Examining Tesla Model 3 Production Goals – Are Targets Even Feasible?

Tesla Model 3


Tesla Model 3

Tesla Model 3

Tesla CEO Elon Musk notes a potential of factory production speed improvement by a factor of 10. Where does this put Model 3 production, and at what point might Tesla achieve this monumentally lofty goal?

The real answer may be “never”, that is until Tesla has more than a single factory … meaning several factories (Ford’s huge list of combined factories globally can output about 16 vehicles every minute). Yes, the former NUMMI (GM and Toyota joint venture) factory is quite capable, and Toyota is well-known for its lean manufacturing and top-notch automation. Additionally, Tesla is continually bringing in newer, more state-of-the-art automation. But, to put it in perspective, the factory under NUMMI’s control made about 385,000 vehicles per year. In 2006, its record year, NUMMI polished off a whopping 428,633 vehicles, with a significant amount of overtime.

Tesla Model 3

Tesla Model 3

Fast forward to Musk’s estimates. Tesla is said to be able to produce about 500,000 vehicles per year by the end of next year (and that’s not with a second factory). Recently, Musk said that now Tesla is looking at being able to hit 700,000 per year at some point (likely with a second factory, or at least the Tesla Gigafactory chipping in heavily).

Musk cites volume, velocity, and density as the three main factors affecting overall production.

According to Musk, there’s still significant potential to expand the facilities in Fremont for greater volume. According to autoblog, this likely means expanding upward, as there is much space in the factory above current structures and robots, to build taller machines.

Musk explained that the world’s top auto factories have a vehicle exit rate of 25 seconds (keep in mind this is not 24/7  and 365 production, nor is it feasible under everyday circumstances). This means that when the line is running, a finished car comes off every 25 seconds, and the line moves at about 0.2 meters per second (Musk compares this speed to that of a turtle). He believes that Tesla can do much better. Velocity exceeding a car every 25 seconds?

In terms of density, Musk refers to the leaps and bounds that have been made with regards to computer chip density. looking at how close together and multiply layered parts are in a computer chip, and comparing it to the Fremont factory, shows that the factory could be significantly more dense. Musk says the density of the Tesla factory is about 2-3 percent and could be increased by 20-30 percent more over time.

Tesla has delivered ~23,000 vehicles thus far this year in the US through July, 47,000 globally in the first half. Current Model 3 orders are just south of 500,000, with an additional 1,800 orders per day at this point. Tesla delivered ~30 Model 3s so far, and it appears as if maybe a handful more have made their way into owner’s hands. Musk’s S-Curve ups this to 1,500 cars per week in Q3, followed by 5,000 per week by the end of the year, culminating at 10,000 per week sometime in 2018.

A factor of 10, when applied to NUMMI’s 385,000-unit average, puts Tesla at 3,850,000 cars per year. Hmm … this means 74,038 cars per week. That’s about ~10,500 cars per day. There are 1,440 minutes in a day, so if Tesla is building 24/7, that’s pretty darn close to a car every second. Perhaps Musk meant 10 percent? But wait, that’s only 388,850 cars per year, and he said 500,000. Even if we use NUMMI’s record year, Tesla would be looking at ~471,000 per year. As you can see, none of the math really adds up.

Musk recently stated during the Q2 earnings call that Tesla has the potential to make 5,000 cars per week sometime in December of 2017. This, while a bit more feasible, is still a car every 3.5 minutes. If Ford could build a Model T every 24 seconds, then why not right?

The premise is still that Tesla has a long, long way to go to get even close to such numbers. It will surely be a sight to see as the automaker moves forward.

Source: autoblog

Categories: Tesla

Tags: , , , , ,

Leave a Reply

47 Comments on "Examining Tesla Model 3 Production Goals – Are Targets Even Feasible?"

newest oldest most voted

This is going to be a fun one!

I’ll start: Why does everyone keep throwing around that net 1,800 orders per day increase? It may have been accurate for 5-7 days after the reveal, but it’s unlikely to be accurate now.

Also, I don’t think GM & Toyota left behind any equipment or production lines when they left. So, it really doesn’t matter how good Toyota is at automation & lean operations.


Yes, no way that 1,800 per day rate was sustained much past the few days following the delivery party.

To say the net reservation rate is still at that rate would be like expecting a movie that made $100 million over a 3 day weekend to continue to make that much money the rest of its theater run. No way, jose.

