Connect with us

Hi, what are you looking for?

CleanTechnica

Cars

Peeking Behind Tesla’s Cost of Materials Curtain

In previous articles, I have already discussed in detail how Tesla can achieve a break-even level with the Model 3. I have also discussed the variable unit cost of capital per Model 3, and I have gone to great length discussing the variable unit cost of labor per Model 3. Today, I would like to discuss the fixed unit cost of materials of Model 3.

Executive summary:

• The estimated cost of materials per Model 3 is $16,493, or ~$16,500.

• The estimated battery cost of materials per kWh is $80 for the long-range battery.

• The estimated unit cost for each of Tesla’s Model 3 is currently at $23,300 (this includes material costs, labor costs, capital costs, and relevant operating expenses).

• The estimated unit cost per Model 3 could potentially approach the level of $20,000.

In previous articles, I have already discussed in detail how Tesla can achieve a break-even level with the Model 3. I have also discussed the variable unit cost of capital per Model 3, and I have gone to great length discussing the variable unit cost of labor per Model 3. Today, I would like to discuss the fixed unit cost of materials of Model 3.

However, before entering this veiled area, I would like to emphasize the difference between material costs on the one hand and labor and capital costs on the other hand. One of the most important differences in terms of manufacturing is the way these two groups of costs behave.

As I have shown, capital costs and labor costs tend to decrease with higher levels of production. However, material costs do not. As the costs of production increase, material costs tend to stay flat, and in fact, depending on the national and global economy, they could go either slightly higher or slightly lower. So, keep this in mind when looking at the graphs below.

In this article, I will begin with the conclusion and move into the details later. To begin, the following two graphs show two different situations regarding Tesla’s Model 3 costs. The first graph shows the effect of placing the burden of all capital costs, labor costs, and operating expenses (or OPEX) on only the Model 3 car.

In this case, OPEX refers to the portion of indirect salary expenses of engineers doing research and development, the portion of indirect salary expenses of office people doing sales, general, and administrative functions involved in car manufacturing, and some overhead.

There are similar designations for people involved with the energy generation and energy storage division of Tesla as well as with Tesla’s other business divisions. However, these folks are not involved with car manufacturing and I have excluded them from the OPEX indirect salary expenses in order to focus only on car manufacturing expenses.

Why have I grouped them as indirect expenses? I have done this only for the sake of distinguishing them from the people involved in direct car manufacturing, such as factory workers and factory technicians. The distinction is only for accounting purposes. Tesla could, of course, not build cars without engineers who design cars, and without business people who develop strategy and market opportunities.

As as you can see, in the beginning, the graph shows a very steep decrease (“production hell”) of production costs, with unit costs decreasing from around $60,000 to about $28,000. This happens as production is increased from 1,000 units to around 3,500 units — after this, the graph starts to slowly flatten out. Today, the steepest portion of the curve is history for Tesla. The company has managed to overcome the more difficult portion of this stage (see the next chart as well for comparison).

The second chart is a little different. In this case, Model 3 is sharing capital costs, labor costs, and OPEX expenses with Model S and Model X. Sharing expenses, in this case, is of great benefit for the ramp-up portion of this new model.

As you can see, the beginning of the unit cost curve starts at around $30,000. This is below the base price of the Model 3. This is in contrast to the previous graph, which started significantly above the base price of $35,000. Moreover, the graph itself is less steep and a bit flatter.

Both charts show the individual curves for the cost of materials, the cost of capital, the cost of labor, and operating expenses, and they both show the unit curve cost, which adds these four costs together. Since the unit curve cost is composed of materials, it will always be higher than the cost of materials. The charts also make it possible to visualize the decrease of the unit cost curve as the costs of capital, labor, and operating expenses decrease.

When sharing costs with Model S and Model X, the charts also show the relationship between costs. Material costs are greater than the sum of capital, labor, and operating expenses. Labor costs are greater than capital costs, and capital costs are greater than operating expenses (material costs > labor costs > capital costs > operating expenses).

Both charts identify the unit curve cost in relation to the Model 3’s base price. They also clearly identify the intersection of the curve with the recently announced cost of the Model 3 via a German news organization, and they also show that the lowest achievable unit cost is around $20,000.

