With Tesla (NASDAQ;TSLA) touting July as the production start of its most important product, and the firm's stock at an all-time high, now is a good time to ask: Can Tesla make enough Model 3's fast enough?
How fast is fast enough? Tesla's latest 10-Q addresses future plans for Model 3 output. It forecasts a ramp of Model 3 production "to 5,000 vehicles per week at some point in 2017 and to 10,000 vehicles per week at some point in 2018."
And, last year Elon Musk forecast total 2018 production of 500,000 cars, of which 400,000 would be Model 3 cars, here.
Tesla has not revised that forecast. Indeed, the more recent forecast of 10,000 vehicles per week implies 520,000 cars per year (and, that is just Model 3 cars, which implies 600,000+ total cars in 2018).
I have a lot of familiarity with the Fremont plant from days of yore, and have an extensive background in automobile manufacturing. Based on everything I know, Tesla will not achieve its forecast of 5,000 Model 3 cars per week by December.
And it won't come close to its forecast of 400,000 Model 3 cars next year.
Let's examine why.
Right now, Fremont is Tesla's only factory, and it's far from "alien dreadnought"
Fremont is Tesla's only automobile factory. I agree with what Elon Musk said at Wednesday's shareholder meeting: Tesla is "bursting at the seams" at Fremont.
It's great that he's giving "serious consideration" to three more factories, and ultimately 10 globally or, indeed, as many as 20. But, right now, and for at least the next two years (because that's the minimum time required to plan and build an automobile factory, even assuming you have the capital available), Fremont is all Tesla has.
The Fremont plant, of course, is pushing 50-years-old. It used to be a GM plant with two assembly lines. After that, it was a GM-Toyota joint venture (NUMMI) that added metal stamping and plastics injection facilities adjacent to the body-paint-assembly facilities. Both Toyota's and GM branded vehicles were made under NUMMI.
Tesla has not expanded the factory since inheriting it, but has totally refurbished everything. And Tesla has done something else important: it hired Peter Hochholdinger away from Audi to serve as Vice President of Vehicle Production. My guess is that it's thanks to Hochholdinger that drive train and battery assembly, which were congesting the Fremont production line, were moved to Nevada.
All the same, Tesla has not yet achieved "alien dreadnought" in automobile manufacturing. While the process has become highly automated with the extensive use of robots, there are limits. Fremont is not much different from any other automobile manufacturing plant. If anything, it's less efficient than most.
In any auto assembly plant, you cannot just crank a big dial like in the movie "Gung Ho" and get more product. There are physical constraints relating to the length of the car, the maximum speed of the conveyor, and assembly stations for several thousand parts to be attached to the body, and part warehousing space proportional to production rate.
It's a dangerous misconception that Tesla has installed more robots and other high-tech manufacturing equipment than other manufacturers. There is no evidence of this. Robots have been used in assembly plants since the 1980s, and Tesla buys industry standard Kuka robots (offices near Detroit).
All the firms that sell manufacturing equipment to Tesla also sell to other automobile companies. Tesla's purchase this year of Grohman Engineering (Germany) and Rivera Tool and Die (Michigan) in 2015 illustrates just how closely Tesla must hew to traditional manufacturing technology. There simply are not any tricks here available only to Tesla.
First Bottleneck: The Paint Shop
Tesla has twice revamped its paint shop, first in 2013 and again last year. Last year's rebuild cost several hundred million dollars.
What was achieved? Seeking Alpha member A Plate of Waffles has studied the permitting application documents, and believes Tesla has installed a brand new paint line (but demolished the old one). The new line is capable of 25 units per hour, which translates to between 125,000 to 140,000 cars per year.
Currently Tesla's maximum capacity at Fremont is only 406,620 cars per year. That assumes 180,720 Model X and Model S cars and 225,900 Model 3 cars.
Already, we're a long way from Tesla's 2018 forecast of 400,000 Model 3s. With a projected 100,000 units Model S and Model X that implies a maximum output in CY 2018 of 325,900 vehicles.
Even assuming you could run two paint lines for 24 hours a day, 365 days, that would add up to only 438,000 cars. But right now there is only one paint line. And a paint line can't be run 24 hours a day.
