Jeff Bezos throws his hat in the ring for an orbital data center megaconstellation, too
- Thread starter JournalBot
- Start date
This line really should not be so matter of fact. It is an extremely speculative statement. It's not clear how this would improve, let alone enable, AI. It also doesn't really say what kind of AI it would instantiate, as current LLM's certainly don't need to be in space. It's not even clear in-space processing is that good of an idea in general. Indeed, there's a lot to argue against it.
Upvote
216
(222
/
-6)
Someone call me when any of these rich idiots have any idea how they are going to cool megawatts of computers in the vacuum of space.
Lines like these are about stock price manipulation.
Upvote
219
(241
/
-22)
It is not that difficult in reality. Especially if you add some heat pumps to increase the radiator temperature. Scott Manley has a nice video about it.
Bigger issue is all the trash being lifted up there and then deorbited. But when have rich ever bothered with impact of their BS on others. If they cared about others they would not be rich.
Upvote
-12
(51
/
-63)
Upvote
15
(24
/
-9)
Rocket and AI guy determines that rockets and AI are critical to the future. News at 11.
Upvote
102
(107
/
-5)
ubercurmudgeon
Ars Tribunus Militum
Bezos joining the chorus of tech billionaires saying that "AI isn't a bubble at all, we just need another trillion dollars to launch it into space, then it'll make sense" is the least surprising news in a long time.
Upvote
160
(162
/
-2)
Heat pumps are not magic. The limit isn't even moving the heat from compute hardware to the radiator. The limit is how quickly the radiator can radiate the heat into space. You're going to quickly saturate the radiator unless it's massive.
It doesn't answer the fundamental question of why you even need this in the first place. Communication satellites solved a problem that has plagued humanity since it first started organizing into civilizations. There is no alternative to satellite comms when you are away from terrestrial networks. It's why comm satellites were one of the first practical applications of space.
Data processing doesn't have that need. If you have a network, you don't particularly care about where the numbers are being crunched. The data will get back to you. Which is why data processing has, typically, been done on the ground and why spaceborne data centers haven't really been attempted. It's not that we can't, it's that there's no point.
Upvote
190
(205
/
-15)
Power you can irradiate scales with T^4 so you can dump a lot of power with sufficient determination (of course all that determination adds extra weight).
I completely agree that the whole thing seems utterly pointless, but maybe I am just too stupid to understand thinking on their level.
Upvote
49
(53
/
-4)
Use of space is non-territorial to promote advancements for the human race.
This... this is not what I thought that statement meant.
I mean, it's nice that we don't have global superpowers warring over who can put their own special satellites in orbit in order to beat the other guy into submission, right?
This... this is not what I thought that statement meant.
I mean, it's nice that we don't have global superpowers warring over who can put their own special satellites in orbit in order to beat the other guy into submission, right?
Upvote
18
(19
/
-1)
Cool, thanks for the link--I need to watch it in full later. Reading the transcript--the catch with his analysis is that 20kW as a base point of discussion...is far FAR too low. A single GB300 rack from Nvidia pulls like 140kW. And that is just the rack, not all the other necessary equipment a satellite needs. And of course you need to have ~200kW of generated on-board power onboard to start with to power the rack as well as the radio etc equipment.
IIRC, the depreciation schedule on server GPUs is ~100% in 5 years, as a matter of business. Which means a LOT of deorbited space junk and very quickly. In addition to all the pollution to get it up there. And also, LEO is extremely crowded to start with--and the bigger you need to make a satellite, the more likely it is to get hit by space junk.
Upvote
135
(141
/
-6)
They know it isn't viable to have datacenters in space. They just lie.
I don't get why, when most people talk about these idiots, we tend to use pointless euphemisms:
Empty promises? Lies.
Overselling? Lies.
Untruth? Lies.
Alternative facts? Lies.
Even when reporters try to do their job and talk to politicians and billionaires and the rest of the lot, they never tell them "you are lying". They always say "it doesn't sound like you are making an accurate statement" or crap like this.
