Excerpt from book on Elon Musk
https://www.axios.com/2023/09/08/walter-isaacson-elon-musk-book-excerpt
Andy
1 Like
Interesting. One thing leapt off the page for me, though:
- So in May 2023, he decided to change the initial build location for the next-generation cars and Robotaxis to Austin, where his own workspace and that of his top engineers would be right next to the new high-speed, ultra-automated assembly line.
Has that been publicly announced by Tesla? That the âModel 2â and/or Robotaxi would be built in Austin first?
1 Like
I havenât seen anything, alot about Gigafactory, Model Y, Cybertruck but nothing about the 2 or Robotaxi. But that doesnât surprise me because they do not say exactly where they are on the FSD curve either, only that they are close.
Andy
1 Like
Itâs just, I thought that the next models were going to be built in the new Mexico facility - not Austin - regardless of where Tesla was in developing FSD.
1 Like
If this gets going for Tesla, I think it will be a huge profit center for them if they get something that really works well. IMHOâŚdoc
2 Likes
Right and that would make alot of sense only they canât get their engineers to move there, according to the article. People in the United States, especially near the border, hear about the cartels and all of the kidnappings in Mexico. I liked Mexico in the 80âs, not so much now.
Andy
2 Likes
Much depends on the timing they have in mind. The Austin plant is up and running and expanding. So new model could probably be made there fastest.
Building a plant in Mexico will take years. Probably at least two, maybe more.
As to kidnapping, etc, I think you expect to hire Mexican engineers who know how to deal with the risks. Tesla certainly does that in China and I suspect in Germany. Musk is from South Africa and is very savvy about international business. All of that is a given.
Its been years ago but I went on a business trip to Mexico City. I was suprised to find an armed guard in the lobby of the company offices there. Probably because of the risks you mention. Companies in Mexico know what precautions are required and see that as a cost of doing business there.
1 Like
Think in terms of logistics, you canât build cars in a factory that is not up and running but that does not mean the car cannot be designed elsewhere and the newly developed technology can then be incorporated in the new factory under construction. Linear thinking, a, b, c, d, is not helpful in analysing complex systems.
The Captain
2 Likes
Yes, in project management it is possible to do some things simultaneously shortening the time line.
But still no new models in Mexico plant for two years.
1 Like
I went back and read the article again. I think you are missing this point which was brought up in the article.
But a problem soon arose in his mind: He had always believed that Teslaâs design engineers needed to be located right next to the assembly line, rather than allowing manufacturing to be done at a remote location. That way, engineers could get instant feedback on how to design innovations that would both improve the car and make it easier to manufacture.
- This was particularly true for a completely new car and manufacturing process. But he realized he would have trouble getting his top engineers to relocate to the new factory. âTesla engineering will need to be on the line to make it successful, and getting everyone to move to Mexico is never going to happen,â he told me.
Andy
4 Likes
So you hire Mexican engineers, bring them to the US for training and then supervise from afar. European companies do this all the time. They use Expatriot programs to maintain communications.
I canât believe all the engineers at Teslaâs plants in China or Germany are from the US. No doubt some are but i suspect only a few.
I am sure Musk will figure this out. I am only pointing out what his thoughts were and why he is doing it in Texas for now. I agree with you Paul, eventually it is going to go to Mexico. Why wouldnât it? It would be so much cheaper.
Andy
1 Like
Itâs not who they are but how close to their work they are. Hands on!
BTW, Tesla brought a guy from China to the US as one of their top managers.
https://europe.autonews.com/move/tesla-makes-china-boss-tom-zhu-no-2-after-musk
The Captain
2 Likes
New Mexico or Mexico, the news is pretty confusing but this is from yesterday:
Tesla will build next-gen/cheaper electric cars in Texas rather than Mexico first | Electrek
Looks like Texas and this article about the new way to build Teslas and the robotaxi makes me want to buy TSLAâŚdoc
1 Like
Iâm going to let you in on a very special secret, EVs are just a stepping stone in Teslaâs journey. AI needs data. The 20 million per year EVs Tesla will make and sell by 2030 will be collecting data to train AI. Already the AI software used to develop FSD is being used to train Optimist robots that are starting to work in the Giga-Factories. These robots are collecting other kinds of data to train GAI, General AI.
