General Discussion: 2018 Investor Roundtable

[wdolson]

If you are referring to Munroe, he is not a robotics expert, rather he is anti-robot. Some task are easier to have humans do, but that does not mean they are not capable of being automated. Seat installation for instance. People can do it, but robots don't get relative stress injuries, and are more consistent time wise.

In one of the videos where Sandy Munro discussed his teardown of the Model 3, he said he was one of the first people in Detroit to get certified for robotics manufacturing techniques and he did say some things are well suited for automation, but there are also a lot of tasks better done by people.

Some things were silly to automate, others worked as expected, and some were punted to be automated later. That is hardly a dead end.

They did run into some dead ends because they realized there was tasks humans could do better.

You are very confused. All Tesla has done is temporarily delayed some of the harder to automate tasks. Super automation is coming, just not as fast as they hoped. They aimed really high and hit just a bit high, but they haven't given up on anything. And whatever they don't get done now will be done later when they build a line for Model Y. The goal remains getting rid of all the people in any position where they can slow things down.

I think they've learned three important things for the next step: 1) complexity is anathema; 2) don't try to do everything at once; 3) integrating improvements into an existing working line is much simpler and effective that doing it all from scratch.

I watched the shareholder's meeting video and other comments Elon has made in recent weeks. He has said he overestimated what robots could do and underestimated how flexible humans were. In the shareholders meeting he said that automation makes sense for simple, repetitive tasks, or for tasks that are dangerous for a human to do, but for many other manufacturing steps, humans are better.

To my ear, that sounds like he's backing down on at least some of the automation plans.

Another point Sandy Munro made was if you are automating a manufacturing technique, you need to design the thing being made from the start to be made by machine. Trying to automate a process that has been done by humans without re-engineering the thing being made doesn't usually work very well. The Model Y may be designed more for automated assembly, but I doubt they will achieve the level of automation Elon was predicting a year ago anytime soon.

I've been doing engineering for 30 years now. I've worked for heavy industry (3 different aerospace companies), small electronics both large scale and small scale production, and some pure software projects. Large scale heavy industry is one of the most logistically intensive things humans do. Logistics is a completely different skill set from R&D. Tesla is an R&D heavy company. Most Silicon Valley companies are.

They failed to learn the true logistical lessons from spinning up Model S and X production. They are learning them the hard way now. Elon thought there was some low hanging fruit in manufacturing that nobody else was doing and tried them only to find out they weren't really low hanging fruit at all, they were very tough and thorny problems other companies had tried and abandoned.

How to make things is one of the most studied subjects in the world. An Industrial Engineering degree is all about how to engineer production procedures and production lines.

Elon is a visionary and an R&D engineer by temperament, he's not a logistics guy. Nobody can be an expert in everything.

 

[30seconds]

To recap:

Model 3 LR has been run on the track showing that it doesn’t overheat, has crappy brakes and is already fast and well balanced.

Performance version will be considerably faster, with better suspension & tires, and better brakes.

As a “car” guy I find this to be pretty insane now that the first model 3 performance is rolling off the line and NO ONE outside of Tesla has driven it yet.

tl;dr
1. Elon sure knows how to create some drama

2. I-Pace discussion is about to end

3. Someone at BMW is about to think “Well zee are all fuuuked now”


Elon Musk on Twitter

 

[Waiting4M3]

Now I am confused.

I recalled from the June 5, 2018 Shareholder Meeting that Tesla had 2 currently working GA lines, with a 3rd under construction.

Zach Shahan’s recollection is the same: 17 More Tesla Shareholder Meeting Highlights | CleanTechnica

He reports that the 3rd line began construction approximately 2 weeks prior to June 5. Today marks about 3.5 weeks after start of construction. That’s consistent with Elon’s tweet today saying that the GA line pictured with 1st completed Dual Motor Model 3 took about 3 weeks to build. I think the tent is the 3rd GA line for Model 3. Is there a reason why my conclusion may be incorrect?

If the tent is the 3rd line, and completely outside of the existing factory building, why would Tesla need to shutdown to build it? I think the 3rd line is inside the factory building, and the tent is a 4th line for AWD.

 

[NicoV]

No not at all, you don't buy something first and see what it is later.It's a different design aimed at better ... let's say the ultimate metric is overall cost per unit.
If that argument is not enough, they said they can likely get to 5k with the 2 existing lines, why would they spend on a forth line today?

Lack of space to get them to 10K/week with the productivity of line 1 and 2.

 

[J X 3]

Lack of space to get them to 10K/week with the productivity of line 1 and 2.

What?
There is no 4th Model 3 line, this kind of nonsense makes me want a 1 way ticket to Mars.
A drop of common sense shouldn't be too much to ask for.

Yeah sure they are gonna spend the CAPEX to get general assembly to 10k per week today and idle it for a few quarters before everything else catches up. It's called the Looney Tunes strategy.

 

[mongo]

What?
There is no 4th Model 3 line, this kind of nonsense makes me want a 1 way ticket to Mars.
A drop of common sense shouldn't be too much to ask for.

