Tesla is dropping the hammer on employee leakers

[WHIZZARD OF OZ]

Question is just how much of it has to be "hand built". If they got the design right it should be well under 20hours per CT. Apparently GS is doing MY in 2.5 hours end to end. If they’re doing their unboxed assembly it might even be less!!

I'm pretty sure GFA will be running 3 shifts 24/7 to get some out the door. Let's do some math:

GFA Production
Date Predictions
	Days	Increase	Units per day	Units per week	Month	Total Units	Shift Employees	Hrs per CT
1-Aug-23​		0%​
1-Sep-23​	31​	30%​	7​	49​	217​	217​	29​	100​
1-Oct-23​	30​	140%​	17​	118​	504​	721​	70​	90​
1-Nov-23​	31​	120%​	37​	259​	1146​	1867​	154​	60​
1-Dec-23​	30​	100%​	74​	517​	2218​	4084​	308​	50​
1-Jan-24​	31​	90%​	140​	983​	4354​	8438​	585​	40​
1-Feb-24​	31​	70%​	239​	1671​	7402​	15840​	995​	30​
1-Mar-24​	29​	60%​	382​	2674​	11079​	26918​	1592​	20​
1-Apr-24​	31​	45%​	554​	3877​	17172​	44090​	2308​	10​
1-May-24​	30​	25%​	692​	4847​	20772​	64862​	2885​	8​
1-Jun-24​	31​	15%​	796​	5574​	24684​	89547​	3318​	6​
1-Jul-24​	30​	10%​	876​	6131​	26277​	115824​	3650​	4​
1-Aug-24​	31​	5%​	920​	6438​	28510​	144334​	3832​	4​
1-Sep-24​	31​	4%​	956​	6695​	29651​	173985​	3985​
1-Oct-24​	30​	3%​	985​	6896​	29555​	203540​	4105​
1-Nov-24​	31​	3%​	1015​	7103​	31457​	234997​	4228​
1-Dec-24​	30​	3%​	1045​	7316​	31355​	266352​	4355​
1-Jan-25​	31​	2%​	1066​	7463​	33048​	299400​	4442​
1-Feb-25​	31​	2%​	1087​	7612​	33709​	333109​	4531​
1-Mar-25​	28​	2%​	1109​	7764​	31056​	364166​	4621​
1-Apr-25​	31​	2%​	1131​	7919​	35071​	399237​	4714​
1-May-25​	30​	2%​	1154​	8078​	34619​	433855​	4808​
1-Jun-25​	31​	1%​	1165​	8158​	36130​	469986​	4856​
1-Jul-25​	30​	1%​	1177​	8240​	35314​	505300​	4905​
1-Aug-25​	31​	1%​	1189​	8322​	36856​	542156​	4954​

On a Graph the first year looks like this:

So I should see mine around March/April I hope.

But more importantly it shows the interdependence between hours required to assemble a CT and the amount of employees per shift also required to reach the 400k CTs per year proposed.

Note that the employee numbers are per shift, so 3 shifts would equal around 12,000 people per day for 400k a year production rate. I heard GA was meant to employ 25k or so? I've reduced the CT hours on the right over a year, by which time they should be meeting the 400k a year targets, and I'm assuming around 4 hours per CT.

Obviously these are just guesses, but they do give a indication of what is involved, and roughly how many people are needed to make up the hours required for CT production. The biggest "hour per unit" reduction, obviously comes from automation and reduction in parts, along with faster cycle times at each station.

Now in comparison a VW Golf takes 20 hours a Toyota Corolla around 18 hours. So a MY is being made in Shanghai at nearly an order of magnitude faster than them, which frankly is insane.

That's nearly like just sending a magnet base frame down the assembly line and the parts jump on there from the side convey belts, magnetically, to make a car.

