Author's Note dated February 4, 2016: We have revised some of the figures based on updated calculations by commenter "embers". None of the revisions changes any conclusion. The author is solely responsible for failing to obtain the updated figures from embers before submitting the article for publication. embers also revised his figures regarding the additional cost that will result from producing cells in Nevada rather than in China. His revisions indicate that the added cost of cells, and hence, the incremental cost of each car, is even higher than is suggested in the article. We have left the figures as originally stated, to be conservative.
In this article, we discuss how the Nevada press has now noticed what we first pointed out - Tesla (NASDAQ:TSLA) has fallen far behind in the promises it made to the State of Nevada about employment and investment.
We then review a recent Seeking Alpha article arguing that Tesla's reduced level of Nevada spending is not troublesome because Tesla actually needs a factory that is far smaller than the one it has promised.
We discuss the logical absurdities with this article, and suggest that the scientific model that supposedly supports it is infected with that most common of spreadsheet problems - garbage in, garbage out.
We touch upon the company's announced plans to move Tesla Energy battery assembly operations to the Gigafactory, and question whether that makes financial sense.
We also address the much larger problem: If the scientific model shows anything about the Gigafactory, it shows that Tesla will cost itself about a half billion dollars more a year by locating the factory in Nevada instead of Asia.
We conclude by noting that Tesla will need to raise billions of dollars to complete the Gigafactory, and speculate whether Tesla or Panasonic (OTCPK:PCRFY) might be looking for a way to call the whole thing off. We take a brief look at their agreements, in order to suggest some possible escape hatches.
It's Official: Tesla's Nevada Forecasts Were Wildly Optimistic
It can no longer be denied: Tesla has halted work on the Gigafactory, with only a one-sixth (or smaller) "pilot" facility actually constructed.
On January 11, we published our own article suggesting that both Tesla and the State of Nevada had grossly overestimated how much the company would spend and how many people it would hire. Eight days later, the Reno Gazette-Journal confirmed what we had written. The Reno newspaper has now updated the story with estimates through year-end 2015.
Both Reno Gazette-Journal articles focus on the promises made by Tesla when it was bargaining with the State of Nevada for its record-setting $1.25 billion package of tax benefits and other incentives. The newspaper's latest story includes several graphs comparing promises with reality.
The first graph compares (1) the number of full-time jobs Tesla had forecast it would give to Nevadans by the end of 2015 with (2) the number that it has actually hired:
The second graph compares (1) the dollar amount that Tesla had forecast it would invest in Nevada by year-end 2015 with (2) the amount actually invested:
In each case, Tesla's achievements are less than 40% of what it forecast.
Mark Hibben: "Hey, Tesla, you made a math mistake. The Gigafactory is just fine at one-seventh size."
So, what's going on with the Gigafactory? Let's recall that Tesla has long justified the Gigafactory by claiming it would achieve significant economies of scale that would dramatically reduce battery costs, thereby turning the dream of a 200-mile range, $35,000 Model 3 car into reality.
Although it would appear to the State of Nevada that the Gigafactory is far behind schedule, Seeking Alpha Contributor Mark Hibben has offered another explanation: the one-sixth sized Gigafactory structure now in place is already large enough to accommodate a fully integrated battery factory that can manufacture 500,000 battery packs per year.
Mr. Hibben published his explanation in a January 11, 2016 article at Seeking Alpha: "Tesla: Panasonic Provides A Vote of Confidence In The Gigafactory".
He offered two pieces of evidence to support his explanation:
- First, Hibben pointed to the recent statement by Kazuhiro Tsuga, the president of Panasonic Corporation, that his firm would invest "up to $1.6 billion" in the Gigafactory.
- Second, Hibben argued that the one-sixth shrunken size of the Gigafactory is actually just right to build batteries for 500,000 cars per year. He claims to have done the math using a sophisticated modeling program, and argues that Tesla now understands this as well.
Hibben's arguments do not stand up to close examination.
For starters, why should a statement by Mr. Tsuga that his firm will invest "up to $1.6 billion" in the Gigafactory remove, as Hibben urges, "any doubts about whether Panasonic is truly committed to the project?"
A commitment to invest "up to" $1.6 billion in the Gigafactory doesn't put any minimum on the investment. Panasonic can meet that commitment by investing anything from $1 to $1.6 billion.