At a production level of 10,000 Model 3 cars per week, the unit cost of labor is around $2,140, the unit cost of capital is about $1,150, and the unit OPEX is around $190. Attempting to lower these costs further might prove difficult and counterproductive in terms of the overall required expenses to do it. Why? Notice the flatness of these three curves on the right side of the chart.

When the curves are not flat as in the left portion of the chart, then this presents the potential to decrease these costs. However, as all sorts of production efficiencies are applied, it becomes more difficult to wring further efficiencies out of the production process, and doing this usually requires different manufacturing methods or technologies, and these methods require expenses, and at some point spending more to produce more makes no sense. Is it possible? Maybe. Does it make sense? Perhaps.

Think about this way: At a production level of 10,000 Model 3 cars per week, the sum of capital,
labor, and OPEX is around $3,500 per car. This is around 17.5% of the total cost per car. How much should Tesla spend to reduce this last 17.5% of costs by a further 10% or $350? ($350 is 1% of the base selling price.) The answer is less than $182,000,000 ($350)(52) (10,000) to recover a portion of its investment. How much less? As long as there is a positive value from the investment, it might be worthwhile to consider these investments. But we don’t really know the specifics of what Tesla could do with that money.

Finally, both charts identify the capital cost discussed previously by Mr. Deepak Ahuja, Tesla’s Chief Financial Officer (CFO). When costs are shared by all three car models, the capital threshold cost of $2,000 is clearly pierced at a production rate of around 5,000 Model 3 cars. This, on the other hand, does not occur until much later (on the first chart) when there is no sharing of expenses.

One of the important lessons shown by both of these charts is the wisdom of Tesla’s management strategy in developing and maintaining production of more than one model before attempting a high-production car. This is shown in the steepness of the first chart when there is no sharing of expenses. At a production rate of 1,000 Model 3 cars, the cost is decreased by $27,220 ($57,862 – $30,642) thanks to sharing costs among all car models.

You may ask, “How?” The reason for this is that in the short term, labor, capital, and relevant operating expenses don’t increase. They stay “sticky” or fixed, and as more cars are produced, the sum of these costs is dispersed among more cars, resulting in smaller costs per car.

This also demonstrates good faith on the part of management in attempting to bring to market a car with a more competitive selling price. As the production rate of Model 3 increases, this will allow the company the possibility of producing the car at a profit.

Another lesson shown by the second chart is that the much discussed German cost of $28,000 is not static. In fact, as the number of produced Model 3 cars has increased, the unit cost of the car has decreased below this level, and it is on its way to a potential cost of $20,000 thanks to sharing expenses. However, as shown by the first chart, if only the Model 3’s costs are analyzed, then this level would’ve been reached only quite recently.

The material cost curve in both charts clearly shows the inflexibility of this category of costs (shown by the flatness of the curve), and the difficulty in bringing an affordable product to market. There are no economies of scale shown by this curve. Tesla’s procurement department will have to get very creative to decrease material costs in the future.

Now, I want to present the more or less (depending on your point of view) interesting aspect of how the material costs were gathered.

The battery chemistry costs are based on a report made in December 2002 by Argonne National Laboratory (ANL). This laboratory published a report titled Materials Cost Evaluation Report for High-Power Li-Ion HEV Batteries. This report estimated (among several things) that an NCA battery, LiNi0.8Co0.15Al0.05O2, could be manufactured for $398.86 (please see the appendix) with a capacity of 25-kWh. This is a cost of $15.95 per kWh.

Using more current material costs, the cost increases to $40.93 per kWh. These costs are, of course, not static, and they are subject to change. As noted above, this is the cost of the chemistry (the chemicals) of the battery. This basically means either basic elements or molecules (compounds). These might be bought in almost pure elementary form with few impurities, or they might be obtained as a mixture of molecules. These mixtures might be obtained either in liquid or solid form.

When these chemicals are obtained in pure elementary form, then the cost of the material is based only on their cost of mining and cleaning them of impurities. Some of these impurities might require expensive and laborious processes to separate them from the required element.

When they are obtained as a ready-to-use mixture, then their cost involves more complex processes. For instance, the cathode of this particular battery can be obtained by what is known as a spray-dry process (see appendix). There are other ways of obtaining this cathode, and the cost varies by the type of processes used.

