An Economic Ode to Optimus

Investment Analysis Clubs Macro Economic Trends and Risks
21

Not really. I’m asking whether or not there exist any jobs in Tesla factories today that are done by a human, that an Optimus could do instead of that human. Can you give me an example?

My take is that there aren’t any such jobs in a modern automobile assembly line, because even the best AI brain technologically achievable today or in the near future requires isn’t good enough. All the jobs that could have been handled by Optimus’ brain were replaced by robots long ago - at least in automotive assembly lines.

And that’s the key - the humanoid form doesn’t really offer much advantage to robots in an auto factory. The whole point of assembly lines is that even the humans don’t need to be flexible, and don’t need to move around very much. They do one job (or related suite of jobs) in one place. And a factory setting - perhaps more than any other setting - is generally designed to minimize the types of obstacles to machinery and wheeled carts moving around.

I have no problem with Musk dreaming of a day when humanoid robots capable of doing useful work autonomously might exist. They have been a staple of science fiction for more than a century now. What’s always held them back, though, is the brains. There’s little use for a multipurpose body if the brain isn’t capable of utilizing that flexibility - and we are nowhere near the point where robot brains make it worth building a humanoid body for that brain.

Where the point isn’t to replace humans, but simply have the robot be a tool for humans in a job, there are going to be use cases. NASA can find use for a humanoid robot, because they can devote an entire team of humans to monitor and operate and intervene with that humanoid robot. It’s like any other drone - some degree of autonomy, but you can load it up with human brains in real time. But we’re nowhere near the point where a humanoid robot can replace even unskilled or semi-skilled workers, because the brains aren’t there.

Or, again, perhaps I’m wrong about that. Is there a job that currently exists on the Tesla factory floor that Optimus 1.0 could take over from a human?

2 Likes
22

The video of front end assemblies is one such job. I suspect there are a dozen of not a hundred more. But, I’m not an expert of Tesla factory processes. So … :slightly_smiling_face: you’ll just have to trust me!

Again. I see a POV difference. I see your and others POV as demanding a robot, perfectly designed for a specific task. As you say we already have had for a long time “dumb” robots like this.
Musk and the other “humanoid form factor” robot developers POV is for a robot that is less perfect for one specific task, but flexible enough to fit into multiple tasks and perform them efficiently.

Is Optimus Teslabot there yet? No.
Is it getting there? IMO, yes.
When? I’m gonna go with more quickly than some think (2050) and less quickly than others believe (2024).

We’ll see. Our opinions don’t really matter.
Counting the Austin company’s Valkyrie, there are at least 6 humanoid robot developers.
They are in a race to get first mover advantage.

I’m thinking Digit, due their building a factory to produce 10,000 bots per year, is a front runner.
Tesla has production facilities, but hasn’t said if they are “mass” producing bots.
The others are all “in planning and development” as far as I can tell.

:slightly_smiling_face:
ralph

2 Likes
23

Not at all. It’s not a demand, or a POV.

My assessment is that robot brains aren’t advanced enough, and won’t be advanced enough for many many years, for a humanoid form factor to be worth building as a commercial product. It’s a great R&D project, for sure - which is why Boston Dynamics converted their Atlas robot away from commercial product and into a pure R&D internal program. But we’re a very long way away from a brain that is flexible enough to make it worth trading away the benefits of a task-specific body design for a generalist body type. And if that brain ever does get developed, it’s far more likely to come out of the dedicated AI shops (possibly including Musk’s own) than out of a side project of an automotive and energy company (one where AI and robotics aren’t even part of the company’s mission statement).

Which is why an “economic ode to Optimus” is rather off the mark, IMHO. If we had an AI brain that was capable of directing a humanoid robot body to the level of a human unskilled or semi-skilled worker, then such a thing would be incredibly useful. But we don’t, and won’t for many many years.

2 Likes
24

Can I change “POV” to “Vision”?

Musk and the other 5 companies listed, have a VISION for a generalist robot with a humanoid form factor. They COULD just build an iterative, incremental stationary “dumb” robot.
But they have a different vision.

This seems a major bottleneck for sure.

So, should Musk and the others abandon their “projects”?

Agility Robots, Digit, plans to mass produce Digit within a couple years.
I invite you to see this PRESS RELEASE:

{ CORVALLIS, OR – September 18, 2023 – Agility Robotics, creator of the groundbreaking bipedal robot Digit, today revealed that the company is opening RoboFab™, a robot manufacturing facility in Salem, Oregon with the capability to produce more than 10,000 robots per year.