Although Hibben did not mention it, the Panasonic president also said, in the same interview, "We are sort of waiting on the demand from Tesla." That statement is consistent with what the (heavily redacted) agreements between Tesla and Panasonic show: Panasonic's obligations to install equipment in the Gigafactory depend on the level of demand Tesla experiences for its cars.
Hibben's second argument leans heavily on a sophisticated spreadsheet model developed by the prestigious Argonne National Laboratory (ANL), which is operated by the University of Chicago for the benefit of the United States Department of Energy.
The ANL's "Battery Performance and Cost Model" ("BatPaC" for short) was created to help design battery factories. Using data about the battery design and production rate, the model provides building area requirements and the costs of building and equipment.
Hibben used BatPaC to determine how large of a factory structure Tesla would need to produce 500,000 battery packs per year. The answer, he reported, is 100,000 square meters - which happens to be 50,000 square meters larger than the "pilot" factory now erected in Storey County, Nevada.
But wait a minute. If the BatPaC model was available in 2011 - a full three years before Tesla announced its Gigafactory - then why did Tesla set out to build a factory that was more than six times larger and far more expensive than the 100,000 square meters required?
Hibben appreciates that this is a problem. His solution is to speculate that Tesla's renderings showing some 930,000 square meters of area and costing $5 billion were merely "WAGs (Wild-Ass Guesses)" (Hibben's term) and were made "in ignorance" of the BatPaC software.
Pardon us if we find it difficult to imagine that Tesla was "in ignorance" of the ANL model. The company boasts some of the brightest minds in Silicon Valley. In setting out to design the largest battery factory ever built (indeed, by some measures the largest structure ever built), the company almost surely was aware of the most sophisticated battery modeling software in the world.
Even if we could be persuaded that Tesla's engineers were ignorant of the ANL model, we would have an even more difficult time believing that the engineers at Panasonic - one of the world's leading battery cell manufacturers - were also unaware.
Appreciating that ignorance of BatPaC is implausible, Hibben offers a fallback explanation. He concedes that Tesla and Panasonic may have known about the BatPaC model, but speculates that they simply couldn't believe the results it generated, so instead of using those results, they decided to go with their "Wild-Ass Guess".
And, oops, it turns out they were off by a factor of seven or so, which evidently they have now figured out.
So, that's how we design $5 billion factories, right? With Wild-Ass Guesses? If Hibben is correct, then I think Tesla investors should be much more fearful than even we have imagined about the planning of the Model X and Model 3.
The outlandish suppositions don't stop there. Hibben holds up his BatPaC calculation to show that Tesla needed only 100,000 square meters for a 35-GWh Gigafactory. However, what Tesla has now built is 150,000 square meters. That's not just a small miss. The company has overbuilt the floor space by more than nine American football fields.
So, according to Hibben, having designed a factory that was about seven times too large, and then later realized it could be smaller, Tesla somehow still managed to build one that is far too large.
But wait, there's one more problem here, and it's a huge one. In the very newspaper report that Hibben would have us believe demonstrates a big "vote of confidence" by Panasonic, the company president appears to be under the impression that the existing structure is merely the first phase of an eight-phase project.
So, does Hibben believe that Tesla has not yet broken the news to Panasonic that those other seven phases are unnecessary? (And, by the way, if the next seven phases are unnecessary, then doesn't Panasonic's "up to $1.6 billion" become "up to $200 million" instead?)
The BatPaC Model, Used Correctly, Requires The Much Larger Gigafactory
In spite of the logical absurdities in Hibben's argument, is it possible that he is, all the same, correct? That BatPaC actually demonstrates that Tesla's pilot-sized Gigafactory is not only large enough, but actually way too large?
Yes, it's possible inasmuch as anything is possible, but it's highly unlikely. Far more likely is that Hibben used the wrong inputs when he employed the BatPaC model.
In the "Comments" section to Hibben's article, one "embers" complimented Hibben on a "very nice piece", but raised a few questions. embers noted that BatPaC uses, as a sample calculation, a battery shape (prismatic or rectangular) that is different from the (cylindrical) Panasonic cells used by Tesla. The commenter asked Hibben to identify the inputs he had used in the "Parameters for the Finished Cell".
embers also noted that Tesla uses 74 cells per "brick" and six bricks per module, and noted that the ANL model used a different set of module pack assumptions.