As discussed before, material costs are inflexible (as more cars are produced, material cost do not necessarily decrease), and they tend to increase as an expense as time passes on. This can be seen with the chemistry of the battery itself between 2002 and 2018. Fuji Corporation made a detailed accounting in 2002 of the costs involved in preparing the cathode used in this battery for an annual production of 1,000 metric tons. These are shown on the appendix. These same costs (as shown in the appendix) were changed by me to reflect current costs. It is quite possible that these changes might not reflect the actual cost of the cathode from, for instance, simple changes in the Dollar—Yen ($ – ¥) exchange rate, differences in the current production processes used, or wages paid.

A feeble attempt was made on my part to attempt to estimate the cost of other battery components (including, among others, the anode, electrolyte, and electronics) in the Model 3 battery. However, due to lack of time (I spent almost a couple of months on this series of articles), I only guesstimated these. Anyone with a better understanding of these costs is welcome to contact me, and I will be glad to update these costs if they differ significantly from those in the table above.

The appendix presents the summary of these and other costs. In addition, labor, capital, energy, and financing (among other) costs are involved, and these are not shown in their entirety.

Some of the feeble attempts I took to obtain battery costs required converting the measuring units from meters or liters in the ANL report to kilograms as shown in the appendix.

The next step involved obtaining the internal volume of an empty battery cell. To do this, I had to adjust the internal volume of the battery cells due to wall thickness (as shown in the appendix), which limits the amount of material placed inside. I did these calculations for two Panasonic battery cell containers — the 18650 and the 21700.

The next step involved obtaining the empty weight of the battery containers. The appendix shows the calculations I made to obtain the weight of the container for the battery cells used in Panasonic’s 21700 design.

Several thousand of these battery cells go into each battery in a Model 3, and the weight of the battery cell can increase the weight of each car by several hundred pounds.

Honestly, I have never even held one of these battery cells in my hand. So, again, there is a lot of guessing involved. One of the largest sources of error is the weight of a single battery cell.

The next step was to aggregate the individual container weights. The weight of all of the empty aluminum containers adds between 95 and 141 pounds (see the appendix). Empty steel containers would have almost tripled this weight to between 274 and 406 pounds.

Initially, I was uncertain on whether the material of the container was made up of steel or aluminum. However, several different sources indicated that the material was aluminum.

The reason for the uncertainty had to do with safety. If one of these battery cells explodes, steel may be stronger than aluminum in containing the damage of an explosion. On the other hand, aluminum is a better temperature conductor than steel, and one of the ways to prevent an explosion is by controlling the temperature inside each battery cell.

I was aware that Tesla was attempting to decrease the amount of cobalt in its batteries due to costs. I would like to think that Tesla was doing this based on humanitarian reasons due to the horrible human cost of the children mining this element. However, decreasing the content of cobalt makes the battery less stable, and hence the uncertainty.

I am also aware of the active cooling system used by Tesla for thermal management. Therefore, I suppose Tesla settled this issue by controlling the temperature inside the battery cells and preventing the temperature from reaching critical levels for thermal runaway events — which added veracity to reports that Tesla used aluminum for the battery cells. I assumed, therefore, that it was easier to dissipate heat from a good heat conductor such as aluminum.

The next step involved getting the weight of the chemicals used. I have a lot of uncertainty in my calculations for a variety of reasons, and I am not satisfied with the results. I calculated a weight of 51.5 grams for the chemicals inside a battery cell container.

However, I obtained a weight of 14.9 grams per battery cell container (see the appendix). I cannot explain the difference, and, therefore, I am using the results based on ANL’s report given its authority and expertise on the subject.

Choosing the weight from ANL is going to create several problems in terms of weight, but I don’t have a better alternative. The numbers work out nicer by using the numbers I calculated of 51.5 grams for the chemicals, but I don’t have any evidence to support this. The combined weight of chemicals and containers (65.9 grams) is close to several internet sources. However, I still do not have a basis for going against the ANL report.

A problem that occurs as a result of using such a light weight for the chemicals is to make the weight of the battery enclosure much heavier. I have highlighted this weight in red in the list of materials (see appendix). Using this result, I obtained a weight for the chemicals of 98 to 145 pounds for either the standard or long-range battery (see appendix).