John Ray

Nice analogy. +1


That is actually a very poor analogy. The lifespan of a typical movie in the movie theaters is measured in weeks. After that point, the demand for movie tickets ends.

The typical lifespan of a generation of a model of car is typically 3-7 years. So your comparison is off by two full orders of magnitude.

It still makes me laugh when folks complain endlessly about Tesla’s reservations, in an industry that typically has NO reservations for purchases for volume production cars.

Tesla sets an all time world record for the number of reservations for ANY product ever put into production, and yet all folks like you can to is desperately and frantically search for some little tidbit of something to complain about.

How are those reservations stacking up for any other upcoming (or even current) volume EVs by any other car maker? Which is getting 1,800/day net new reservations on ANY day?


“I don’t think GM & Toyota left behind any equipment or production lines when they left.”

It turns out that this is wrong to a surprising degree, altho you’re certainly correct to say that the 1800 TM3 orders per day couldn’t possibly be sustainable.

Quoting from Wikipedia:

NUMMI auctioned off the press lines, robots and other equipment to Toyota’s other US factories while Tesla purchased over $17 million of manufacturing equipment and spare parts in 2011, at significant discounts compared to new equipment.


Alonso Perez

They left pretty much everything large behind. Certainly all the heavy stuff, like the presses, ceiling cranes, etc. It costs too much money to disassemble and remove equipment like that

Most people don’t understand what a tremendous coup getting that factory for $50 million was. It’s crazy cheap. Impossible to build for less than 20 times that. To Toyota it was just depreciated asset and cost center. It’s crazy how accountants see the world.


According to Wikipedia, the presses for shaping body parts were auctioned off.

And anyway, Tesla is (or was, until the Model 3) using aluminum rather than steel for its car bodies, so it needed new presses — or at least new dies for those presses — anyway.

Yeah, those giant aluminum sheet presses seen in many photos of the Fremont plant… those were bought used by Tesla and shipped in.


I agree it was a crazy good deal. In several ways Tesla owes its existence to the stupidity and shortsightedness of GM and Toyota.

I see now that some equipment was included with the building. But my point was that the article makes it sound like Tesla moved in and just fired up an optimized Toyota assembly line to start making their cars. That’s pretty far from reality.


Ummm…do you know how much environmental damage a manufacturing plant with a paint facility can cause?
Toyota got away with not having to pay for the clean up for an older factory that would take millions to renovate to modern standards + got Tesla stock which they sold for at least a few million dollars of profit.
Honestly, whether Tesla succeeds or not, Toyota got away with smelling like roses while making themselves a huge profit + PR gains.


Possibly a lot. Was that actually the case at NUMMI? It wasn’t that old of a plant, so I doubt there was significant contamination. Not like it was a 50 year old chemical plant.

In any case, GM and Toyota killing the 90s EV mandates, and then abandoning EV development, left the door open for Tesla. Selling a great factory to Tesla for a song has enabled them to possibly succeed.

Ad van der Meer

Comparing NUMMI with the Tesla factory is nonsense. Tesla ripped out most of the equipment and replaced it with state of the art machinery and robots. They are also still adding structures to the complex.
I am not saying he can pull off 3.7 million cars per year, but using NUMMI as a starting point can not result in a good analysis.


Yes, production of 500,000 per year at the Fremont facility seems quite feasible. In addition to the points you made about the modernization of the factory, which came close to 500,000 even in its original set-up, Tesla will be building the battery packs and the power trains at the Nevada Gigafactory, and the remainder of the production process, for electric vehicles, is far less complex than the ICE cars that GM and Toyota made previously (fewer compoents).

So I don’t understand what Musk was talking about regarding the speed of the production line (x10), but you don’t have to believe that to believe that Tesla can produce 500,000 EVs at Fremont, assisted by the Gigafactory.

In addition, Musk has already said that the Model Y, which could get Tesla’s combined production into the low millions, will require another factory. So I don’t understand what point this article thinks it is trying to make about the overall feasibility of Tesla’s production plans.

Alonso Perez

It can, for many reasons. Think throughput. Just the logistics of parts and materials coming in and cars going out. Trains come in and must be processed at the rail yard, and other stuff is shipped in by truck. This takes space and time. 10,000 cars per week is at least 20,000 tons of materials and parts per week, without including factory consumables.


Right, that is most definitely another bottleneck which would likely prevent the Fremont plant from reaching 10x its original optimal production capacity. The sheer volume of parts which would have to be shipped into the plant every day, and distributed to the appropriate holding bins and/or the various assembly stations, would be staggering. There are only so many loading docks, and they can only unload trucks and move things in at a certain rate of speed, even if they have tractor-trailer rigs waiting in line at loading docks working 24/7.