The Tesla Model S uses 16 modules (or 7,104 cells) for an 85.67 kWh capacity. Assuming Tesla uses a similar module approach for the Model 3, did you assume 11 modules (for 58.90 kWh) or 12 modules (for 64.26 kWh) based on a 3.5 kWh/mile energy requirement on the UDDS driving cycle (row 173 in the Battery Design tab)?
Unfortunately, at least as of the date of this article, Hibben has not answered the questions posed by this commenter.
embers was kind enough to email us as well. He revealed his name and background. We have agreed to preserve his anonymity, but confirmed his identity.
The commenter has a BS in Aeronautical Engineering from California Polytechnic State University at San Luis Obispo, an MS in Aerospace Engineering from San Diego State University, and many years of experience in aerospace, missile, and automobile design.
For the past six years, the company of which embers is the founder and president has been working on the design of a new type of battery for electric vehicles. While it will be an alternative to the lithium-ion battery, there is no prospect that it will be completed in time to be used in the Model 3.
As embers wrote to us, "Obviously I am pro-electric vehicles, but I am anti-hype and fluff."
embers told us how, after reading the Hibben article, he ran his own simulation using the latest version (version 3) of the BatPaC spreadsheet to determine how large and expensive the Gigafactory would need to be to manufacture the 35 gigawatt-hours of battery packs required for 500,000 cars per year. As he warned (and as we have confirmed), the ANL simulation is very sophisticated, comprehensive, and not for the faint of heart.
Because the BatPaC simulation is designed for prismatic rather than cylindrical cells, embers adjusted some parameters (for example, cell dimensions and weights, modules) to account for the differences. He acknowledges that his results are imprecise. However, he is confident that they are much more accurate than the results obtained by Hibben.
embers calculates that to manufacture batteries for 500,000 cars, Tesla needs about 773,000 square meters of factory area. He figures another 20% for administration, services, and amenities, for a total of 892,000 square meters (which is within 4% of Tesla's Gigafactory plan).
As for cost, embers calculates that a total of $10.5 billion will be required, about two-thirds of which is equipment cost. He notes that the BatPaC simulation is quite granular, breaking down the square footage and capital equipment costs for each "process step".
Surprised at the large cost variance between what Tesla has forecast ($5 billion) and what he calculated ($10.5 billion), embers checked his capital cost estimate against a study by the National Institute of Standards and Technology on the manufacture of lithium-ion batteries.
According to the NIST, a good rule of thumb is that capital cost (to build the factory) will be $3-4 per cell to be produced each year (exclusive of the costs of research & development or equipment design).
Using the $3 lower bound, the resulting capital cost is still $9.6 billion. Reducing that figure by 30%, which is Tesla's forecast of the battery cost reductions it will achieve at the Gigafactory, the figure shrinks to $6.7 billion.
In view of the BatPaC model results and the NIST cross-check, embers questions whether perhaps Tesla has been unduly optimistic in estimating the total capital costs.
Like embers, we do not know what assumptions Mr. Hibben used. And even if Hibben were to share his assumptions, we would not be qualified to judge whether those of Hibben or embers are more reasonable.
However, unless we are the victims of an elaborate prank, it is clear that embers is a serious student of electric batteries, and his analysis would seem to challenge the work and conclusions of Hibben.
We have posted images of embers' spreadsheet results here. We wish the quality of our images were better, but the detail and density of the spreadsheet were daunting.
Knowledgeable readers who care to review embers' spreadsheet are, of course, encouraged to weigh in with their informed opinions. Mr. Hibben, of course, is invited as well. If anyone wishes for us to pass along a message to embers, perhaps asking to be put directly in contact with him, we will do so. He likely would be able to share more detail about his work.
And, of course, we are hopeful that embers will review the comments to this article and use that forum to address any questions that the commenters may have about his work.
Powerwall and Powerpack Assembly: More Gigafactory Absurdity
In Tesla's most recent shareholder letter, the company announced that, "[f]aced with growing demand for Powerpacks and Powerwalls, we have accelerated our plans to expand manufacturing capacity. In early Q4, we relocated production from Fremont to an automated assembly line at the Gigafactory."