Initial construction of Agility’s 70,000 square foot robot factory began last year, and it is set to open later this year. Agility anticipates production capacity of hundreds of Digit robots in the first year… }

Hundreds in the first year.
So, they seem to think they’ll have a working, economically viable, humanoid bot, doing generalist type tasks … In a couple years.
Ie not “many many years”.
This, IMO is the FIRST PRINCIPLES THINKING process.
Which in some ways is “baby steps”.

Will this cohort of Digits have a quantum compute brain?
Very unlikely. <Wry sarcasm?>
Agility Robots seems to think it’ll be “good enough for release”.
I expect it to be slightly more generalist (ie intelligent) than the stationary “dumb” bot.
Ie FIRST PRINCIPLES.
Have an “end product”, but also have functional “deliverables” at “delivery dates” along the way.
As “better” is developed, it will be added to the current product and we’ll get a “model refresh”.

Don’t let “perfect” be the enemy of “good”. By Voltaire?

Boston Dynamics, IMO, has a different vision.
They (DARPA) are not interested in “commercial”. :biting_lip:
They vision a more military, security, and law enforcement application.
IMO.

I own some Tesla. I do NOT think of me as a fanboi.
I try to keep an open mind, and not get emotionally attached.
So far, I have faith that MY vision of Tesla “investing thesis” is intact.
LOLOLOL.
If the thesis changes, how will I respond?
Right now, I’ve a LTBH position, and a trading position, with Cash Secured Puts and Covered Calls.
So I could be biased? LOLOLOL

We’ll see.
:ukraine: :israel: :heavy_check_mark:
ralph

3 Likes
25

If a company’s goal is a completely automated assembly line, you don’t really need an AI brain. All you need is programming to do the limited number of tasks you need done. Perhaps you could add to that some mobility capabilities, whether it be walking (like Honda’s Asimo has been doing for a couple of decades - or like Boston Dynamics, ummmm, dog format) or more likely a simple tracked vehicle which doesn’t need to worry about the complexities of balance.

Frankly, I wonder why a human form is all that important. If you no longer have humans crawling all over your assembly line you don’t need to design around them. You can use smaller spaces or bigger spaces or tighter clearances.

—Peter

2 Likes
26

I did - and I’ve seen Digit before. It’s nothing special, and I don’t expect Agility Robots to amount to very much - for exactly the reasons we’ve discussed here. Amazon and some other investors have been willing to dump money into it to see if they can make it work. But without better brains in those robot heads, there’s just not enough value compared to all the other robot form factors that are better designed to do things in a factory setting, like Amazon’s doing. I think Digit ultimately loses out to a much more efficient and better “bespoke” robotic warehouse worker in the end (or itself ‘morphs’ into a less humanoid version of itself that is better adapted to a specific job).

It’s not a “should” question, and it’s not the same answer for Musk and the others.

There’s enormous commercial benefit to continuing to research robotic solutions to various industrial and commercial needs - both the “bodies” and the “brains.” That’s why there are so many robotics firms out there pushing the cutting edge, and why they continue to attract investment. To say nothing of academic and other non-commercial research into those fields.

But that’s very different from saying that any humanoid robot is ready for near- or intermediate-term commercialization, which is what has been suggested for Optimus (and what Agility is trying to do). And it’s different from the idea that a car and energy company’s side project in robotics is at all likely to be one that contributes meaningfully to either the field of robotic development or to that company’s bottom line.

Google has done lots of this type of stuff in their Moonshot factory (now X Development) and their Advanced Projects teams - side projects that were gambles on ground-breaking technology. Whether Glass, or Loon (balloon network), Wing (aerial drones for delivery), Malta (molten salt energy storage), Makani (clean energy from kites), Tango (an internet of things project), Ara (a modular cell phone), or Jacquard (a smart fabrics technology) - all of them did some interesting work but never resulted in a commercially viable product. None of them had any material impact on the value of Google/Alphabet.

Again, you noted upthread that you would be happy if someone could poke a hole in this investing thesis. The investing thesis is based entirely on the idea that if we had different technology than exists today and is likely not to exist for quite some time, and if it does exist is not likely to be developed in-house at a car and energy company or exclusively available to that car and energy company, these types of things would be commercially viable. That’s not a strong basis for an investing thesis for Tesla.

Tesla has a solar division, but investing in Tesla probably isn’t a great way to get exposure to new developments in solar technology any more (if ever it was). It’s clearly a side project to the company, and a moribund one at that - so folks who have real talent and skills in the solar industry are likely to go to other companies that are more dedicated to that project. And that’s a market segment that is at least nominally part of Tesla’s mission.

