Musk throws cold water on Cyberab, Optimus production plans

Stocks A to Z
1

https://www.investors.com/news/tesla-is-elon-musk-lowering-expectations-with-this-cybercab-comment/

Excerpts from the article:

Musk, in a social media post to X Tuesday evening, warned that early production for the Cybercab, which has been touted to be designed for true autonomy, and the Optimus humanoid robot will be slow.

“The speed of the production ramp is inversely proportionate to how many new parts and steps there are,” Musk said on X. “For Cybercab and Optimus, almost everything is new, so the early production rate will be agonizingly slow, but eventually end up being insanely fast.”

Amid Musk’s warnings, Tesla reports fourth-quarter earnings and revenue on Jan. 28. Tesla announced on Jan. 2 it delivered 418,227 electric vehicles in the fourth quarter, down 15.6% from 495,570 in Q4 2024.

Analyst consensus has Tesla EPS falling 38% to 45 cents with revenue declining 3.7% to $24.76 billion, according to FactSet.

To sum up Musk’s comments: “We aren’t doing well today, but BIG, BIG things are coming in the future!”

4 Likes
2

Said every CEO ever in history when explaining a bad report.

3 Likes
3

There’s probably no need to ramp up CyberCab production to high levels this year - need to have a couple of areas of approved Robotaxi service first. Unless they’re making CyberCabs with manual controls for the testing/permitting phases, which seems unlikely to me.

Optimus is going to take what it’s going to take. Boston Dynamics’ Atlas is expected to start at $320k, so with Tesla targeting an order of magnitude less for that, production will need time to ramp. Even the Chinese company Unitree’s H1-2 is $128,900 - and hands are extra.

It’s actually good to get some realism out of Musk on timelines, IMO, anyway.

4

This is no mystery. There is no market for either one of those products. There might be in the future of course, but currently there is almost no point in manufacturing either of these.

1 Like
5

And even if there is a market, the products are not autonomous.

Minor detail.

But you have to think more “futurist” I suppose. Mars, data centers in orbit, more rockets involved, I believe.

Just another iteration of “never mind the goals we didn’t meet from last year, or even weeks ago.”

Looking forward to the earnings call. The last one was relatively sober, I’m looking for more entertainment value in the next one.

The recent “need 10 billion miles” for autonomy was a big funny. This so dwarfs the (real world) miles that Waymo has to get to even where they are today (even after now driving 150 million miles of true autonomy, 10 b vastly dwarfs 150 million).

It’s great to have some competitors making real progress so now there is a point of reference, whereas before there was very little so people can just make up anything with no way to empirically check.

Now we can check: just look at Waymo.

6

Data centers in space.

I haven’t run the numbers, and maybe there a case for it but…

SpaceX is mostly putting stuff in low orbit. A higher orbit is less than 1/3 as efficient in terms of lift and capability.

A data center, to be useful, has to be in geosynchronous orbit - or there have to be a thousand of them, and if your data is on the one that just few past you have to wait another hour for it to be back. Unless all you’re doing is manipulating the numbers you’re beaming up, which seems to imply you also need a data center on planet earth.

But let’s ignore that for the moment. Data centers, as we all have learned (endlessly) require vast amounts of electricity. Solar panels, even the best, will only get you so far. They work for minimal applications, maybe, but data centers?

They also require vast amount of water for cooling, but let’s say that being in cold outer space takes care of that for you. Not to be a wet blanket, but I know the ISS requires significant cooling when it’s in the sun because there’s no atmosphere to take away the heat caused by the solar radiation. (It also requires heating when it’s in shade, so maybe there’s a way to “heat sink” and bring it back later?)

And then there’s latency. Maybe not a big deal if you’re just using the data center to, you know, hold data, but if you’re doing anything real-time like customer transactions or missile calculations that 2 second round trip is a real killer.

I’m not even going to get into lofting racks of chips into space requiring thousands of tons of lift, not to mention assembling them in space into some kind of coherent module, but maybe that’s all been figured out.