Does such a move make sense? Note that Tesla is not making battery cells at the Gigafactory (and for good reason - it lacks the many "partners" that were supposed to have signed on a year ago). The company is merely assembling the cells into Powerwalls and Powerpacks at the Gigafactory.
So, after receiving shipments of the battery cells from Panasonic in California, Tesla will then transport the cells from California to Nevada for assembly, and will then transport the assembled Powerwalls and Powerpacks back to California for shipment around the continent and around the world.
Why incur this extra expense and logistical complexity, especially as there appears to be plenty of extra room in the Fremont factory? The only reason that occurs to us is to mollify state officials in Nevada, who are already embarrassed about how far Tesla has fallen short in its promises about both investment amount and employment levels.
Adding the unnecessary expense inevitably will cut into the Tesla Energy margins. In an increasingly commoditized business with many established competitors, investors might well scratch their heads at Tesla's decision.
The analysts on last November's conference call asked questions about the margins that Tesla Energy expects, but, disappointingly, none of them questioned moving the battery assembly operations to Nevada.
Was This Nevada Gigafactory Such A Good Idea In The First Place?
It's worth pausing a moment to consider the economics of the Gigafactory in light of the BatPaC model.
Using the parameters selected by embers, and assuming annual production of batteries for 400,000 Model 3 cars and 100,000 combined Model S and Model X cars (at 85 KWh each) - in other words, Tesla's 2020 forecast - approximately 20 million labor hours would be required to perform the cell manufacture work that is now performed in Asia.
The company has promised the State of Nevada that it (and its "partners" such as Panasonic) will pay its Nevada employees an average hourly wage of $26.12. Add another 10% for benefits and, in round numbers, Tesla's Gigafactory labor cost is $29 per hour.
This compares with approximately $2.00 per hour in China (and that figure may be on the high side, as iPhone workers are reported to make $1.50 per hour). In other words, a difference of about $27 per hour.
So, producing 35 gigawatt-hours of batteries per year in Nevada would result in an annual cost increase (relative to Asia) of about $540 million (the product of $27 and 20 million labor hours). That works out to an added cost of about $1,100 per Tesla car.
There is also the cost of transporting the battery packs from Nevada to California, where they would be mated with the cars. In addition, one must amortize the cost of the Gigafactory itself. Including those costs would significantly increase the $1,100 per car handicap.
Tesla has projected that it will achieve substantial savings from economies of scale. While that may well be true, those same economies of scale, presumably, could as easily be achieved in an Asian factory, but without the added labor costs.
In short, what the BatPaC model illustrates quite powerfully is that by building its Gigafactory in Nevada, Tesla may well have saddled itself with a huge cost disadvantage.
"There must be some way out of here", said the joker to the thief...
The Gigafactory shackles Tesla to a single cell manufacturer (Panasonic), a single battery architecture (cylindrical), and a single battery chemistry (lithium-ion).
In a world where battery research is proceeding rapidly, with technology breakthroughs frequently in the news, one can question the wisdom of the company's decision to vertically integrate its battery production. There may, indeed, be a good reason why companies such as Apple (NASDAQ:AAPL) (and General Motors (NYSE:GM)) outsource their battery manufacture.
Reflect, for a moment, on what would happen if in two or three years an enterprise such as the one founded by embers were to announce a battery breakthrough that would enable greater energy density, greater range, and lower costs. (And, no doubt, there are other engineers like embers out there, applying enormous brainpower to the problem of battery chemistry and design.)
Tesla would be saddled with outdated technology and would have no way to change course, except at a prohibitive expense.
Where we are, then, is with the company needing billions more to complete the Gigafactory at a time when it has severe operating losses and cash burn. And with Tesla apparently having failed to sign up any of the "partners" who, under its own schedule, were to have been on board a year ago. And with a Gigafactory plan that, examined closely, does not seem to make a lot of sense.
So, is it too late for Tesla to escape its Gigafactory shackles? Perhaps not.
Under paragraph 1.1(c) of the "General Terms and Conditions" that are part of the October 1, 2014 agreements between Tesla and Panasonic Corporation, both parties pledge to use reasonable efforts to cause Panasonic's sub-suppliers to establish "operations" at the Gigafactory. However, "Tesla acknowledges that [Panasonic] cannot guarantee that its sub-suppliers will agree to establish operations at the [Gigafactory]."