Building more loading docks means less room for other things inside the plant. Tesla can’t just keep cramming more and more things into the same space, no matter what Elon’s basic physics equations say.


In the short run, it doesn’t matter if he can produce 500K a year. What matters is can he keep his reservation holders happy? After that is met, will demand be there for anything like 500K? If there is, he won’t be able to see over the pile of investor money to tell. 🙂

Anti-Lord Kelvin

Steven, when you wrote this:”Tesla has delivered ~23,000 vehicles thus far this year. The automaker delivered ~47,000 over the course of 2016″, you should specify that are figures for US deliveries alone, because, so far, Tesla delivered more than 76k cars last year, and 47k up to June 30th.

Jay Cole


Yes that is mixed up, should be 23,000 in the US, 47,000 for first half of this year 2017. Will grab that now, thanks for pointing it out!


This year 100-120k. I don’t expext Tesla to make more than 400k in 2018. Of course I want them to produce 700k in 2018 but…


Once tax credits are gone for Tesla per the current rules, demand for M3 in US will drop a lot. China may be the biggest market ultimately but Tesla will have to (at least) assemble there.

John Ray

Thanks for the article and bringing a much needed dose of reality to the situation.


Battery packs and motors are made at the Giga factory. I don’t know how much drivetrain assembly went on at the factory in the Toyota days, but M3 is probably more just bolting the subassemblies together. I would guess it is easier to make an EV than an ICE. Skateboard design and no plumber’s nightmare for emissions, fluids, transmission… Add in 10 years of advances in robotic assembly and I can’t image why Tesla can’t get to 500-750,000 a year.

I would guess moving the cars off site and distributing to customers is as big a hurdle. Maybe they will just switch on the auto-pilot and have them self deliver. I can image getting a text from your car “I just rolled off the assembly line and MapQuest estimated arrival in your driveway Tuesday at 4:03 pm. I will update my ETA accordingly.”


I’m not sure how I would feel about my new car telling me it was stocking me down? Sort of creepy.

But as long as they don’t use Apple Maps I’m sure they could make something like that work with a trivial amount of programming.

(sideburn –> Apple Maps)


While that would be cool, I think we’re several decades from that happening.


I have no idea if they can pull it off. We do know that the Fremont factory has done it in the past. And Tesla has hired a lot of production people from other automakers.

But it is going to be production hell!

Good luck, Tesla people! We are rooting for you!

Did reddit’s violentacrez slip a story in? Because it is like insideev’s hasn’t been reading this really cool site that has a ton of information about EV’s. That site is called insideevs.com…. You can’t compare NUMMI manufacturing numbers to Tesla Fremont factory numbers because Tesla keeps expanding it, doubling it in size: http://insideevs.com/teslas-new-fremont-factory-plan-double-size/ So effectively Tesla is building an additional NUMMI factory right next to the existing NUMMI factory. They effectively will already have 2 NUMMI factories. The sales numbers are US only as the other poster noted. No mention of the Solyndra site they leased either. http://insideevs.com/tesla-signs-lease-deal-500000-square-foot-ex-solyndra-site/ Can Tesla build more cars on twice the Fremont footprint + Gigafactory? Do I even have to state the obvious? That’s before even considering the improvements in robotics since the 1980’s/1990’s and now, or the future “alien dreadnaught V1.0” that is scheduled for mid-2018. Are cutting edge modern robotics more capable of building cars faster than the old robotics that were installed 20-30 years ago in NUMMI? Yes. Absolutely. So much more info about Tesla’s plans that can be found right here on insideevs that it hurts. This is a bad sign with this following the heels of the “Why Tesla Model… Read more »

Yes, the suggestion that Tesla can’t build more than 500,000 cars at the Fremont assembly plant just because the NUMMI plant wasn’t designed to produce that many, is downright silly.

But then, Elon’s claim that Tesla can produce 10x the cars from the same space, if not equally silly, is at least rather questionable.

I think the reality will prove to be somewhere in the middle. Even 3x the maximum NUMMI production would IMHO be an impressive achievement by Tesla.

