Wiki Model 3 delivery estimator

[Troy]

Not necessarily. It shows 2250/wk in April, and Tesla's current outlook is to hit 2000-2500/wk my the end of March.

Actually, 2,000-2,500/week was a misunderstanding. Elon was talking about the Grohmann module production line increasing existing capacity by 2,000-2,500/week. Later they clarified that the Grohmann line alone will add 2,000-2,500/week but they can get to 2,500/week by the end of Q1 without the Grohmann line. Here is Tesla's statement to clarify the misunderstanding: Tesla - Current Report

 

[WileyTheMan]

Almost two years to the date of reserving is my estimate. We shall see!

 

[suwaneedad]

@suwaneedad, I would need to see the exact wording they have used. The video is here. You can give me links to a specific time in the video if you want. Alternatively, if you click CTRL+U on this page, you could search the transcript and show me an exact quote. Otherwise, this will have to wait until I listen to the conference call again. There is a question after 36:33 and Deepak says the following:

Actually it was in the Feb shareholder letter. I can't square the circle, after reading this statement, that production will continue to ramp immediately after attaining 5000 units/wk. "As we shared previously, in order to incorporate our learnings and be capital efficient, we intend to start adding enough capacity to get to a 10,000 unit weekly rate for Model 3 once we have first hit the 5,000 per week milestone."
Anyway, if at some point you notice there's something disjointed between your model and when Tesla is estimating deliveries, on either side of the 5000/wk point in time, it could be due to this "intend to start adding" comment. I read that as "intend to start spending on incremental capacity" but obviously I could be wrong.

 

[brian98]

Troy,
I love your enthusiasm but it surely must be more accurate to take the True production estimates from your sheets of Vin regs i,e Jan 570 Feb 670 March 780 April 885 UNTIL we actually see some significant movement upwards in production rates.
Feb has come down on VIN's estimates from 1179 on 6 Feb to 644 today and therefore Feb production is likely to be well under 1000 per week and probably nearer 600.
I know Elon has said 2500 on existing capacity by end of March but there is No sign of any significant production increase other than 100 per month.
At that rate it will take 19 months until production reaches 2500 per week.
Unlike Elon it may be better to under promise and over deliver.
Thanks again for your great work.
Just give a thought for those of us in UK who are looking at 2020 for deliveries!!!

 

[Troy]

Hi, @brian98. The easiest way to compare the estimator to the VIN charts is probably to look at the production total since Tesla started making the Model 3. The estimator shows 8,602 units as of today (see cell D1208 on the data tab). The user VIN chart suggests 6,817 and the NHTSA VIN chart suggests 10,618. You can see these last two numbers here.

The big difference between the NHTSA and user VIN data could mean that Tesla has made those cars but they have not delivered them for one reason or another. It could be lack of battery packs or suspension coil replacements or some other reason. These two charts used to show numbers that were very close. Let's see what Tesla will do. They have 40 days left until the end of Q1 to reach 2,500/week.

 

[brian98]

Thanks Troy,
I wonder where they are storing these alleged nearly 4000 cars.
I recognise lag between NHTA and VIN sightings but at current production rate of 640 per week that is a lot of cars to be missing.
As you say we will find out in 40 days if they have succeeded in quadrupling production.
Thanks for trying to understand/justify my concerns.

 

[schonelucht]

I don't think they're storing 4000 cars anywhere in particular. The difference is likely due to Tesla running a bit ahead of actual production start with NHTSA registration, cars on transport all over the US, time difference between car in production, versus VIN assigned versus car delivered and then finally indeed a substantial (but not 4000) cars in various stages of assembly and quality control. The take away of it all is that Model 3 is in a sort of steady state of production somewhat below or near 1000/week. Elon promised a step up in production by the end of this quarter. Should that happen, it'll be clear enough pretty much immediatly after these cars start rolling from the factory floor.

 

[ℬête Noire]

Thanks Troy,
I wonder where they are storing these alleged nearly 4000 cars.
I recognise lag between NHTA and VIN sightings but at current production rate of 640 per week that is a lot of cars to be missing.
As you say we will find out in 40 days if they have succeeded in quadrupling production.
Thanks for trying to understand/justify my concerns.