'Let's Stick Together, C'mon C'mon'
( Roxy Music )

Sponsored

 

[cvalue13]

Question is just how much of it has to be "hand built". If they got the design right it should be well under 20hours per CT.

once and after the production line is running and shaken out, I’m sure it will be speedy (assuming no parts constraints)

I’m more interested in the time between now and when they get the line running and shaken out

because that’s the time before they begin making retail units

 

[JBee]

once and after the production line is running and shaken out, I’m sure it will be speedy (assuming no parts constraints)

I’m more interested in the time between now and when they get the line running and shaken out

because that’s the time before they begin making retail units

So you don't think it will just transition between pre-production and production? How much do you think is missing internally?

I imagine that as they bring the input individual lines up to speed, a few will not ramp as intended with issues that need to be fixed with extra attention but the rest will ramp as fast as the stragglers.

Now which things are the stragglers on ramp?

I can tell you which ones shouldn't be the reasons for a delay:

1. Casts, cabin structure, doors, windows
2. motors, drivetrains, suspension, wheels, brakes etc
3. Interior, seats dash, infotainment etc

These are virtually "off the shelf" versions from the other already running Tesla lines.

The one that is of most concern is the 4680 production constraints for the structural packs.
The packs we already have had in production for the MY, just a bit smaller, so no real risks there, except cell supply.

The way I see it is that Tesla has already adopted many of the proposed system upgrades for CT into their already existing models. I'm keen on seeing the M3 highlander config atm.

 

[cvalue13]

So you don't think it will just transition between pre-production and production? How much do you think is missing internally?

I imagine that as they bring the input individual lines up to speed, a few will not ramp as intended with issues that need to be fixed with extra attention but the rest will ramp as fast as the stragglers.

Now which things are the stragglers on ramp?

I can tell you which ones shouldn't be the reasons for a delay:

1. Casts, cabin structure, doors, windows
2. motors, drivetrains, suspension, wheels, brakes etc
3. Interior, seats dash, infotainment etc

These are virtually "off the shelf" versions from the other already running Tesla lines.

The one that is of most concern is the 4680 production constraints for the structural packs.
The packs we already have had in production for the MY, just a bit smaller, so no real risks there, except cell supply.

The way I see it is that Tesla has already adopted many of the proposed system upgrades for CT into their already existing models. I'm keen on seeing the M3 highlander config atm.

I guess I don’t know if we’re talking past one-another here

the RC1 trucks we’re seeing now are not being built on the production line proper.

there continues to be a separate, temporary, RC “pilot line” at GFTX, relocated from Fremont to GFTX, for accomplishing a significant portion of these builds “by hand.”

It’s an iterative, step-wise, process of Increasingly attempting to utilize more and more of the production line and less and less of the pilot line.

so increasingly, parts are being created on the line tooling (eg castings) to validate that tooling, others may be partially assembled on line tooling (eg welding the castings) to validate that tooling, while many other processes that will eventually be automated (eg windshield install) are being done by hand while the line is being completed - and this work is being done eg a rolling chassis on dollys, in a part of the factory that isn’t the production line, proper.

as tools or assembly processes are validated, increasingly more steps will be done on the production line proper. At some point in validation that will convert from RC1 builds to RC2 builds.

Those RC2 builds will be what, for example, may be delivered to insiders at the “delivery event” - they’re basically advanced prototypes, given only to buyers for whom Tesla can monitor and regulate still remaining issues in the builds.

after whatever requisite number of RC2s are built, there’ll then be some run of ‘confirmation units’ on the production line proper.

after all THAT checks out, there’ll be built the first production unit on the validated production line.

(the term “RC” in Tesla speak is roughly equivalent to what most manus term “production validation vehicles” - vehivles being built for purpose of testing parts and assembly processes)

and mind you: if Tesla will be producing more than one variant, there’ll be iterations of the above process for both portions of the line relevant solely to one variant or the other

for some visual distinctions here:

here are some sanctioned views of the production line and production line area:

Meanwhile, here’s where we’re seeing units being finished out, and with stacks of parts boxes around, etc. (from Joe vids):

meanwhile, leaks like these show the RCs on dollys for manual movement between build stations

Note this recent tweet from Joe

 

[JBee]

I guess I don’t know if we’re talking past one-another here

the RC1 trucks we’re seeing now are not being built on the production line proper.

there continues to be a separate, temporary, RC “pilot line” at GFTX, relocated from Fremont to GFTX, for accomplishing a significant portion of these builds “by hand.”