Compare the hardware required to build a car vs. the hardware required to build an Optimist robot. Compare the value of the work that an Optimist robot can do vs. the value a car can produce. Simple math, how much does a $15 minimum wage worker cost. The Optimist robot wonât have the sad habit of goofing off.
A $15 minimum wage worker might cost $30,000 per year. If the Optimist robot costs $15K it will be amortized in less than a year. Not only amortized but they will be feeding Dojo to become more efficient and with many more skills.
Steve, the Optimist robots wonât be wining about the JCs, at least not yet. 
The Captain
I am pretty sure that building the robot would be far more difficult. The basics of car design are well known: aerodynamics, speed capabilities, function, acceptable price point, and so on. The recent explosion of electric vehicles and the fact that other companies already have self-driving, pilotless cars on the roads tells me that 1) building them is not so hard, it just took someone to prove the economic viability, and 2) the software is hard, but not insurmountable. Indeed multiple vendors already have it in the field and there are sure to be more coming.
By contrast there are no Optimus or Optimus-like robots doing anything particularly productive at the moment, more than that, nobody really knows what jobs, if any, they will easily and functionally translate to (although I am sure there will be some.) Nobody knows what price people will pay for them, or why, or when. And, contrary to popular opinion, Iâd say that âdoing human thingsâ with some ability is harder than âdriving between the linesâ even with some safety.
Imagine programming a robot to cook a meal, clean the house, watch the children. Or wait tables, wash dishes, eject an obnoxious customer. Or watch patients, disinfect a bedpan, know when to summon a doctor. (There will be simple things they can do, but then that wonât make them worth as much will it?)
So far self-driving cars are pretty much cars. Robots are Roomba and dishwashers, or monstrous flailing arms in factories: single use mechanical devices. Over time, yes, youâre right, they will be feeding each other - but then look how long itâs taken to get relatively simple, straightforward self-driving cars to, well, self-drive.
Long way to go.
This is what I though a few months ago but I changed my mind. The robots we know are not humanoid robots, they are all special purpose robots like the Roomba and dishwashers you mention and the monstrous flailing arms in factories. Optimus is none of that, itâs something completely new that requires a mental reset.
When Musk first talked about Optimus he mentioned the robots would first work in the factory. Believe it or not, that is the key to the mental reset. None of the robots we know are based on machine learning, on AI, they are controlled by algorithms written by humans. The latest version of FSD, v12, has gotten rid of most of these algorithms and relies almost entirely on the machine learning what good humans drivers do. Same for Optimus, they will not run on human written algorithms. They will be taught like children are taught, my watching what adults do.
Teslaâs factories are the ideal robogardens to train the robots! Children do take a decade or two to learn, I think Tesla can cut the robot learning time to 5 to 10 years. The first customers will not be people but factories.
Optimist does take thinking outside the box!
The Captain
2 Likes
Why, exactly? In a car factory, robots are going to observe only a very limited slice of human interaction. At best, theyâll learn only the behaviors that comprise being an auto factory worker - only some of which will translate well to jobs outside an auto factory. Theyâll have relatively few subjects to observe, since auto factories are among the more automated work environments, and Tesla factories especially so.
Your objection has merit, Tesla factories are not be be-all and end-all of robot training. Letâs analyse what I wrote
I used the term robogardens in reference to kindergartens to imply the starting point of their learning, just like children. The key descriptor of the Optimist robot is âhumanoidâ and teaching humans is a time tested method.
Once Optimist has itâs ABCs at Tesla the next step is other factories where they can be taught new skills under controlled conditions. Only later can Optimist be trusted on the general public.
My argument makes sense when you think of Optimist more as âhumanoidâ than as robot. Of course, only time will tell how it all develops.
The Captain
1 Like
I mean, yeah. So again, why would Teslaâs factories be an ideal place to do that? We donât just stick little kids in random adult environments. Or even random older kid environments. They go to kindergarten - a place that is specifically designed solely for the purpose of early childhood education, to be taught a curriculum for that purpose by employees whose sole task is to educate young children.
If the idea is that these robots are going to be âtaughtâ rather than programmed, why just stick them in an auto factory and hope that the things they can pick up there have some relationship to what they need to âlearnâ? Especially since thereâs no reason to think that âauto factoryâ is going to provide them with any better learning opportunities than any other place where people are, much less an environment that was designed to teach them things?
1 Like