the service for the owners of our Model S to connect to and charge their Tesla vehicles at a Supercharger station as a standalone deliverable and have accounted for it as a separate unit of account under ASC 605-25, Revenue Recognition - Multiple-Element Arrangements. In determining the amount of sales revenue allocated and deferred for this deliverable, we have used our best estimate of its standalone value, which represents the price that we would charge our customers if we were to sell it separately.
Our best estimate of the value of this deliverable is based on a cost-plus model that includes the estimated cost of energy that will be consumed plus a per-car estimate of forecasted Supercharger network depreciation, lease and maintenance costs expected to be incurred over the estimated eight-year life of the car plus a reasonable margin.
In my first article, I argued that surely the deferred revenue was intended to pay only for operating costs, not capital costs, and within operating costs, perhaps only for electricity. Even so, I concluded that $500 was too little.
But it turns out Tesla claims the revenue allocated to the Superchargers covers operation and depreciation... plus "a reasonable margin."
It's also the first time Tesla disclosed its extraordinarily short "estimated life" for the Model S, though to be fair they did include that factoid in subsequent 10-Qs. (Presumably the cars self-detonate after eight years).
Because we do not spend money on traditional advertising and consider our Supercharger network as our only major marketing effort from a business perspective, we recorded a portion of the costs related to our Supercharger network to cost of automotive sales and another portion to selling, general and administrative expense. This allocation was based on expected usage levels at Supercharger locations. Generally speaking, Superchargers located on more frequently traveled routes with eight or more charging stations were those that we estimated would have a higher utilization rate and are recorded to cost of automotive sales. Supercharger stations expected to have low utilization rates serve more as a marketing function for Tesla and we recorded these costs to selling, general and administrative.
So Tesla is assigning expenses to either cost of goods sold (COGS) or selling, general and administrative (SG&A) - the way it's done is essentially arbitrary. Again one has to wonder why the only place it's been disclosed in is a letter few people have read?
Many of us have speculated before about the reasons Tesla's SG&A is rising not just on an absolute basis, but per delivery. It now turns out that 60% of Supercharger costs go to SG&A, and since these costs have been rising faster than deliveries, this is in fact a factor behind the SG&A anomaly.
Notably, this methodology did not have a material impact to our reported automotive gross margin. Had we included all Supercharger costs within costs of automotive sales for the full-year 2013, our automotive gross margin would have decreased by approximately 0.1%.
This is misleading. I'll show the math later, but the short of it is that Tesla is talking about the SG&A costs for a single year. If you're going to compare SG&A to COGS, you have to include the SG&A expenses incurred in every year you provide service - eight according to Tesla, a lot more in the real world.
Implications of the letter: cost of the Supercharger stations
The first thing to note is that the letter allows us to better calculate how much the company has spent on the Supercharger network, and the cost per site. Previously, we simply divided net book value by number of stations and said "this doesn't account for depreciation." Now we can account for that.
Although the exact depreciation period is not disclosed, Tesla says it's 12 years for the Superchargers and 5 years (or less if the lease is shorter) for infrastructure improvements to those. I'll assume that averages out to 10 years - it wouldn't change the numbers much if we used 9, 11 or other similar number.
First we get the number of Supercharger stations open at the end of every quarter, according to Supercharge.info. (Tesla doesn't disclose a Supercharger count every quarter, but its numbers are very close to those of Supercharge.info)
Superchargers open by the end of the quarter
From this we calculate the number of total Supercharger months that have gone by since the network started. I do this by assuming that, if for example 50 stations opened in a given quarter, then by the end of that quarter they had accumulated 50 / 2 = 25 Supercharger months.
Supercharger months elapsed by the end of the quarter
And from that we get the average age:
Average age of a Supercharger station, in months
Well, all those numbers to arrive at the conclusion that the average Supercharger is just over one year old. I won't pretend we know figures to this level of certainty, so let's cut Tesla some slack and reduce that to a year. In other words: the Superchargers have lost 10% of their original worth to depreciation.