Yeah sure they are gonna spend the CAPEX to get general assembly to 10k per week today and idle it for a few quarters before everything else catches up. It's called the Looney Tunes strategy.

Then what is your take on Elon's e-mail?
Specifically:

paint shop output, GA3, bringing up the new GA4, End of Line and Module Zone 4 at Giga

 

[mongo]

Elon thought there was some low hanging fruit in manufacturing that nobody else was doing and tried them only to find out they weren't really low hanging fruit at all, they were very tough and thorny problems other companies had tried and abandoned.

The lowest hanging fruit is not necessarily low hanging fruit, especially on an old tree. Orchards that have been in business for years using the same methods are not going to risk disrupting the revenue chain to try a new system (or possibly even invest in the attempt).

Analogy warning: Reusable rockets were one of the ways to cut launch expenses, but that was a tough and thorny problem that others (experts, none the less) gave up on.

If your first lightbulb doesn't work, or 900th, it doesn't mean the problem is unsolvable.

Some problems only become solvable with increases in technology. ICE allowed heavier than air flight.

From the video interview, I did not get the sense Munro was up on the latest robot tech. The advances in machine vision make today's robots far beyond his "blindfolded one-armed builder". Although the concepts he developed for designing the parts so that even a B.O.B. can assemble them are still relevent to modern manufacturing.

It will be interesting to see what they do with Y...

 

[NicoV]

What?
There is no 4th Model 3 line, this kind of nonsense makes me want a 1 way ticket to Mars.
A drop of common sense shouldn't be too much to ask for.

Yeah sure they are gonna spend the CAPEX to get general assembly to 10k per week today and idle it for a few quarters before everything else catches up. It's called the Looney Tunes strategy.

Elon confirmed that nonsense.
My take on this is that GA1 and GA2 did not achieve their target productivity and that with that experience they built GA3, which turned out to be so much better that they built another one, i.e. GA4. They have to produce now on GA1 and GA2 with lower margins until GA3 and 4 can produce whatever the target rate they’re targeting. Seriously, producing cars in a tent does seem like a plan B to me. All of this is fallout of production systems that didn’t achieve their targets.

 

[J X 3]

Then what is your take on Elon's e-mail?
Specifically:

It's the 4th overall, the 3rd for Model 3.
GA1 is S&X and there were 2 for Model 3, this (GA4) is the new one that was mentioned at the shareholders meeting as being better than the existing 2 (GA2 and GA3). But there is no 5th overall, 4th M3 like some people insanely insist.
The quote from the leaked email mentions "bringing up" GA4 and at the shareholders meetings Musk says that they are "constructing a third" and they started construction about 2 weeks ago. Yesterday he mentioned it took them 3 weeks for the new line.
at 42:18

 

[SwTslaGrl]

paint shop output, GA3, bringing up the new GA4, End of Line and Module Zone 4 at Giga

Imho, the tent could be the new "End of Line" not the new GA4..

 

[Starno]

Nico Rosberg

 

[hobbes]

Nico Rosberg (2016 Formla 1 world champion) has a long video driving a Model X in Monaco and France on his youtube channel - judging from the Tesla smile, won´t be long until he gets one...

Some time last year he was spotted at Fremont with Elon IIRC, so maybe there is some connection.

EDIT: @Starno beat me to it . Shouldn´t have written such a long comment I guess.

 

[hobbes]

I thought I read somewhere it was whoever was heading China Service is now Worldwide. Can’t recall where I read that.

I do rember where you read that . See here:

Robin Ren, Tesla’s head in China, has now taken over responsibilities of Worldwide sales.

 

[wdolson]

The lowest hanging fruit is not necessarily low hanging fruit, especially on an old tree. Orchards that have been in business for years using the same methods are not going to risk disrupting the revenue chain to try a new system (or possibly even invest in the attempt).

GM spent billions on automation in the 80s and it didn't work. Elon has admitted their experiment with automation was not a full success either. This is from April:
https://arstechnica.com/cars/2018/0...epeated-car-industry-mistakes-from-the-1980s/

In the interview with CBS Elon said they have a "fluffer" robot that was supposed to place a sound deadening blanket inside the battery pack, but it couldn't pick it up correctly. I human could do it with ease. They did some testing an determined the blanket was not doing anything perceptible for noise, so they eliminated it.

Robotics have improved since the 1980s, but there are still a lot of tasks humans can do with ease that takes a long time to get a robot to do correctly. And if a robotic machine has some kind of fault, it can become deadly very quickly. Humans can do dangerous things too, but a human arm flailing around doesn't weigh a ton and most human "failure" modes involve less movement than unexpected movements in unexpected directions.

Analogy warning: Reusable rockets were one of the ways to cut launch expenses, but that was a tough and thorny problem that others (experts, none the less) gave up on.

If your first lightbulb doesn't work, or 900th, it doesn't mean the problem is unsolvable.

Some problems only become solvable with increases in technology. ICE allowed heavier than air flight.