I don't expect there is 4000 cars but there is lots of places cars can be in the logistics pipeline. On the road, in service centers getting prepped for handover.

Customers in TX (with more contact w/the Las Vegas coordination team because of local rent seeking laws here) have been told Tesla has a logistics staging location in Lathrope, CA. TX cars in particular sit there a while before moving because we have to pay in full before they get shipped into the state.

I've not seen anyone find that location and report on it. *shrug*


EDIT: It is possible that Tesla is already in the midst of installing the equipment from their German engineering group, and this might be leading to something of a lull where production is flat as they have resources focused on getting that up.

 

[RTPEV]

Actually it was in the Feb shareholder letter. I can't square the circle, after reading this statement, that production will continue to ramp immediately after attaining 5000 units/wk. "As we shared previously, in order to incorporate our learnings and be capital efficient, we intend to start adding enough capacity to get to a 10,000 unit weekly rate for Model 3 once we have first hit the 5,000 per week milestone."
Anyway, if at some point you notice there's something disjointed between your model and when Tesla is estimating deliveries, on either side of the 5000/wk point in time, it could be due to this "intend to start adding" comment. I read that as "intend to start spending on incremental capacity" but obviously I could be wrong.

I think you make a fair assumption that they will probably hit 5000/week (or some number in that neighborhood--Tesla thinks it may be 6000-7000 before they actually top out, but I'll chalk that up to potentially unjustified optimism) at which point they will assess and start spending on the ramp to 10K. At this point I would expect a flattening of the production rate as they put their orders in for new equipment, have it built, installed, increase orders from suppliers that may need a few months lead time, etc.

 

[suwaneedad]

I think you make a fair assumption that they will probably hit 5000/week (or some number in that neighborhood--Tesla thinks it may be 6000-7000 before they actually top out, but I'll chalk that up to potentially unjustified optimism) at which point they will assess and start spending on the ramp to 10K. At this point I would expect a flattening of the production rate as they put their orders in for new equipment, have it built, installed, increase orders from suppliers that may need a few months lead time, etc.

Not to mention, by then lots of energy will be pointed toward Model Y. Ensuring the new capacity can serve both models, deciding whether Fremont can really continue to be the sole assembly site, etc. The number of assembly plants built by carmakers in the Southeastern US is worth noting; I can't envision Tesla doubling-down on its Fremont investment...too many good reasons to decrease reliance on CA labor costs, CA union efforts, earthquake damage risk, etc. Myriad carmakers have expanded to the Southeastern US, as has Boeing. If Tesla doesn't follow suit, as well as potentially in a place like Germany to serve the European market more seriously, I will be exceptionally surprised. Not that I'm suggesting these will be 2018 announcements/investments. Allllll that said, I don't think we'll ever see 10k Model 3s per week...tapering the top end off to some degree in order to make room for Model Y feels much more likely to me.

 

[ZeApelido]

The largest variable in delivery estimation is Tesla, not Troy's modeling. It's silly to not account some Bayesian prior into the estimation, as well as give some uncertainty.

For instance the original estimator said I would get the car in late October...but some Bayesian logic would have given a later result way back when.

Now here we are again, assuming over 2000 by end of March. A true 2000, not extrapolated. And there's no uncertainty to that number? Doubtful.

What is the probability distribution for production rate at end of March? To me, it's 25% above 2000, 60% chance above 1500, 90% chance above 1000.

 

[ℬête Noire]

Now here we are again, assuming over 2000 by end of March. A true 2000, not extrapolated. And there's no uncertainty to that number? Doubtful.

What is the probability distribution for production rate at end of March? To me, it's 25% above 2000, 60% chance above 1500, 90% chance above 1000.

The "actual" number is still the 2500, Tesla's most recent target, the 2000 (er, 2250 maybe, or is that for first week of Apr?) Troy is using would be the median (the 50/50 point) he's selecting.

What's your basis for those numbers in the last paragraph?