According to Tesla's latest 10K, the net book value of the Supercharger network was $166.6 million at the end of last year. Dividing by 0.9 we get the money invested by the company: $185.1 million. With 588 stations, that works out to $314,800 per station.
Tesla's letter to the SEC also confirmed that the value of unfinished Superchargers is included there, but with 588 stations working, the impact of an additional 10 or 20 half-built sites will be minimal. No matter how you square it, the cost of the Superchargers is about twice the $150,000 repeatedly stated by the company, an issue I covered previously.
If you're wondering if the stations have gotten incomparably bigger since those cost statements were made in 2013, I'll point out that currently Tesla currently gives numbers of 611 stations and 3,600 plugs, or 5.89 plugs per site, and that it's always been somewhere between 5.5 and 6, as you'll see in the comments section of my previous article. Whatever fluctuations there have been in the number of chargers per station cannot explain a doubling in price (especially given Tesla's increased scale and experience).
(Note: Tesla's website actually says "611 Supercharger stations with 3,600 Superchargers." Actually, by "Superchargers" they mean "plugs." In this article, a "Supercharger" and a "station" are the same. Usually, a Supercharger has three chargers, each serving two plugs - hence the typical number of six plugs per station).
Implications of the letter: cost of the Supercharging service
Tesla tries to convince the SEC that this accounting issue is a non-event by mentioning it only spent $1.6 million operating the network in 2013. But for most of the year and most Model S owners, there was no network to use. To put things graphically:
In December 2013 usage was 500MWh. The previous month, about 300MWh, and less before that. All in all, I count about 2GWh delivered during the year. It may be 1.9, it may be 2.1 - it doesn't matter. What matters is that, if their costs were $1.6 million, then they spent about 80 cents per kilowatt-hour - just operating the network. That doesn't include depreciation.
How much is depreciation, by the way? Earlier I calculated the number of Supercharger months elapsed by the end of a quarter. Rounding the cost of a station down to $300,000 (better for Tesla), and with depreciation of 10% a year, we get that any month of depreciation costs $30,000 / 12 = $2,500. So let's go back to the table showing cumulative months and subtract so as to show only the incremental months, i.e. how many Supercharger months go by during a quarter.
Supercharger months elapsed per quarter
We multiply by $2,500 to get depreciation:
Estimated Supercharger depreciation, in 1000s of USD
(Someone will be asking why don't I simply get the net book value of the Supercharget network and work out depreciation expense from that. Well, using the quarterly valuation numbers I don't see a way to account for the depreciation that happens during the first quarter a station is open; e.g. a station that opens October 1st will only show up in the accounts with the value it had by December 31st, by which time it would have suffered three months of depreciation. By contrast with the actual Supercharger count I could correct for this. It's not a big deal but especially in 2013 and 2014 it can skew the numbers as the number of stations opened in any given quarter was relatively large compared to the number of stations existing before that quarter).
Now we know the dollars, but we still don't know the kilowatts.
Tesla has never disclosed Supercharger usage in a consistent manner. It has always been an infographic here, a tweet there, a Facebook post reposted on InsideEVs over there - and this irregular stream essentially stopped in early 2015. So we cannot make a chart showing the capital costs per kWh. But we can work out some numbers.
For 2013, we have costs of $585,000, divided by 2GWh this is 29.3 cents. Of course usage may have been higher than 2GWh, but it would take 2.34GWh to drive it down to 25 cents.
Combined with operating costs, this means Tesla was spending about $1 per kWh, or 7 times more than most owners pay if they recharge at home. Given that a Model S covers about 3 miles per kWh, this represents a cost of 30-35 cents per mile. A similarly-sized combustion sedan, getting 20 miles a gallon, would spend only 10 cents if gasoline is $2 a gallon.
For 2014, we can look at this infographic from February 27, which shows cumulative consumption at 25GWh. Subtracting a bit over 2GWh from 2012 and 2013, plus whatever was consumed in January and February 2015, we can conclude 2014 usage was about 20GWh.