I think the same fate awaits Tesla’s robotics program (which isn’t even part of Tesla’s company mission). Tesla needs to have a lot of resources to work on solving a very specific (but no longer cutting-edge) AI problem in vehicle autonomy. To get talent and justify the resource spend, though, it needs to have a much bigger “playground” for those engineers than just working out the last few details in FSD on a problem that Musk keeps projecting will be solved in the next year or so. But Optimus doesn’t make any sense for Tesla outside of being a nice AI day talent-attractor. I think Tesla’s not really going to be in the robot-building business, any more than it’s really in the solar roof manufacturing and installation business - because the time-frame for having commercially viable humanoid generalist robots is far longer than the time-frame for finishing (or abandoning) FSD.

27

Guilty! I own TSLA for that “vision”.
I’m aware of the holes that have been exposed. Those holes are the same ones exposed a … a year, 2 years, 5 years, a decade ago, and continually repeated with some “model refresh” from the critics?

And Tesla/Musk keep First Principles Thinking and keep proving the Shorts wrong?
(Although the shorts did ok in early 2022?)

An opinion. :slightly_smiling_face:
And, I disagree. :slightly_smiling_face: For the explanations and examples given.
I SWAG I’m correct! :slightly_smiling_face:

I continually watch Tesla news for anything that will suggest an investment change is needed.
The most obvious, to me at least, is China and Geopolitical risk.
I don’t see any “tech” news that suggests to me that I need to change my position… Yet.

:person_shrugging:
ralph

SWAG - a guess based at least in part on “evidence”.

1 Like
28

You seem to have jumped from a Tesla factory to an auto factory.

I’m very skeptical that there is a commercial case for selling robots to use in a factory anytime in the next 3 or 4 years. But Tesla can take the long view and will delivery (or not) on whatever time scale they want.

But Tesla has lots of factories (or parts of factories) that are not auto assembly lines (which have probably already been fairly well optimized with expensive large stationary robots). But they still build things like Superchargers, wall chargers, powerwalls besides batteries and components that are built before auto assembly (seats, motors, power electronics, etc.) When/if needed they could probably run a non-auto assembly line at 5 or 10% over the needed throughput and give the robot engineers a day or two a month to try out a robot doing/learning/debugging in a real factory after simple training succeeded at a workbench mockup. This is easily years away.

In the mean time maybe Optimus can learn to sweep the factory floor

Mike

2 Likes
29

Only the human being and the computer are multi-jobbed. Well that and the Rotti but we are not talking about guarding sheep or pointing at dead game.

The multi-purpose humanoid has a huge upside in working on an AI network for replacing human labor in all factories of a certain size—the economics work. Just designing one more mechanical arm does very little. That is unneeded.

This is a massive industry that Musk is uncovering. He will begin to immediately save on the labor of 10s of thousands of workers in his plants while expanding the number of plants Tesla has. Currently, Tesla employs 127k employees. Like I said 10s of thousands will be replaced over the next several years.

Because human labor is so expensive fully formed robots can replace them. This broadens out to other industries if designed for that. The costs come down with production ramping up.

Skipping this opportunity is a mistake.

If all we do is design mechanical arms we get less of a return decades out.

We already have mechanical arms and still mobile employees are necessary.

1 Like
30

You hit the nail on the head. Tesla’s vertical integration gives it the advantage of being able to develop and deploy humanoid robots in-house without having to deal with customers and the press. Elon Musk said at the very beginning that Optimus would start life in Tesla’s factories. At first I was intrigued by this statement but as events unfold the beauty of this logic is showing itself. Why fight the infidels?

Translation of the Arab proverb, “The dogs bark but the caravan goes on its merry way.”

To be humanoid, form factor is not enough, the brain (GAI) also needs to be humanoid in the sense that it needs to learn what it takes to keep humans alive and safe. That harks back to Isaac Asimov’s three laws of robotics.

  1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
  2. A robot must obey orders given it by human beings except where such orders would conflict with the First Law.
  3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

Asimov didn’t have the tools (the Positronic Brain). Learning via neural networks is the first real approximation technology has to deal with this issue. Optimus must learn to coexist with humans and Tesla’s own factories are good sand boxes for this development.

Americans suffer from the Instant Gratification Syndrome [IGS]. They want Optimus to be born Omniscient, Omnipotent, and OmniEverythingElse. God like.

The Captain

It Ain’t Happening

that way…

4 Likes