The next step was to use the the weight of all the containers and the chemicals inside of them to estimate the weight of the battery enclosure itself.

Regarding the stated cost per kWh, in the last shareholder’s meeting, CEO Elon Musk stated that Tesla will “definitely reach a cost of less than $100 per kWh at the pack level within two years.” My estimate of the cost of materials is below this cost (see appendix). However, keep in mind that my estimate does not include labor, capital, and an allocation of operating expenses, which would need to be added to the cost of materials.

The cost of the module container appears as $0 (see appendix) since the cost of the modules is not separated from the cost of the battery enclosure. The cost of the modules can be obtained by weighing and evaluating the type of materials used, the manufacturing processes, etc. I don’t have access to them, so I must gross the weight of the modules and the enclosure together.

As you can see, the chemistry cost portion is rather inflexible to cost changes. The only way out of this inflexibility would be to change the chemistry of the battery (subject to performance and safety issues) or change the technology of the chemistry itself (using solid state batteries versus semi-liquid type batteries). These changes require extensive research and development, which may take months, years, and in some cases decades.

However, in the immediate present, the cost of some of the components of the battery may be more amicable to cost reductions. For instance, the battery management system (BMS) cards are used to control the charge and battery state of each battery cell. Each module uses one of these BMS cards for this purpose. There are four modules in each car battery. My understanding is that some of these components are designed by Tesla’s engineers to attempt to control their cost. It is possible that the actual boards are built by an outside contractor.

As an aside note, one of the interesting results of this excercise was obtaining the energy in a single battery cell, 18.2 Wh (54,300 / 2,976). This is interesting by itself since there are many estimates of this number going around. I used this to estimate that the Gigafactory currently has the potential to produce around 46 GWh of battery cells. This means an incredibly automated production line or s (around 80-battery cells per second) and enough battery cells to produce 5,000 long-range Model 3s, 10,450 Powerwalls, 1,200 small Powerpacks, and 600 large Powerpacks per week.

I estimate that Tesla is very nearly at full capacity based on my own estimates. In other words, to reach the goal of producing beyond 5,000 Model 3 cars per week, Tesla needs to add additional battery cell lines, pack lines, module lines, and battery enclosure lines.

The next two quarters will be challenging for Tesla. The challenge will come not from trying to ramp initial production as in the near past, but the challenge will come from expanding production with current capacity until new equipment is installed.

After obtaining the weight and the cost of the battery, I moved on to the rest of the major components that I could think of. I found several different sources that guided me as well in obtaining a reasonable estimate for weight and cost of the remaining components.

Another source of uncertainty (more of it) had to do with the weight of the glass and the weight of the electric motors.

In any case, all of the above (plus the appendix) provides an estimate of the material costs of the Model 3.

These material costs taken together with the capital costs, the labor costs, and the relevant portion of operating expenses provides an initial framework to understand Tesla’s financial situation and the place that the Model 3 plays in it.

This analysis is different from my first article exploring a break-even analysis of Tesla. The major difference is that the break-even analysis was a “top-down” analysis and this analysis on material costs is an attempt at a “bottoms-up” analysis.

The following continues the same spreadsheet analysis, which I started creating in my series of previous articles on Tesla’s:

1. Break-even analysis
2. Cost of capital
3. Cost of labor

The table below shows six different scenarios with known and unknown information. The first scenario from the left presents the historical information for the first quarter of 2018. This scenario (as well as the rest) contains speculative numbers in terms of average selling prices (ASP). However, ASP are used to initially get a similar automotive segment revenue as the reported revenue for this segment. This means that, individually, the ASP might be off for the first scenario, but as a group they reflect actual sales.

The ASP are slightly lowered for the fourth scenario, which utilizes actual reported production numbers for the second quarter. The fifth scenario keeps the same ASP presented in the break-even article, and the final scenario presents the lowest known possible ASP.

In addition, it shows the cost of capital presented in the cost of capital article for all six
scenarios. This cost as discussed decreases with increasing production.

The next table below shows the labor costs presented in the cost of labor article for all six scenarios. This cost as discussed, also, decreases with increasing production.