(I have cropped the image to show only the relevant part).
With $4.6 million in depreciation expense for that year, we're talking about 23 cents per kWh.
Finally, depreciation expense for 2015Q1 was $2.7 million. According to this infographic, by April 22 cumulative consumption was 29.5GWh.
As of December 16, according to Tesla's Hawthorne information screen, cumulative GWh were 16.75.
This time the calculation is more uncertain than before: just over four months, and almost 13GWh, separate both dates. But if you go with 9GWh (3GWh x 3 months) consumed between January and March 2015, then depreciation per kWh was 30 cents. In other words: even more than in 2013.
"But those numbers are misleading because most of the costs are fixed"
That seems a valid argument: if the amount of electricity delivered doubles, that doesn't mean the cost will double. It's likely that a lot of the operating costs (customer service) are fixed, or will increase less than usage. As for depreciation, it's a function of how many stations are in operation relative to the number of cars using the network - if utilization rates increase then depreciation per kWh will go down because the depreciation expense for a given station will be the same but said station will be providing more kWh.
The point is, there is no way to know because Tesla doesn't disclose this data. They haven't mentioned operating costs for 2014 and 2015. They haven't provided consistent data on usage.
Nevertheless, there are some things we do know.
First, as shown in the previous section the depreciation expense per kWh has been relatively stable, at 20-30 cents per kWh from 2013 to early 2015. Depreciation expense per kWh is directly related to utilization, as explained a couple paragraphs above.
Second, the real-world charging and utilization rates will always be far lower than some electric vehicle enthusiasts imagine. We can glean info from some dashboards or screens in Hawthorne and Amsterdam showing some detail - these were turned off in early 2015 (roughly at the same time as the company stopped divulging info by other means).
First: although the Model S can take up to 120kW, and Tesla's new chargers can pump up to 135kW (shared between two cars), the actual average charging rate per car is about 55kW. Factors such as power tapering when the battery is getting full, hot weather, sharing a charger with other car (usually every charger serves two plugs), and malfunctioning hardware all affect the charging rate. Here you can see an example from February 19, 2015: 40 cars times 38 minutes charging per car equals 24.33 car-hours. Given that the station delivered 1,375kWh, this implies a charging rate of 54.3kW. I have many other photos, and they all show similar rates.
And about utilization, the dashboard allows you to look at "infamous" (for overcrowding) stations such as San Juan Capistrano, and look at how "full" they actually are. As you can see in the image above, San Juan Capistrano delivered 2,579 kWh over the previous 24 hours (Wednesday-Thursday). It has 4 chargers with 7 stalls, an unusual setup.
With 480kW among the four chargers, the station could theoretically have delivered 11,520kWh - it delivered only 22.4% of this number. So even a station famous for being always full is not putting out even 25% of the energy it could in theory provide.
In fact, let's calculate the depreciation associated with a given utilization rate. We assume a $300,000 station with 3 chargers and 6 parking spots, total power of 360kW, and depreciation per month of $2,500 as calculated before. A utilization of 10% would give 360kW x 24h x 30 x 0.1 = 25,920kWh.
Remember that we had calculated depreciation at $2,500 per month. So, even if the average station gets to 10% utilization, or half as much as the most overcrowded stations in the world, you still have a depreciation cost of 10 cents per kWh. Going as crazy as San Juan Capistrano would only drive this down to 4-5 cents per kWh. In other words, depreciation is a significant expense - there's no way around that, and anyone "calculating" the cost of Superchargers without including it is only fooling himself.
It's not at all clear that Tesla can get anywhere near 10%. The figures are from early 2015. They are already using valets/attendants at some stations - Fountain Valley and San Mateo in California are the ones I know. At $15 an hour and 3,500 hours a year, one attendant costs upwards of $50,000 per station, per year. Then you have things like mobile Superchargers with their accompanying diesel-fired generators, tow trucks for when a whole station is down, lease (sometimes in very expensive cities), repair crews, call center operators, etc.