To our knowledge, no sub-suppliers have yet agreed to establish operations at the Gigafactory. So, perhaps paragraph 12.2 might come into play. It defines a "force majeure" event to include an event beyond the reasonable control of either Tesla or Panasonic that prevents them from performing their contractual obligations. If the problem cannot be cured within a certain time period (and here, the contract redactions obscure that period), then the affected party can terminate the agreements.
Tesla could argue that, despite all commercially reasonable efforts to-date, no sub-suppliers are likely to locate at the Gigafactory, thereby preventing the company from performing its cell purchase obligations and freeing it from the contractual bonds.
Of course, that same argument is open to Panasonic. And it may have a second escape route. Paragraph 11.1(a) of the General Conditions states that "[u]pon completion of construction of the Factory", the parties will negotiate in good faith to reach a "Factory Lease". However, if no Factory Lease is signed by a certain target date (again, hidden by redactions), then Panasonic can terminate the Gigafactory agreements with no liability to Tesla.
So, when is the "Factory" deemed to be complete? Is just the pilot size Gigafactory now in place sufficient? If not, how much more is required?
One might think the "analysts" who participate on the quarterly conference calls would be more curious about these questions, especially those affiliated with the investment banks that are underwriting more TSLA shares to credulous investors. Based on past conference calls, one would be wrong.
There is, of course, another way out for both Tesla and Panasonic: mutual agreement. They could simply decide to call the whole thing off and terminate their Gigafactory agreements.
Panasonic, which has spent only about one-sixth as much as Tesla in Nevada, would be free to remove its equipment and move it elsewhere.
Agreed termination would be more costly for Tesla. The company would have to abandon its $300 million investment (or sell it, likely at a loss) and make reparations to the State of Nevada for the tax benefits received to-date. However, that might be a small price to pay for escaping future capital expenditure and employment obligations that could swallow the firm.
One is reminded of the answer given by Benjamin Disraeli when asked the difference between a disaster and a catastrophe. Disraeli was a prominent English politician in the 19th century, and is remembered for his many Parliamentary clashes with his arch rival, William Gladstone. A disaster, said Disraeli, would be if my good friend and distinguished colleague, Mr. Gladstone, were to fall into the Thames river. And a catastrophe would be if someone were then to pull him out.
Similarly, it may be a disaster for Tesla to abandon the Gigafactory, but it may be a catastrophe if the company pushes forward to complete it.
A Note About Predictions
We're with Yogi Berra: "It's tough to make predictions, especially about the future." (Yes, we realize others may have said something like that before Yogi did, but he said it best.)
We've speculated quite a bit in this article, and done so using a cloudy crystal ball, so there's no doubt that we'll be off target in some instances.
After all, our forecasts and conclusions are starkly different from those of Goldman Sachs, Morgan Stanley, JPMorgan Chase, and Deutsche Bank - that is to say, they are contrary to the forecasts of Tesla's equity and debt underwriters, option writers, and lenders.
Then again, those same sell-side analysts, only a few years ago, exuberantly forecast that, this year, Tesla would be selling hundreds of thousands of power trains to other automobile manufacturers. (In fact, it will sell none.)
So, while we are confident our forecasts will prove inaccurate, and some of our speculation completely off-target, we are willing to say with even greater confidence that our forecasts will be less mistaken than those of the Tesla analysts from Goldman Sachs and Morgan Stanley.
A Note About the Other Contributors to this Article
No man is an island. No one person can hope to master and synthesize the multitudinous details of an enterprise as complex as Tesla Motors. I have used the first person plural ("we", "us", and "our") in this article because my views are in large measure shaped by the careful research and thoughtful analysis of others.
Regarding BatPaC, I obviously owe an immense debt of gratitude to Seeking Alpha member embers. I am also indebted to investor.gator for his analysis of the economics of Tesla's battery assembly.
I have also relied on the insights of some deeply knowledgeable posters at the Yahoo Tesla message board, chiefly temagami, metalapocalyptica, dynamic violence, and n0m0renancy.
As always, these posters are not responsible for my errors, nor do they necessarily agree with all points in my analysis.
A Note About How To Contact Montana Skeptic
I am on Twitter (Montana Skeptic@montanaskeptic1) and can be reached via email (Montana.firstname.lastname@example.org).
Disclosure: I am/we are short TSLA VIA LONG-DATED OPTIONS.
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.
Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.