Even 21 hours a day (which would yield 316,260 cars) would be difficult and expensive to achieve. So, I'm setting the theoretical limit on the paint line at 316,260 cars per year. Combined production of Model 3 and Model S/X combined would be around 416,260 for a year. This is within several thousand vehicles of the maximum NUMMI production.
But let's go ahead and look at the production line limitations.
Production Line Limitations
Because Tesla has not expanded the Fremont factory, it is reasonable to expect the two assembly lines have a nominal capacity similar to NUMMI. Here is the NUMMI capacity:
Truck Line (Model S and X): nominal capacity 180,720 jobs/year (48 jph)
Small Car Line (Model 3): nominal capacity 225,900 jobs/ year (60 jph)
From 2000 to 2009, NUMMI produced an average of 154,522 Tacomas on the Truck Line, less than the nominal capacity. Because the truck line is slower than the small car line, Tesla uses it for Model S/X production.
Tesla forecasts Model S/X production to be no more than 100,000 units in 2017. Presumably this is demand constrained.
The small car line is faster and is slated to assemble the Model 3. The 225,900 capacity figure is based on two shifts with straight time. The old UAW contract at Fremont allowed two 15-minute breaks per shift (labor law minimum is 10 minutes), so total production time is 15 hours per day.
The highest output for NUMMI was roughly 420,000 on both lines, which squares with the maximum output for each line with a little overtime (177,840 + 222,300 = 400,140 units with straight time).
Currently Toyota has 13 holidays, but a Glass Door search says Tesla has nine paid holidays. That allows 251 straight time production days, not counting any shutdowns for changeover. So, let's assume the maximum possible for Tesla, with straight time, is 225,900 Model 3 cars per year.
Can Tesla produce Model 3 cars on both lines (that is, using the old truck line for the Model S, the Model X, and the Model 3)? In theory, yes. But it would take considerable time and money to build the Model 3 on the truck line alongside the Model S/X. Typically you want to put cars on an assembly line that have similar footprints and share at least some parts. The Model 3 does not fit the bill on either count.
So, we're at 225,900 Model 3 cars per year. And that's assuming no line shutdowns for changeover, which is highly optimistic, if not unrealistic. It also assumes, of course, perfect execution by Tesla in parts procurement, and no problems on the production line despite the Model 3 being a bottom-up new design. Both those assumptions are, given Tesla's history, charitable.
But isn't a Tesla a simpler car to produce?
Some Tesla advocates are convinced Tesla's line will run faster because, they say, the Model 3 is more simple than anything built during the NUMMI days.
Color me unconvinced. There is very little chance the Model 3 will have fewer components than the Corolla or Vibe that were built on that line before. Remember, the engine and transmission counted as one part and usually took about one or two minutes out of hours of assembly.
Moreover, there is far more to laying out an assembly line than simply putting the assembly stations closer together. The best engineers from two of the largest companies in the world could only get about 225,000 cars per year off this line (straight time). Tesla, which is not nearly as efficient as either GM or Toyota, will not do better, and Elon Musk parking his sleeping bag at the end of the line won't change that fact.
What if Tesla uses three shifts?
Some commenters at Seeking Alpha like to calculate production capacity by using seven-day work weeks, each with 24-hour days. Sounds great, but it's completely unrealistic.
There are many reasons assembly plants don't run 24 hours per day. First and foremost is cost. To have three shifts, you need to subtract 30 minutes per shift for lunch and another 30 minutes per shift for legal breaks. This gives a maximum of 21 production hours.
The lunch need not be paid, but breaks must be. In practice, then, you wind up paying eight hours for seven hours of work because if you stiff workers out of their lunch, you are taking away 6.25% of their pay just because you want to run 24 hours. If Tesla were to attempt that, Musk would need to write an even more compelling letter on why not to unionize.
In addition to a minimum 6.25% wage increase, a three-shift assembly plant will probably have to shut down for more than the short lunch and work breaks. There are maintenance items such as cleaning spray paint equipment, changing weld tips, and myriad other tasks that cannot be done while production is running.
However, during these shut downs, the pay clock will continue running for 3,000 production workers, unlike a two-shift operation. So, add even more cost.
Also, in a three-shift operation, skilled trades typically will be scheduled at two shifts with daily overtime. Skilled trades are harder to staff and more expensive, so paying them eight hours for seven hours of work becomes difficult. So, add in two 10-hour staggered shifts for skilled trades maintenance. Electricians should be able to easily take in $100k per year with this scenario.