Upvote
112
(127
/
-15)
Well, because the licensing works on a first come/first serve basis and you have to demonstrate you do not interfere with folks higher in the chain than you there is an incentive to file very large speculative constellations to block out your rivals. If you are only planning to launch a thousand satellites it doesn't matter that you wont hit the milestones to launch fifty thousand, just that you have forced the next company to prove they wont interfere with your fifty thousand satellites.
On the thermal side its all doable - no one in the industry is suggesting it can't be done technically, just that it can't be done economically. It all comes down to what $/token you think these things can offer and how that compares to a terrestrial application, and also who would actually use them over running their own datacenters. Not a problem for SpaceX since they are the end user, but it is a problem for the others.
On the thermal side its all doable - no one in the industry is suggesting it can't be done technically, just that it can't be done economically. It all comes down to what $/token you think these things can offer and how that compares to a terrestrial application, and also who would actually use them over running their own datacenters. Not a problem for SpaceX since they are the end user, but it is a problem for the others.
Upvote
27
(29
/
-2)
Can we stop pretending that the FCC chairman, who has for years been advocating for SpaceX and accused the previous FCC of "regulatory harassment" against them, is some principled independent arbiter here?
Upvote
103
(110
/
-7)
Upvote
31
(41
/
-10)
Instead of this, they need to concentrate on reducing the power demand for data centers.
I'm waiting for something that can do what a human brain can do with the same or less mass and power.
I'm waiting for something that can do what a human brain can do with the same or less mass and power.
Upvote
29
(29
/
0)
They need to make a second Doomsday Clock, but this time for Kessler Syndrome rather than nuclear war.
Upvote
7
(17
/
-10)
Never gonna happen. Just like Mars colonies and all the other space-based fantasies these fuckwit techbros dream about.
It's fucking exhausting that people give these delusional assholes any credence whatsoever when they puke out such utter nonsense.
It's fucking exhausting that people give these delusional assholes any credence whatsoever when they puke out such utter nonsense.
Upvote
51
(61
/
-10)
“Space-based data centers will be a complement to terrestrial infrastructure.”
Okay, but why? Do the computers think more smarter after being tangentially exposed to rocket science? Is it the awe of being in space that drives them?
Well, come to think of it, yes. Except "them" is people with space-related businesses and the thing being improved is their share price. Yes, the investors do think the computers will become more smarter in space. Lines can go WAY up if the sky isn't even the limit. Stonks.
Okay, but why? Do the computers think more smarter after being tangentially exposed to rocket science? Is it the awe of being in space that drives them?
Well, come to think of it, yes. Except "them" is people with space-related businesses and the thing being improved is their share price. Yes, the investors do think the computers will become more smarter in space. Lines can go WAY up if the sky isn't even the limit. Stonks.
Upvote
42
(44
/
-2)
Upon rereading, you are right. Instead of "artificial intelligence," the "birth of SkyNet" would work too. I also considered "mass surveillance" but we already have that planet-side.
I'm generally a fan of Eric Berger's writing, but this is a fair call-out.
Upvote
16
(19
/
-3)
The 20 kW was just the initial starting point, based on an estimated Starlink v3. He does go into comparing power needs of server racks, and how that affects the cooling needs.
This, not the cooling, is the reason data centers in space is a stupid idea at the current state of space technology. Costs to build and launch, lack of serviceability and upgradeability, all means a lot of resources are spent putting things into space when it could be done much cheaper on earth. The only reason they want to do it is to avoid the regulations needed to build them on the ground.
Upvote
28
(36
/
-8)
All these rich assholes know is stonk manipulation. How do you think they got rich? Actual knowledge and ability?
Upvote
13
(22
/
-9)
The 20kw number was how much a Starlink passively radiates without a dedicated radiator. Manley was priming people with that example to help us understand how much heat satellites are already dispelling. He noted that 4kw do get dispelled via Starlinks antennas though, and probably more via lasers.
From that radiator-free design he begins considering what would be needed to scale up. He notes that 20kw is around what a traditional server rack consumes, but that ML servers will likely be 5x or more.
Upvote
39
(39
/
0)
Labor exploitation? Although that's been a thing for like forever.