That’s my point though. Since they have doubled the space, they don’t need 10x. They can just increase the number of lines they are running. They don’t even need 2x to hit Model S + X + 3 target volumes, even if they didn’t increase the number of lines at all and just improved the existing lines. (Alien dreadnaught Version 1.0) So the question of whether Tesla can do 10x is completely irrelevant to the question of whether they can produce their stated target volume in a plant that is twice the size as it was when it was the old NUMMI plant. _______________________________ This is making up Fear, Uncertainty, and Doubt where there simply aren’t any supporting facts to suggest that Tesla CANNOT do what they have stated they will be doing. Fear that they might not do something is irrelevant. Uncertainty that they can do something is irrelevant. Doubt that they can do something is irrelevant. That’s just FUD. Now if somebody had a source of actual information (not just idle speculation) that there was some hard limit they were going to hit. Like if there are only 200,000 Headlight trees left in the world, and once they… Read more »

“A factor of 10, when applied to NUMMI’s 385,000-unit average, puts Tesla at 3,850,000 cars per year. Hmm … this means 74,038 cars per week. That’s about ~10,500 cars per day. There are 1,440 minutes in a day, so if Tesla is building 24/7, that’s pretty darn close to a car every second.”

In an article about math being off, this is a whopper. A car every second is 86,400/day, if we are talking 24/7 production at that rate. 10,500 cars/day is a car every 8.23 seconds. Quite a difference.


To be fair, he did say: “As you can see, none of the math really adds up.”



Yeah, I was going to say that. 8.2 seconds could be doable, and the cars would drive themselves off the end of the production line to their parking spot.

Ocean Railroader

I think in terms of automation of building cars it might have reached it’s peak considering you rarely see any humans in scenes building cars.

I think Elon’s claim that an auto assembly line can be sped up by 10x… in fact, in his earliest statement along that line he said 10x-20x… is aspirational rather than practical. Sure, if you look at an episode of the documentary series “How It’s Made”, you can see touchless light industry production lines working at eye-blurring speed. In theory, automotive assembly lines may be able to move at that speed. However… In theory, there is no difference between theory and practice. But in practice, there is. As they say: “A chain is only as strong as its weakest link.” And an assembly line is only as fast as its slowest bottleneck. If Tesla makes all the assembly stations in the Fremont production plant touchless, and speeds them up to as fast as the machines can handle, that might well cause car bodies to move down the line at 10x their previous speed. But what happens when they get to a process which, due to physical constraints, cannot be practically sped up? Paint sprayers only work so fast; trying to speed them up would be counter-productive. Now, that’s not to say there’s no way to speed up that process; perhaps… Read more »
John Ray

Well thought out. Industrial engineers study just this sort of thing. Once you remove one bottleneck, the next slowest process becomes the bottleneck and so on. Assembly lines can be optimized based on the cycle time of each process, but there is a practical limit.

The article says: “In terms of density, Musk refers to the leaps and bounds that have been made with regards to computer chip density. looking at how close together and multiply layered parts are in a computer chip, and comparing it to the Fremont factory, shows that the factory could be significantly more dense. Musk says the density of the Tesla factory is about 2-3 percent and could be increased by 20-30 percent more over time.” Aaaaand this is one of the reasons why I say Musk’s vision here is aspirational rather than practical. Computer chips have been successfully shrunk by multiple orders of magnitude because teeny-tiny transistors and circuits work just as well as larger ones. Contrariwise, auto bodies can’t be shrunk, and neither can the robotic arms that are used for touchless auto assembly. Furthermore, on a practical basis, if you try to cram machines in too tightly, “cheek by jowl” as they say, then that makes it much more difficult to make any changes to the assembly line. For flexibility, not to mention maintenance, there needs to be some space maintained between lines and between machines on the line. Try to pack the machines and lines together… Read more »
John Ray

Agreed. And, while robots may be smarter today, I bet they are only marginally faster. There is a physical limit to how fast a robot can move based on mass, inertia, etc. All the modern computing power in the world cannot overcome that. And some tasks are going to take what they take – laying down a weld for instance.



Those light industry touchless assembly machines seen on “How It’s Made” can and do work at eye-blurring speed because they’re handling objects which are small and don’t weigh much.

Compare to the giant robotic arm, made especially for Tesla, that can lift and rotate an entire car body, as seen in the video linked below.

Speed that up by 10x? Hmmmm… very likely not! The cube-square law holds: Scaling something up by X amount means X^2 the surface area and X^3 times the mass (and weight). Moving very heavy objects around that fast is both dangerous and very stressful on both the car bodies and the robotic arms. They just weigh too much to be moved that fast, period. Even if it was possible, moving a car body at 10x the speed might cause it to buckle if the movement started or stopped too fast.


