Last week Tesla (NASDAQ:TSLA) formally announced the Gigafactory, an audacious plan to boost production of Tesla cars to 500,000/year while reducing the cost of the battery pack by 30%. When completed in 2020, it will dramatically increase the worldwide output of Lithium ion batteries to about the total worldwide output of today and have cost $4-5 billion dollars. I'm a Tesla fan, and even I find these numbers a little scary.
Probability and Consequence
The risks of such a venture are very real. Tesla will go into debt, to the tune of $1.6-$1.84 billion dollars in convertible notes, which will come due around the time the Gigafactory is planned to reach full production. How do investors and analysts assess the risk? Clearly the risk of failure of the Gigafactory is substantial, but is not certain.
For this article, I decided to adapt the approach I learned while working in the aerospace industry. Such companies periodically drag their senior technical staff and managers into Risk Management Class where they learn the company's preferred approach to Risk Assessment and Mitigation. There isn't really a universal approach to risk management, but what I learned seems appropriate here.
The basic approach is to divide the risk assessment into two problems:
1) For a given adverse event or outcome, what's the probability of occurrence of the event.
2) Assuming that the adverse event occurs, what's the consequence or impact of the event.
It's the combination of the two that is used to determine risk. Suppose that a company embarks on a venture that might result in the company becoming insolvent (a severe consequence). This might still be considered low risk if the probability of occurrence is small. Conversely, an outcome might be considered low risk even if the probability of occurrence is high because the consequence has little impact to the company.
John Petersen's recent article on the Gigafactory did a credible job of assessing the consequence portion of the Gigafactory risk: If the venture fails, Tesla is saddled with a mound of debt it can't repay and creditors who don't want to convert it to Tesla shares. Tesla goes into receivership, and Tesla shareholders are left holding the bag. One has to be a little careful about defining what constitutes failure, but for the sake of argument, I'll assume that Petersen has correctly assessed the consequence portion of Gigafactory risk.
What I didn't see in his article is an assessment of probability of occurrence. That's the missing piece of the Gigafactory risk assessment I would like to try to fill in. This is only a first cut, and this is something I'll be reviewing and updating in the coming year(s).
Where Did the Numbers Come From?
What disappointed me about the Gigafactory press release was the sheer paucity of information. We're told that through economies of scale and other things vaguely alluded to, the Gigafactory will achieve a cost reduction of battery packs of 30% and that it will produce 500,000 packs per year when it reaches full production in 2020. Tesla offers no substantiation for either number. We're told that the Gigafactory will employ renewable energy sources such as wind and solar, but we aren't told how large a contribution these will make to the Gigafactory's power usage. We don't even know if the $5 billion price tag includes the solar and wind generation or if these will cost extra. Am I belaboring the obvious to point out that too little information can be worse than none at all, from a PR perspective?
After a little digging, I was able to come up with a credible source for the 30% cost reduction estimate. In 2012, the Chemical Sciences and Engineering Division of Argonne National Laboratory (ANL) published "Modeling the Performance and Cost of Lithium-Ion Batteries for Electric-Drive Vehicles". Accompanying the paper is the Excel spreadsheet model they used, which anyone can download.
The paper describes the design and construction of the battery pack, as well as the design of a hypothetical battery factory to come on line in (you guessed it) 2020. Through optimization of the design of the battery cells, as well as the design and layout of the factory, the ANL model projects that the cost in $/kWh of battery capacity will be about $155/kWh, almost exactly 60% of the usual $260/kWh assumed for Tesla's current battery packs. The ANL model assumes a production rate of 100,000 packs per year.
Of course, there's always a lot of uncertainty built into these models, which ANL acknowledges and tries to quantify, so the range is roughly 20%-50% cost reduction.
ANL didn't assume any dramatic breakthroughs in battery technology (although their baseline battery design is not cylindrical), so this makes the estimated cost reduction conservative in my view. In fact, battery innovations such as I referred to in my previous Tesla article have the potential to reduce battery pack costs much more dramatically by increasing cell capacity by a factor of 2-4 and thereby reducing cell count proportionately.
Running with the ANL number for now was the right thing for Tesla to do, given uncertainties about future battery technology availability. The ANL number represents a high confidence achievable baseline. The actual battery technology that is used remains to be seen, but it probably won't be the Panasonic 18650 battery currently used by Tesla.
As for the 500,000 pack annual production capacity, this is just the maximum production capacity Tesla expects from the Fremont, CA assembly plant. The ANL report makes clear that Tesla doesn't have to build 500,000 packs in order to achieve the 30% cost reduction.
And it's not like Tesla builds the Gigafactory, then flips a switch and suddenly starts producing 500,000 packs per year. The Gigafactory is planned to ramp up production gradually over a course of three years starting in 2017. A first year production of 50,000-100,000 packs is probably reasonable, so the cost savings should kick in fairly early in the life of the Gigafactory.
By ramping up production gradually, Tesla can avoid overextending itself by building capacity it doesn't need. This could be especially important for the planned wind and solar arrays, which could be very expensive, but also needed to avoid overloading local electrical generating capacity.
100,000 packs is about triple Tesla's planned vehicle production for 2014. There's certainly room for debate about whether demand will be at that level in the 2017-2020 time frame. I consider it reasonable given that pack production will be supporting three Tesla models, including the lower cost model.
The ramp up to the 100,000 production level in order to achieve the 30% savings also makes clear that the success criterion is not quite as hard as the 500,000 production level implies. The sky doesn't fall in on Tesla if it doesn't sell 500,000 cars a year. This production level is a goal for Tesla to grow into, and this growth process may extend well beyond 2020.
Subjective Risk Assessment
Although analysts try to be objective about it, risk assessment is invariably a subjective process based on experience and judgment. My subjective assessment of probability in this case is that the probability that Tesla will not be able to sell the minimum production of 100,000 cars per year starting in ~2018 is low. That still makes the overall risk moderate, in my view, based on severity of consequence. Musk is betting the farm on the Gigafactory, putting the entire company somewhat at risk. But then, disruption is never risk-free.
Disclosure: I 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.