Or maybe it’s all just super prep for the upcoming SpaceX IPO, where Musk investors don’t bother with mundane and practical details in favor of grand visions of the future. Dunno. It’s worked out well for them so far, I’d have to admit - but at some point reality sets in (see: predictions of FSD for the past decade, dancing robots, etc.)

Good luck on this one boys, but I think I’ll sit it out. (While not dismissing the idea that there is a nicely profitable business in satellite launch and other “spacey” ventures as may come about. Humans transporting to Mars? Not one of them, probably.)

1 Like
7

Since some people don’t ever believe Musk/Tesla, maybe they’ll believe what Nvidia says?

Instead of relying on fresh water for cooling through evaporation towers, as many Earth-based data centers do, Starcloud’s space-based data centers can use the vacuum of deep space as an infinite heat sink.
Emitting waste heat from infrared radiation into space can conserve significant water resources on Earth, since water isn’t needed for cooling. Constant exposure to the sun in orbit also means nearly infinite solar power — aka no need for the data centers to rely on batteries or backup power.

and:

Starcloud projects the energy costs in space to be 10x cheaper than land-based options, even including launch expenses.

2 Likes
8

The way datacenters work is that your data isn’t on just one of them. It is spread out on multiple datacenters, and there are multiple copies. That’s why you can open up your Gmail in Tokyo.

LEO is around 500 miles altitude so that’s probably good enough for most applications. There is such a thing as a sun synchronous orbit where the the data center would be in LEO, but in sunlight 24/7. So no heavy batteries needed.

Some serious players are looking into this, like Google and Microsoft. So they must think there is something there.

Still a few things I don’t understand though. One, you have to get to space, which means it will cost a lot. Then you have to put everything together. That’s solvable, but a hard problem. Then it is hard to maintain.

Also cooling. Space is cold, but it is a vacuum, right? And vacuum is an insulator. So I don’t know how you do it without heavy cooling components.

Seems like it was be easier to buy up land in Haskell Texas, but maybe not.

1 Like
9

That’s not Nvidia, that’s a blogger at the company, whose job it is to find interesting things that talk their book. This is one of them. She never mentions, for instance, that the total staff of this amazing enterprise numbers … 12.

The cool animation on the splash page linked shows s 5gw center-in-space, which is cool. It also shows that the solar panels needed to power it are … 4km x 4km, or more than 6 square miles. That’s certainly quite a few truckloads of panels to get up there. And to assemble. And to fix if anything goes wrong. How often does something go wrong? Often enough that data centers on the ground are staffed 24/7/365.

Now, like McLovin’ and others, I’m not saying there’s nothing here, just that there are stupendous hurdles and those talking it up are all those with something to gain by it. SpaceX ipo, Starcloud, Nvidia, even companies with ground-based centers in play or proposed because they’re in the “how high can this go?” mode and likely won’t stop until they’re forced to.

But I’ve now done some reading. Here is what looks to me like (pardon the appropriation of a phrase) a fair and balanced view from someone with some experience in this sort of thing:

Before we get nerd sniped by the shiny engineering details, ask the only question that matters. Why compute in orbit? Why should a watt or a flop 250 miles up be more valuable than one on the surface? What advantage justifies moving something as mundane as matrix multiplication into LEO?

That “why” is almost missing from the public conversation. People jump straight to hardware and hand-wave the business case, as if the economics are self-evident. They aren’t. A lot of the energy here is FOMO and aesthetic futurism, not a grounded value proposition

He has all kinds of snazzy charts and analysis, which I will reproduce one or two here to entice you to look further, if you care. (And after the charts I will give you a short snippet of his final conclusion, which may surprise you.)

IMG_0792
IMG_07921640×2360 212 KB

IMG_0793
IMG_07931640×2360 219 KB

Here's the headline result: it's not obviously stupid, and it's not a sure thing. It's actually more reasonable than my intuition thought! If you run the numbers honestly, the physics doesn't immediately kill it, but the economics are savage. It only gets within striking distance under aggressive assumptions, and the list of organizations positioned to even try that is basically one.

Hint: that “one” is Musk.

1 Like
10

Nothing confusing companies with the people running them. He he