It would be so easy for Tesla to put an end to the speculation and disclose operating costs for the network in 2014 and 2015. That would show whether the 80 cents per kWh seen in 2013 were a fluke or whether the trend continues.
The attentive reader will have realized that if you know the number of Supercharger months elapsed per quarter, you can calculate utilization that way as well - for the whole network. I didn't make a chart because there isn't enough data to cover every quarter, but we can be fairly sure that:
- In 2013, consumption of 2GWh and Supercharger time of 234 months gives utilization of 3.3%.
- In 2014, consumption of 20GWh and Supercharger time of 1,845 months gives utilization of 4.2%
- In 2015Q1, consumption of 9GWh and Supercharger time of 1,095 months gives utilization of 3.2%
All of these numbers assume 120kW per charger and 3 chargers per station. Assuming 135kW would simply make all the utilization percentages numbers 11% lower (120 / 135 = 0.89), without altering the trend. In fact, since 135kW chargers have mainly appeared in 2015, my analysis favored 2015 over previous years. In short, there is no evidence the utilization rate is increasing, although of course absolute usage is growing fast.
Perhaps you have seen news reports of queues, waiting to charge at stations, etc. If so, you might be wondering how can utilization rates be so low. Well, that's not the topic of this article but, on top of the real-world charging rate being about half the nominal 120kW, there are two factors:
- Demand is very uneven. Nobody charges Tuesday at 5am.
- If you look at a Supercharger map, you will notice that lots of them are in countries or regions with very low population and/or very few Tesla customers.
So it's totally possible for the network as a whole to have a very low utilization and yet for some stations (or even whole cities or regions) to be overwhelmed.
And you also see why my previous estimates are useful: they give us a ballpark number that in fact is very good at illustrating the issue. Supercharger utilization in 2015Q1 may not have been 3.2% - it may have been 3.9%. But it doesn't really matter. The whole of the network cannot get anywhere close to the 20% that the busiest stations do.
If Tesla achieved 10% utilization, then depreciation would be 10 cents per kWh. At 5% utilization, it's 20 cents per kWh. That's what matters.
$500 doesn't buy you much these days
Just because you don't know everything, doesn't mean you know nothing.
We know that Tesla defers less than $500 per for each Model S (and now X) for the Supercharging service, to be used during an estimated life of eight years. We also know that this figure is supposed to cover "the estimated cost of energy that will be consumed plus a per-car estimate of forecasted Supercharger network depreciation, lease and maintenance costs... plus a reasonable margin."
Tesla's hasn't explicitly said if it's only including here the 40% of Supercharger expenses included in COGS, or all of it. Let's give them the benefit of the doubt and assume it's the former - that would mean its estimated costs, for 8 years, would be not $500 but $1,250.
So let's do the math:
- 20 cents per kWh in capital costs, as calculated before (this is only if Tesla ever achieves 5% utilization).
- 5 cents per kWh in the actual electricity. Of course this is ludicrously low, but it will serve to illustrate the problem.
- 5 cents per kWh in non-electricity operating costs. Again absurdly low, but let's explore the possibility.
With this set of impossible assumptions we get a cost per kWh of 30 cents. And with that in mind, Tesla's $1,250 give us a total of 4,167kWh, or just over 500 kWh per car, per year. That's only 1,500 miles.
But go back to the figures for 2015Q1. We already estimated 9GWh, at a moment when there were on average about 60,000 Model S on the road (56,800 delivered until December 2014, plus part of the 10,000 delivered during the following quarter). In other words: the average Model S was already doing about 50kWh per month, or 600kWh per year. And that's in a quarter without the holiday travel that July or December see.
The last numbers available are for early 2015, but we know that it was growing. In my December 2014 article, I noted that the annualized per-car usage had risen from 335kWh in June to 522kWh in October.