I could go on about the extra costs of a three-shift operation, but suffice to say that it is almost never done in the industry due to high costs and lower productivity and lower quality. One reason for three shifts is ill advised industry competition (Taurus and Camry production championship). And, let's remember, the Model 3 margins are already going to be skinny enough with a two-shift operation.
But what if Elon is determined, and thinks he can run a three-shift operation, and run on weekends, and make money? For starters, let's assume three shifts (21 hours production), work every Saturday, but not Sunday (UAW double time), and hold the number of company holidays at nine (otherwise pay double-time). Any more scheduled time would elicit visits from the U.S. Labor Department and the UAW organizers.
That gives us 303 production days * 21 hours/day = 6,363 hours. If we use the 400,000 jobs per year and divide by production hours, we get 62.86 jobs per hour, which is very close to the NUMMI run rate for this line.
So what, you say? Elon is going to Mars! Figuring out a simple three-shift schedule and some maintenance issues should be a piece of cake. Well the logistics will be hard enough and Tesla has not proven to be great at the logistics and planning end of this business.
Cost is a huge problem
A problem even bigger than logistics is cost. The example above shows that a 6% wage increase will be a starting point and eight hours on Saturday would legally need to be paid at time-and-a-half (add 8.33% to overall wages). If there is any slip in the schedule, then there is no choice but to run on Sunday.
And then there's the problem of shipping parts on Saturday and Sunday. When you extend into Sunday, the old NUMMI rules were a minimum of four hours scheduled and double-time pay. California requires double time after 12 hours in a day, or after the eighth hour on the seventh consecutive workday.
So, technically Tesla could get away without paying double time. But Tesla already has had one brush with a unionizing effort, and this is exactly the kind of operating scenario that will give an organizing effort the momentum it needs. Before, the union was simply promising increased safety and rational schedules, but with a three-shift schedule, organizers could promise 6%-10% pay raises, or more.
Consider the fact that there is not much difference in manufacturing costs between a large expensive car (e.g. the Model S) and a smaller car (Model 3). While revenue of the small car may be 60% less, the manufacturing costs may only drop 20%-25%. So, there will be margin compression. Several years ago in an interview with Betty Liu, Musk admitted that the Model 3 would be a lower margin car, maybe as low as 15% (said in a very low mumble). It is likely that with a three-shift operation the Model 3 would never be profitable.
Even Bob Cratchit got Christmas off
We have figured it is theoretically possible, if everything goes perfectly and we're willing to incur exorbitant labor costs, to build 400,000 cars per year on the Fremont line.
But 500,000? Well if you worked every day of the year for the legally maximum 21 hours, and used the 62,86 jobs per hour that got us to 400,000, you get annual production of 481,821. This is still short of 500,000 units per year, but logistically it's even crazier than the 400,000 unit per year scenario.
The labor costs per car would be high but the human toll would be higher. Even Scrooge gave Bob Cratchit the day off on Christmas. Such a schedule would guarantee unionization before the year was out.
At this point in the discussion, what happens in a normal auto company is a responsible accountant stands up and says, "Wait a minute - can't I build another green field assembly plant with much lower operating costs rather than this craz 365 days/year scheme?" And the accountant would be right and all the engineers would go about planning another assembly plant. At last week's shareholder meeting Musk mumbled some statements that indicate that may be the direction Tesla will be heading.
But that takes time. There's not enough time between now and the end of 2018 to build another Tesla plant. It would take about 24 months to build a new assembly plant from ground break to start-up. And it does not sound like Tesla even has a site selected, let alone plant design and contract letting required before ground break.
And it takes money. Right now, Tesla doesn't have the extra $2 billion or so required to build another assembly plant. Given the current state of the Model 3 product program for the second half of 2017 there will be production totals of 2,000 - 5,000, I predict. If Fremont cannot get two shifts at full production by year-end, 2018 production will be lower than maximum, a generous estimate might be 150,000 Model 3's in 2018 CY. This will be a significant revenue reduction from what many are projecting and will affect borrowing ability.
With its soaring share price, can it do another capital raise for that purpose? That may very well be what Musk had in mind at the shareholders meeting when he spoke of three or 10 or 20 more plants.
But that, dear readers, is another story for another day.
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.