It’s an iterative, step-wise, process of Increasingly attempting to utilize more and more of the production line and less and less of the pilot line.

so increasingly, parts are being created on the line tooling (eg castings) to validate that tooling, others may be partially assembled on line tooling (eg welding the castings) to validate that tooling, while many other processes that will eventually be automated (eg windshield install) are being done by hand while the line is being completed - and this work is being done eg a rolling chassis on dollys, in a part of the factory that isn’t the production line, proper.

as tools or assembly processes are validated, increasingly more steps will be done on the production line proper. At some point in validation that will convert from RC1 builds to RC2 builds.

Those RC2 builds will be what, for example, may be delivered to insiders at the “delivery event” - they’re basically advanced prototypes, given only to buyers for whom Tesla can monitor and regulate still remaining issues in the builds.

after whatever requisite number of RC2s are built, there’ll then be some run of ‘confirmation units’ on the production line proper.

after all THAT checks out, there’ll be built the first production unit on the validated production line.

(the term “RC” in Tesla speak is roughly equivalent to what most manus term “production validation vehicles” - vehivles being built for purpose of testing parts and assembly processes)

and mind you: if Tesla will be producing more than one variant, there’ll be iterations of the above process for both portions of the line relevant solely to one variant or the other

for some visual distinctions here:

here are some sanctioned views of the production line and production line area:

Meanwhile, here’s where we’re seeing units being finished out, and with stacks of parts boxes around, etc. (from Joe vids):

meanwhile, leaks like these show the RCs on dollys for manual movement between build stations

Note this recent tweet from Joe

Thanks for the extra info, I hadn't realised the distinction between the imported pilot and production lines at GFTX.

Although their might be some assembly in the pilot line by hand, I'm not sure what parts, if any, that it will produce for the actual production line. I also don't think they needed the pilot line to get the production one "online" as such, beyond maybe some hands on training, rather they needed the pilot one to make some CT's in numbers to get some units for external testing. Note Fremont is undergoing extensive refitting as well for Highlander, so maybe they wanted CT out of Fremont?

I'm also not sure what parts they are manufacturing that will need to be adjusted, customised or adapted to make the line work. Those individual part lines already need to be working before assembly can start. To the point I don't think bringing the assembly line up to speed is much of making the already manufactured parts fit together, because they already do, but more about getting the work stations setup for fast cycle fitting times including training the humans and the bots that handle it (about 2min for 40 stations on the M3)

Are you thinking the CT line is a unboxed, more parallel approach as they described in their last investor conference? Or a more conventional line. I think they are already going to do the unboxed manufacturing with CT and there won't be a single line. I also think they are already going to do it for Highlander as well. Tesla is known to be a fast iterator, they've had 4-5 "conventional" models of M3/MY already.

My thinking is that there is 4-5 main module lines, that then come together for "final" assembly of the separate modules. If so, then the question becomes which of those lines become the bottleneck for ramp. Some of those lines will by their nature be easier than others to achieve, which in turn means they can continue to be optimised while the harder ones are ae sorted out separately, all without affecting the final assembly line at all. Except of course that such a slower module line will determine final assembly numbers.

Following on from that, I'm wondering which modules, or sub-assemblies, are they making that "do not" resemble already existing sub-assemblies from a MY in GFTX. I can't think of any "new" one except the SS panels, and I can't think that it will be a problem to fold those in high numbers. That was the reason for the low polygon unpainted use of SS panels in the first place, not because of the exo etc.