 

[sreams]

What is the probability distribution for production rate at end of March? To me, it's 25% above 2000, 60% chance above 1500, 90% chance above 1000.

Yeah... but that isn't based on anything apart from a personal feeling.

 

[ZeApelido]

The "actual" number is still the 2500, Tesla's most recent target, the 2000 (er, 2250 maybe, or is that for first week of Apr?) Troy is using would be the median (the 50/50 point) he's selecting.

What's your basis for those numbers in the last paragraph?

Tesla history of goals vs. actual, current production levels / ramp trends. A bayesian prior is never perfect, but still is useful to give a better estimate in the end (on average). And you update it as you get more information. So I guess my max probability pick would be about 1700 / wk. And I don't mean 'if we extrapolate rate of different lines blah blah blah", but actual average production rate at the end of March.

Of course I could be wrong. Just don't ignore the prior and just input what Tesla says. Fair enough that Troy may have adjusted a bit lower than Tesla's goal, so that's a start.

Often betting markets bring decent probability estimates. When you put your money down, people get more realistic.

 

[ZeApelido]

Yeah... but that isn't based on anything apart from a personal feeling.

 

[ℬête Noire]

@ZeApelido

Not only did Trop drop it below Tesla's official goal numbers but what's also happening with that curve is that it's getting smoothed out over multiple weeks, as production is likely to have leaps and such but good luck trying to model those with granularity easily with any meaningful accuracy. The exact date of getting the new equipment turned on for production is going to be hard to nail down and it's a substantial influence on the production number. So the future is dragging the Apr 1 number up, too.

Steep edges like that are hard to nail exactly, in a lot of ways it's not so much whether they hit it rather it's WHEN they hit it. However ultimately it's probably not even all that material to estimated delivers as the difference gets buried in the [hopefully] much larger production volume down the line, certainly buried within the variance of possible outcomes even a couple months out.

Another thing about the priors here that you may be missing, this is Tesla bringing in-house a previously [somewhat failed] out-sourced engineering task. They've had pretty good success executing that in the past, when they bring their focus onto it. This is also something that Musk has had a lot of success with at Space-X.

 

[smak]

If Tesla has the manpower, and expertise to make enough in-house, why don't they do it in the first place?

 

[mrtian97]

When Tesla said dual motor mid 2018, do they mean dual motor with the Long Tange battery or Short range? Any idea?

 

[Troy]

When Tesla said dual motor mid 2018, do they mean dual motor with the Long Tange battery or Short range? Any idea?

I see AWD and Standard Range as two separate options that will become selectable in the design studio. Imagine a checkbox that says AWD. They will add that in Mid-2018. What does this say about short range vs long range? Nothing. That's a different checkbox that will be added later. Therefore, the way I read it, Mid-2018 for AWD means only LRD.

In this scenario, SRD would be a selectable option as soon as SR becomes available. You would just select the SR checkbox and the AWD checkbox and there you have your SRD. However, they might make the SRD available before SR because of higher gross margins. They need to show that the Model 3 situation looks good before raising more money for the Model Y. Tesla seems to be more focused on the Model Y than I would've thought they would be at this stage.

 

[MattD]

@Troy I figure I would share some data as it might be relevant to the conversation. I am a former Tesla employee who reserved with a @Tesla.com email. I reserved on August 24th 2016 which was my last day on the job. I made 2 reservations.

The reason my delivery estimate is important is that I still got a certain amount of priority (but not active employee priority) I have been narrowing down where I stand in line as it is a unique situation.

It looks like I am after the at the end of the current owners but before Day 1 in store reservations.
My Tesla Delivery Estimator says February-April. This was recently changed from Oct-December.

I have a feeling I am going to get the configuration email at the end of February (or whenever the pool of current owners runs out) with delivery towards the end of March but who knows. Anyway hopefully my unique situation provides some clarity as to where things are at.

Thank you so much for taking the time to provide these delivery details and data! To everyone else, the model 3 is an epic car and will be worth the wait

Fun Fact: I met Elon Musk at the Falcon Heavy Launch. One of the best days of my life...

Matt