Upvote
18
(18
/
0)
Yeah, looks like these guys are literally paid to create Kessler Syndrome ... especially as sun-synchronous orbits are basically a type of polar orbit, which means that all these "datacenters" will have a direction of travel that deviates by ~ 90° from 99,9% of the stuff that already is up, which maximizes delta-v and therefore impact energy and debris creation rates.
Conspiracy mode on :
They are paid by aliens to lock humankind in on earth under a scrapyard-filled sky for the next 200 years while they reposition their fleet into our solar system.
Conspiracy mode off.
Last edited:
Upvote
-2
(11
/
-13)
It was just an estimate. You can ramp up cooling and having a flat satellite is a good choice for cooling.
As to power, there is a reason that every proposal is for sun synchronous orbit (always exposed to the sun). So you need a lot of solar panels but that is well developed tech. Especially, when it has to survive 5-10 years max due to aging of other hardware.
I would be more worried about radiation induced errors, on the other hand, who would notice those in LLMs
Upvote
-8
(9
/
-17)
The shade created by the huge solar array will provide sub-freezing temps for "cosmo-thermal cooling" (opposite of geo-thermal heating).
Upvote
0
(5
/
-5)
All those additional bit-flips from all that extra cosmic radiation orbital data centres get exposed to could speed up the evolution of genetic algorithms, you know ...
/s
Upvote
45
(45
/
0)
Ars is the perfect publication to either explain how these space-based AI data centers make any financial sense, or to show that it’s absolutely stupid (my bet is on the latter). Frankly, I’m disappointed that there hasn’t already been an investigative article on this, instead of just parroting the space company press releases.
This shouldn’t be hard. I feel like anyone in the space industry should be able to do some back-of-the-napkin math to show that it doesn’t make sense even with very generous assumptions (i.e., assume Starship is working and launches cost roughly what they’re currently projecting).
Some things to include:
This shouldn’t be hard. I feel like anyone in the space industry should be able to do some back-of-the-napkin math to show that it doesn’t make sense even with very generous assumptions (i.e., assume Starship is working and launches cost roughly what they’re currently projecting).
Some things to include:
- you have to compare against terrestrial data centers with the same latency (distance) to cities as low earth orbit
- whatever magic hand-waive they want to do for collecting heat (to make waterless cooling work in space) is also available for terrestrial data centers. Current data centers “need” water for open-loop cooling because that’s very cheap, but obviously they won’t have that in space.
- at least mention the problem of handing off user sessions, since the data center that was “close” to the user a few minutes ago is now over a different continent; for long AI queries, this can happen between when the user asks the question and when an answer is finished.
- AI data centers “need” lots of grid power, but they’ll be forced to provide their own power in space. Compare that price to them providing their own power for terrestrial data centers (yes, solar panels in space can get sunlight 24/7, but I suspect considerably cheaper to install panels on earth)
Upvote
109
(111
/
-2)
Don't give them talking point ideas.The shade created by the huge solar array will provide sub-freezing temps for "cosmo-thermal cooling" (opposite of geo-thermal heating).
Upvote
-1
(3
/
-4)
Perhaps we need to figure out the space junk problem before approving 10s of thousands more SATs. And decommissioning by burning up in atmosphere doesn't appear to be a reasonable approach.
Upvote
10
(14
/
-4)
It'll be called "Skynet", of course.
I mean, obviously it's 20 years later than it was supposed to happen, and it's happening in the most stupid way possible, but Terminator got the apocalypse mostly right (without the addition of climate change, too!).
Then again, this is fucking Bezos who's got more rocket envy against Musk than most out there. So naturally he's going to unzip his equipment in a megaconstellation pissing contest jumping on the bandwagon that's traveling at supersonic speeds to total environmental hell both on Earth, and above it.
The sad part is that the on-earth part will take a shit-ton longer to fix itself once we're gone than most of the shit up in orbit. But at least the space (traffic) jam is comparatively faster at correcting itself.
Gonna take the Earth a few millennia to get over the case of homo sapienitis it contracted...
Upvote
-2
(2
/
-4)