So I suppose if they do it as above, and there is no new process to learn or optimise, I think the question for production numbers really comes more down to part availability, and not achieving assembly speeds with the line. Sure there will be a few months to get numbers up, but ultimately, I think by early next year the rate will be set by cell production numbers.

 

[cvalue13]

Thanks for the extra info, I hadn't realised the distinction between the imported pilot and production lines at GFTX.

that distinction should be increasingly collapsing as portions of the line are validated, as I understand it

Although their might be some assembly in the pilot line by hand, I'm not sure what parts, if any, that it will produce for the actual production line.

not sure I follow? There’ll be any number of parts the line is making (eg BIB, SS panels, etc.) and any number of parts being supplied from outside suppliers (eg trim, seats, etc.).

I think the distinction of RC Line vs validated production is largely, eg, the processes for incorporating these parts into the vehicle itself. How much is automated, how much is a “production line” validated process of hand installation, vs an ~ad hoc installation process.

rather they needed the pilot one to make some CT's in numbers to get some units for external testing.

I suppose, though part of this validation process is to utilize for the first time production parts built on production tooling. For example the recent Munro vid of the beta suspension shows the tie-rod having a one-off CNCd central adjusting nut. Now in TX, they’ll possibly the first time using the production part, either built by production tooling or by a production supplier. And validating assembly processes of that part and tooling.

Are you thinking the CT line is a unboxed, more parallel approach as they described in their last investor conference?

I really don’t know, but personally, I took that presentation to be pretty clear that the unboxed method will be in their next generation vehicle

And photos from the vehicles getting a symbol to date, don’t appear to be taking that unboxed approach

If so, then the question becomes which of those lines become the bottleneck for ramp.

yes, which portion of the production line process becomes the bottleneck, or which of the production parts, becomes a bottle neck. Obviously, the batteries are a well suspected potential bottleneck. but part of this validation process of determining whether there are others, anticipated, or especially unanticipated.

Can you clean your doing great?

Following on from that, I'm wondering which modules, or sub-assemblies, are they making that "do not" resemble already existing sub-assemblies from a MY in GFTX. I can't think of any "new" one except the SS panels, and I can't think that it will be a problem to fold those in high numbers.

None of that really relates to the current time. In the production validation process. All of those other models also went through this process.

Put a different way, I don’t think we’re here talking about what’s holding up production. I think we’re here talking about the fact that there is a process that takes a certain amount of time in order to get there.

in the past, the first delivery events to insiders have been of RC2 units. And in those instances, it’s been 3 to 4 months before the first production unit comes off the validated production line.

likewise, before the delivery event, there is a months long process to get to those. That’s where we are now.

 

[HaulingAss]

I think we maybe need to brace ourselves .. Production speed may surprise even the most optimistic.

I'm optimistic on production speed but I think people are expecting those speeds to be reached too quickly. It takes a lot of time and re-working of virtually every step of the process. A breakthrough improvement in the slowest process requires new innovations in other processes and sometimes the entire line must be reconfigured. The speed of gigacasting is probably the most limiting factor because the entire line must be configured around that.

At the beginning, there will be surplus gigacasting production capacity. As the lines come up to speed, they will maybe eventually match the casting speed. At that point, assuming the casting machines are already fully optimized, the only thing to do is to add more gigacasting machines. It creates a step-function in the speed of production.

All of this takes not months, but years, to become mature, even at "Tesla speed".

That is just considering production speed, but it must be remembered that is dependent upon a steady supply of parts and raw materials. Tesla can only ramp as fast as their suppliers and the purchasing guidance provided by Tesla. Tesla cannot take undue risk by guiding suppliers for too fast of a ramp or Tesla will have to pay the contractual penalties or stockpile parts for too long.

There are two ways to look at speed:

1) Production speed
2) Ramp speed

I'm bullish on #1 and think people are over-estimating #2. The best way to guess actual ramp speed is to look at the production Tesla is guiding for. Of course, Tesla doesn't know what kind of problems they might run into either, so they have to account for unexpected issues. And it's important to parse what Tesla says in terms of how they think of ramp speed. They will often talk about burst rates rather than average rates. If everything goes better than expected, the ramp will be slower than it could have been if Tesla had guided their suppliers higher (so Tesla has to be careful not to guide too low). Relying on Tesla's guidance is an imperfect method, because they don't know either, but it's the best we have.