You do realize that the video you linked to was just them having fun showing off their cool robot, like in the movie “Dave”, right?

That’s not actually a motion that Tesla would have to speed up 10X on their factory floor.

But more to the point, Tesla doesn’t need 10X to hit their targets. They don’t even need 2x. So debating whether they can do 10x is like debating how many angels can dance on the point of a needle, or the head of a pin.

It might be a fun mental exercise, but it is completely irrelevant to whether Tesla can hit their target numbers.


(PS — the answer is 88 angels.)


Well they got 1.8 billion in their over prescribed bond offering, so they won’t be hamstrung by lack of cash anytime soon.

If they do over 40k by the end of the year, I would be surprised. So surprise me.

So much for no demand for their bonds. What did the yield end up being? _________________ In the conference call, they set the official Q3 2017 (July+Aug+Sept) Model 3 sales/delivery guidance at 1500 (not to be confused with production target, which will be higher). They also stuck to the 5K/wk by the end of Q4 2017 target, while refusing to commit to a specific ramp-up rate. So if they start Q4 in the first week of Oct at 375/wk and end at 5,000/wk, Tesla’s theoretical range of total deliveries in 2017 would be between 11K to ~60K. 40K would indeed be a very strong showing, and would require them to push through the steep part of the S curve by November. To put it into perspective, anything over 30K would be a massive record-breaking showing for 2 quarters of sales (all of which would be US sales). In fact, 30K in 6 months would be TWICE the sales rate as any other EV has ever had in the US (2011-2016 data here: http://insideevs.com/monthly-plug-in-sales-scorecard/ ). It has taken a full year of sales for any other EV to get close (or to) 30K in the US. I think everybody should be… Read more »

I realize I messed up the math (my bad), but I still think that anything over 30K would be a success.


In order to produce 500,000 cars a year at the Freemont factory I heard that Elon Musk is going to expand the factory by building an underground facility using the Boring company. So Model 3 is gone to be assembled at 100 feet below street level in huge tunnels. Just kidding.


“Tesla is said to be able to produce about 500,000 vehicles per year by the end of next year”

Steve, I think there is an error in your data. This is the production goal for the Model 3 only, excluding Model S & X.

Fat Bloke on Tour

Interesting to get the correct numbers.
My understanding is that the 500K target is for the whole company.

Tesla Luxury and Model 3.

However sources differ on this this.

Unfortunately Tesla has made contradictory statements about that 500,000 number too. They started out with a very clear message of 500K Model S + X + 3 (+whatever else they might build) by 2020. Then they put in their official SEC papers that this was changed to a RATE of 500K S+X+3 sales by the end of 2018 (this implied that they could produce fewer units for the year, but by the end of the year they would be building 40K S+X+3’s per month.) Then they dropped the “rate” qualifier. I initially assumed this was just sloppy talk, and that they didn’t really mean they would build 500K S+X+3’s in total in 2018. Now I don’t know. The official break down of targets is that Elon says he “guarantees” they will end 2018 building 40K cars/mo, and they “expect” to start the year at 20K/mo. With 100K Model S+X sales, they certainly could build 500K cars in 2018 if they hit their targets. But to make it worse, there has been some more talk where it sounds like they are saying 500K Model 3’s a year. in 2018? in 2019? Hard to know. They aren’t doing a good job of… Read more »
Fat Bloke on Tour

Still not sure that the BEV community understands how vehicles are built — it might be new stuff to an SV billionaire wearing L plates but it is pretty basic stuff to Big Auto.

No magic involved just a lot of sweat and tears.

Fat Bloke on Tour
The basic question being asked — can FAP produce 500K vehicles pa? Then the answer is yes — no real issue. FAP is essentially a very large — and growing — roof which provides dry space for production and Man Eng to show their stuff. Never been so a fair bit of speculation … Tesla FAP started with a line for the Model S which was pretty basic — this now seems to be the prototype line. Then there is the newer all singing / all dancing Luxury line that currently builds the Model S and the Model X. Pretty large and complex vehicles with a significant amount of options — my thoughts are this will run at 2 minute tack time / 30 cars per hour very approximately. Normal hours / double shift = 110K vehicles pa based on 46 / 47 work weeks. Might make more if then can get the run rate up a bit — 36 cars ph would be the max. Model 3 is different — it is at best premium not luxury and it has very few options. Consequently one line working at 50 plus vehicles ph has the potential to meet the mythical… Read more »