Finally, consider the depreciation expense during 2015Q1. Depreciation expense that quarter was $2.7 million, so in order to get depreciation per hour down to 20 cents per kWh would have required usage of 13.5GWh. This would imply a consumption per car not of 50kWh/month, but of 75kWh/month or 900kWh/year - which is to say nearly twice the figure that we estimated Tesla could pay for with its $1,250-per-car pot. Remember that even this estimate includes unrealistically low costs of operation, both for electricity and for non-electricity expenses.
It's a catch-22. If utilization is low then depreciation costs are high. If utilization increases, even if just to drive depreciation down to $0.20/kWh as in the example we used, then the Teslas on the road are consuming far more kWh than the company estimated. Either way there is not enough money.
There is only one possible conclusion: Tesla's estimated Supercharging costs are much lower than reality.
What is the effect of not including all Supercharger cost into COGS?
If 60% of Supercharger expenses go to SG&A rather than COGS, that would be $1,250 - $500 = $750 in extra costs. This would be a loss of 0.7-0.8% for the company's gross margin.
Perhaps you think that's negligible, but then again the company excluded all costs after the eighth year. Taking the $1,250 above, and extending them proportionally another 7 years (to reach 15), would give us (7/8) x 1,250 = $1,100. Together with the $750 they didn't include for the first 8 years, that's nearly $2,000, or a 2% decline in gross margin.
If you extended that to 20 years, as seems fit to do for a $100,000 sedan, the cost going into SG&A rather than COGS would be $2,600 - approximately a 2.5% fall in the gross margin. Gross margin is only 20-25% to begin with. And needless to say, $2,600 times the 107,000 Teslas on the road is a not insignificant amount of money.
And again, this is all calculated with Tesla's own estimated costs - which as we have seen are wildly unrealistic. The idea that one can brush aside Superchargers as an accounting irrelevance is absurd.
I have not calculated how expensive the Supercharging service is - only insiders can do that. But I have proven Tesla is under-budgeting for it.
And I know something else: Tesla always includes the revenue deferred for Superchargers in its quarterly reports, under the section of revenue recognition. But in the latest report it has disappeared.
So few data points, so much obfuscation
Tesla has published very few concrete facts about the Superchargers. With that in mind, consider:
- The originally stated cost of $150,000 per station was half the real price.
- Tesla estimates "the life of the Model S" as eight years. Undisclosed until the SEC asked, unexplained after that.
- They are only allocating about 40% of costs to COGS, dumping the rest on SG&A - this overstates the company's much-touted gross margin. Undisclosed except for the letter to the SEC, and hard to justify even if disclosed.
- They have stopped disclosing revenue deferred for the Supercharging service. No notice, no reason given - it simply disappeared.
- They first provided data on Supercharger usage inconsistently, and since a year ago hardly at all.
- Even if the $500 allotted per car are only supposed to cover only 40% of costs, and only during 8 years, the number is too low. Even with extremely generous assumptions it would provide about 500kWh per car-year - we already know annualized usage is at least 600kWh.
You get the point. We're way past three strikes - and that's only in the Supercharger issue.
An additional proposal to the SEC, if they're reading: Make Tesla disclose how much it has spent repurchasing cars. For a program involving hundreds, probably thousands of cars, it's stunning that the company hardly talks about it. In fact it's only mentioned in the 10Qs because it overlaps to some extent with direct leases and with the resale value guarantee, so when a car on one of these two programs is repurchased (before the lease is up, or before the period for RVG repurchase arrives), Tesla has to account for that. Except for that, there are zero mentions of these buybacks.
Yes, it has nothing to do with the rest of the article. But I haven't seen anyone bringing it up so I might as well do so.
Acknowledgements: I would like to thank Bryce_in_TX for calling attention to the SEC letter in a Montana Skeptic article. The analysis in this article is only mine, as are any errors or misinterpretations. I'd also like to thank the Tesla Motors Club users who took photos of the dashboard... but those were very long threads, and I didn't write down users' names. So I'll just thank the TMC community - we may disagree on a lot of things, but they do a great job at gathering info.
PS: the tables I showed exclude most calculations for reasons of brevity. If you want to see those, just send me a message and I'll email you the Excels.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.