As long as the economics of Cybertruck production pans out, and they are not limited by raw materials, I expect Tesla will continually increase production speed through 2026 and beyond, to levels that will continue to make traditional auto analysts look like the fools tools they are.

 

[JBee]

Ok so you think there will be a distinct changeover from RC2 to Retail CT units, not just a transition between the two?

I'm thinking it will be more subtle, and units will start collecting for a month or two, then they start some retail deliveries. At that point ramping will be underway. Everyday will see some numbers made, and over time they also add up.

Maybe I'm being more optimistic here, but I think the whole point of CT was to build vehicles faster, because faster is also cheaper, if you can use the same line and capex to produce more in the same time. You want to be able to pump those things out in numbers.

Substantial parts of the CT are ridiculously cheap already (cast is $150, which is close to resource value) so the only thing you can really do more is also bring down capex for the line, and increase its throughput, to achieve the lowest possible price per unit made (no paint shop etc). Both of those sort of force the line to also be simple and scalable, meaning in turn their overall complexity "should" be less. Which it turn should mean it's less likely to have a manufacturing hiccup.

I suppose the point I'm trying to make is that I don't see enough that is different on the CT to warrant the same profile ramp as a M3, neither do I think EM would of committed to a line that wasn't faster, simpler and more cost effective than it. The lessons on M3 would have been a dominate factor to change CT production drastically, and even optimise the CT design for it, not to complicate it, rather to uncomplicate it. Or as EM puts it, the manufacturing line is the product, not the vehicle itself.

Either way I think we'll see an official release fairly soon and a couple of deliveries, upon which we will get heaps of new information. A factory line tour would be cool, like they did for MY in GFTX.

 

[Coolbreeze704]

I'm optimistic on production speed but I think people are expecting those speeds to be reached too quickly. It takes a lot of time and re-working of virtually every step of the process. A breakthrough improvement in the slowest process requires new innovations in other processes and sometimes the entire line must be reconfigured. The speed of gigacasting is probably the most limiting factor because the entire line must be configured around that.

At the beginning, there will be surplus gigacasting production capacity. As the lines come up to speed, they will maybe eventually match the casting speed. At that point, assuming the casting machines are already fully optimized, the only thing to do is to add more gigacasting machines. It creates a step-function in the speed of production.



It time for you to have an avatar. Here you go

 

[CyberGus]

What I want to know is:

if they're already building Cybertrucks, then what is the Gigafactory extension project for??

 

[cvalue13]

What I want to know is:

if they're already building Cybertrucks, then what is the Gigafactory extension project for??

robo/next gen cheapy

 

[madquadbiker]

What I want to know is:

if they're already building Cybertrucks, then what is the Gigafactory extension project for??

The replacement for the discontinued Smart EQ, shall we call it the Model 1 where it’s made in one piece like a Dinky toy.

 

[PilotPete]

I can’t wait until one of the Optimus robots gets called in HR because its AI discovered X/Twitter/Instagram/YouTube and it was hooked on getting “likes”!

 

[jerhenderson]

they're threading the needle as some employees require their phones and taking photos - thus the 'photo passes'

to date I think Tesla hasn't terribly minded the occasional leaked photo of a certain variety

but I suspect this restriction now is due in large part to a few of the more recent leaks having betrayed some information that is material non-public information regarding production status.

read those tea leaves how you may

require their phones? I work In a maximum security jail... no one.. and I mean no one... requires their phones.

 

[cvalue13]

require their phones? I work In a maximum security jail... no one.. and I mean no one... requires their phones.

Thankfully Tesla’s assembly plants aren’t max prisons

notice all those QR codes on the Tesla parts?

Sponsored