This table, also, shows the cost of materials. I have changed these slightly from previous scenarios bit there are three cases where the cost of materials reflect the information discussed in this article. These are included for Model3 in the third, fourth, and sixth scenarios.

The table below shows the result of the previous assumptions. To begin, the difference between the first and second scenario explores the effect of decreasing the cost of materials. The third scenario also explores this, but it includes the cost of materials discussed in this article. At the same time, it places the cost of materials for Models S and X between the first and second scenario.

The costs for labor and capital do not change. Notice that that the resulting gain or loss is quite similar in all three cases.

In other words, if the cost of materials for Model 3 as discussed in this paper reflects in any way Tesla’s cost of materials for the Model 3, and then assuming the rest of the numbers are not too off from reflecting actual costs, then Model 3 has been contributing profits. Now, these profits can still not overcome operating expenses, but they are doing something positive to get Tesla on the road to profitability.

The fourth scenario makes use of Tesla’s reported car production numbers. As mentioned, it has lower ASP, and the cost of labor and capital reflect the higher production numbers by decreasing compared to the first three scenarios. The cost of materials reflects the discussion in this article for Model 3 while the cost of materials is lowered a bit further for Models S and X.

This decrease is justified in my opinion due to lower learning costs for these two models, reflecting higher productivity per employee. I don’t have hard numbers for this, but I simply had curiosity to see the result of this change.

As you can see, it is possible that Tesla may have managed to overcome the inertia of the bulk of operating expenses. This will probably not be reflected in the reported results of the second quarter. Why? Tesla in my opinion tried helping its customers by not triggering the federal tax credit quota, and despite producing more cars than in the first quarter, it did not deliver all of them.

Reported profits for the second quarter will, therefore, be depressed. However, this situation should switch in the third quarter as normal deliveries for the third quarter plus postponed deliveries from the second quarter are reported as sales. At the end of the day, these are accounting games. They do not reflect production problems. On the contrary, if you compare total cars produced in the second quarter versus the first quarter, they are almost 80% higher. These numbers in my opinion reflect Tesla’s coming of age in being able to do this while its actual production numbers continue to increase and improve.

The fifth scenario shows the break-even scenario discussed in the break-even article. The ASP is higher in this scenario than in all of the other scenarios. I have allowed this scenario to reflect knowledge gained in terms of costs for labor and capital, but it does not make use of the acquired knowledge of the cost of materials for Model 3. Rather, it maintains relatively high costs of materials to show the necessary numbers to break even.

If my numbers are plausible, then you will have noticed that the fourth scenario offers the possibility of achieving profitability at a lower production level than the break-even article concluded. Such is the power of information. The fourth scenario has lower ASP and lower production numbers than the fifth scenario, making it more difficult than the break-even case to reach a break even level.

However, the fourth scenario has lower material costs for Model S and Model X compared to the break-even case, and the fourth scenario leverages the gained material cost knowledge from this article for Model 3. As discussed some time ago, Tesla’s business model is highly sensitive to small changes in material costs. This is the real kicker in terms of achieving a break-even level and overcoming it to reach profitability despite a beginning handicap against it in terms of lower units produced, lower average selling prices, higher capital costs, and higher labor costs.

The sixth scenario shows a higher production level for Model 3, and it shows the lowest ASP. In addition, it makes use of the cost of materials of Model 3 discussed in this article and it maintains the cost of materials for Model S and X at the same level as the third scenario. Finally, this scenario leverages the knowledge learned in the labor costs article on the automotive segment’s overhead and indirect labor costs included in operating expenses. This is meant to show Tesla’s future potential ability to leverage the Model’s 3 manufacturing prowess by lowering selling prices while increasing profits.

I have attempted to introduce in each article a new cost element, and to maintain the explored costs from previous articles. In addition, I have presented in all of them a larger level of production than Tesla’s current level of car manufacturing. I have done this to present a production level which I think embodies the ability of the company to significantly reduce its average selling prices. In addition, it reflects on the company’s long-term plan of promoting a sustainable mode of transportation that is good for the environment.