From the video interview, I did not get the sense Munro was up on the latest robot tech. The advances in machine vision make today's robots far beyond his "blindfolded one-armed builder". Although the concepts he developed for designing the parts so that even a B.O.B. can assemble them are still relevent to modern manufacturing.

It will be interesting to see what they do with Y...

Munro's specific knowledge of robotics is probably out of date. I once could design a high frequency RF amplifier, but I haven't in 30 years. But I still know the basic concepts. He may not be up on what exactly a modern robot could do vs one 30 years ago, but he knows conceptually what needs to be done to apply robotics to manufacturing.

This is a tangent, but I think the media tends to hype on every little mistake Tesla makes in part because Elon is one of the most honest CEOs out there. In a world where everyone is spinning everything to a point where some are spewing things that not only a factually false, but don't even hold together logically, when someone comes out and says, "I screwed up and this is what happened..." the media assumes that the admission is only the tip of the iceberg instead of the whole story. They assume if Elon is admitting anything is wrong, the entire company must be on the verge of utter collapse.

 

[22522]

Multiple lines let you improve process with minimum disruption to production, as the other lines keep running.

They can play leap frog with each other and compete.

 

[mongo]

It's the 4th overall, the 3rd for Model 3.
GA1 is S&X and there were 2 for Model 3, this (GA4) is the new one that was mentioned at the shareholders meeting as being better than the existing 2 (GA2 and GA3). But there is no 5th overall, 4th M3 like some people insanely insist.
The quote from the leaked email mentions "bringing up" GA4 and at the shareholders meetings Musk says that they are "constructing a third" and they started construction about 2 weeks ago. Yesterday he mentioned it took them 3 weeks for the new line.
at 42:18

It's not insane to question how the data aligns.

In the SH call (June 5th mid afternoon), the 3rd GA already had been built with cars going through it, so that doesn't seem to jive with the tent not being built on May 31, unless they prebuilt everything, got the tent done, then moved it out, bolted it down and commissioned it all in 5 days. Which is possible.
In that senario the tent is the third Model 3 GA and they already had 2 GA model three lines. If that is so, then the second of the original two lines is having issues (email's GA3), and the 3rd line which we had cars on it June 5th, is still being brought up on June 16th (e-mail's GA4).

OR
There were two established model 3 GA lines at the time of the SH call, plus a third that got going in only ~2 weeks. In parallel, due to GA being the bottleneck, a fourth line was being built in the tent at the time of the SH call and is now being commissioned (same design as 3rd line). Thus also aligned with GA4 not being supported much vs the 3rd line which was needed for >5k.

I agree the reported line assembly timing makes the starting points potentially align (about 2 plus 1.5 weeks ~~ 3 weeks), so I could be off, but the plant shutdown aligns with the third line bring up pointing toward interior install.

And possibly the GA4 includes the tent build time. I tend to lean toward the optimistic side, so would like to think the first 2 Model 3 lines are working ok, and only the two new ones need tweaked, but reality trumps optimism.

 

[electracity]

2. I-Pace discussion is about to end

Nah, the iPace is well designed and well positioned against Tesla. It is a significant car, although the reality of the iPace will probably be plenty of bugs and low production. Plus no charging network. So I doubt it takes a lot of sales from Tesla over the next few years, but that doesn't make it an unimportant car.

The higher luxury end of the car biz is not where the main EV battle is going to take place anyways. There will be a lot of great $75K+ EVs, but those cars are not particularly important. It will be interesting how Tesla allocates resources to update the S/X relative to other needs.

 

[electracity]

The lowest hanging fruit is not necessarily low hanging fruit, especially on an old tree. Orchards that have been in business for years using the same methods are not going to risk disrupting the revenue chain to try a new system (or possibly even invest in the attempt).

Car manufacturing is not an "old tree". It's a highly competitive business where companies improve or die. Successful car companies today are the winners of a brutal competition conducted over many decades.

The opportunity for innovation has always been at the gigafactory, not in most of the processes in Fremont.

 

[HG Wells]

Since it is the weekend and its fathers day. I would like to wish Elon a happy fathers day.
May he find a few (just a few) hours to spend with his kids. They don't stay small for long.

 

[J X 3]

@ mongo

2 lines might potentially reach 5k on their own and doing up to 3.5k at the time , burst. 3rd line is better, they don't need a 4th anytime soon.
Musk said putting cars through it, doesn't mean they exit all that fast and if they do, doesn't mean they are good enough for delivery.
GA is just a fraction of production , downtime doesn't mean it was just for GA. And ofc if they scale GA ,they need everything else to scale or they can't supply it with parts , there wouldn't be a point in increasing output for GA without increasing output everywhere.
And this is not installing a toaster, it takes time. I'll remind you about what was said on AWD, they start this month, ramp in July+August and volume surely by Sept.

PS: You guys are focusing on the wrong thing anyway, volumes are getting there , the focus shifts to margins.and how fast they can get M3 margins to 15% and then 25%. They need close to 15% in Q3 and hopefully 25% by Q2 2019. And once there is a better understanding on margins, the focus shifts on demand.