Expensive cars were not the original goal of the company even though I am aware that this is the current state of affairs. Expensive cars were a means of leveraging financial strength to launch affordable electric vehicles, and this is unfortunately lost in today’s media. However, I, for one, continue to believe in the company’s mission statement, and the proof, as they say, is in the pudding.

The first car, the Tesla Roadster, was an expensive and low-volume car; the Model S car is also expensive, but it is manufactured in significantly higher quantities than the first model; the same goes for Model X; but the Model 3 is a higher-volume car which I hope I have managed to show has the potential for lower average selling prices.

These last three models share a common theme (absent on the first) of a systematic organized infrastructure to permit reproducing cars like with a cookie mold cutter. This means a reproducible product with as few differences between similar models as humanly possible while at the same time providing a very high-quality product.

Moreover, the significance of the Model 3 lies in the realm of possibilities for future lower priced cars. It serves as the means of developing the in-house know-how to allow Tesla to leverage manufacturing knowledge for even higher production electric cars at even lower average selling prices. This takes time, effort, large amounts of capital, and a dedicated, enthusiastic, motivated, and talented group of people willing to cooperate together for a common cause.

My last threee articles have provided a basis in understanding Tesla’s Model 3 costs, and I hope it has also provided some insight with regards to the current debate on electric car costs. Are electric cars viable? Are they affordable? Are they expensive to build? Are the batteries too expensive? And so on….

Keep in mind that my articles on Tesla are a thought experiment to attempt to recreate the company’s business model in text. This experiment does not reflect the actual reality of the company. At best, it is only an approximation of it. It contains numerous estimates, errors, and assumptions, and I urge you not to make investment decisions based on it.

Let me know what you think about it.

Sincerely,
Eric Kosak

Disclaimer: I have, sadly, no positions in the stock of this company. The above is not a recommendation for investment. If you seek investment advice, consult with a professional, and if you do invest, then do it wisely and without allowing the loss of any of your investments to injure you financially. Remember that money invested in the stock, bond, option, futures, and real estate markets can and might go to zero.

There is a list of sources attached as a separate set of appendices for this article. As you may see, several of the screenshots have underlined descriptions, and these are the links to the actual sources. However, the confounded CMS (content management system) of WorldPress makes a mess of formatting and links since it prefers screenshots to complex tables. This is the reason you cannot simply click on the original tables’ links, and in order to give credit where credit is due, I have placed these sources at the end.

In addition, some readers appear to prefer briefer articles, and to accommodate these readers, I have placed several parts of the article in the appendices. These parts for the most part show calculations and methods used.

Here are the appendices to this article: Appendices for Peeking behind Tesla’s cost of materials curtain

 
Don't want to miss a cleantech story? Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!
 

Have a tip for CleanTechnica, want to advertise, or want to suggest a guest for our CleanTech Talk podcast? Contact us here.
Advertisement
 
 

Written By

I live in a magical time of science and technology in which Merlin himself would be amused. I am just learning about it. A goal of mine for many years has been to learn and understand as much as I can about it, and, yet, everyday I find out something new. I hope I can share this. I believe in being independent; therefore, I enjoy reading on a variety of things. For instance, what can I do to improve the energy consumption of my house? How can I reduce the negative impact on my health due to the things I buy and do? Can I improve my life by choosing one type of energy over another? How can I repair a leak in my bathroom? What do I need to do to grow my own vegetables? To improve the world, my mom always says, we start at home. I have learned that to succeed at something I must fail many times, and with each failure I learn something, and I also try to live by the golden rule of doing unto others as I would like them to do unto me.

Comments

You May Also Like

Cars

You have probably been reading about the atmospheric river that has brought incredible rain, flooding and destruction to California this winter. That atmospheric river...

Cars

Following up on our stories about the best selling plugin vehicles in the world and the automotive brands that sell the most plugin electric...

Cars

After looking at the top selling plugin vehicle models of 2022 across the world, let’s look at the auto brands that sold the most...

Cars

Electric vehicles are becoming more popular with consumers, but even Tesla’s lineup took several years to become profitable. With the automaker’s most recent round...

Copyright © 2023 CleanTechnica. The content produced by this site is for entertainment purposes only. Opinions and comments published on this site may not be sanctioned by and do not necessarily represent the views of CleanTechnica, its owners, sponsors, affiliates, or subsidiaries.