Continuing my rant about how Space-based data centres need to address the massive hurdle of cooling and almost no one is saying anything about it, here's the CEO of Voyager Technologies pointing...
Continuing my rant about how Space-based data centres need to address the massive hurdle of cooling and almost no one is saying anything about it, here's the CEO of Voyager Technologies pointing it out too. They were one of the first to say they were going to work on Space-based Data Centres, and are also building a lot of neat space hardware, largely focused on space stations.
As an ignorant person, I assumed the main benefit of space data centres would be easy cooling. Learning that that is basically the opposite makes me wonder what benefit there is to space data...
As an ignorant person, I assumed the main benefit of space data centres would be easy cooling. Learning that that is basically the opposite makes me wonder what benefit there is to space data centres? Less rules? Less chance of citizens protesting? If Elon is involved I assume it's mostly lies and grift.
It makes no logical sense. Full stop. With the hyper loop you could say “well it it was possible it would be faster even though we don’t need that” as SOME level of reason. This just sounds like...
It makes no logical sense. Full stop. With the hyper loop you could say “well it it was possible it would be faster even though we don’t need that” as SOME level of reason.
This just sounds like something that might make sense to the layman and 100% does not.
It is easier to get power on earth. Easier to cool. Easier to maintain. Easier to build.
You would need to be out of space on the surface AND in the ocean for this to make sense or be at the point where orbital lift was trivial (skyhook, space elevator, whatever) for this to even start making sense
Okay, that's kind of what I assumed. I should have thought more logically about it. My brain was just like, space is cold, data centres need cooling, so obviously that's why they are suddenly...
Okay, that's kind of what I assumed. I should have thought more logically about it. My brain was just like, space is cold, data centres need cooling, so obviously that's why they are suddenly obsessed with it.
Space is cold in the sense that the average temperature of the few molecules floating around are close to absolute 0. The main mechanism for one cold thing to cool another thing (water cooling a...
Space is cold in the sense that the average temperature of the few molecules floating around are close to absolute 0.
The main mechanism for one cold thing to cool another thing (water cooling a data center or wind cooling your skin), heat actually needs to be transferred between molecules in a process called convection.
There are so few molecules in space that the heat doesn't have anywhere to go. If you were instantly teleported to space without a suit it wouldn't actually feel that cold, you would just die from lack of oxygen.
Things in space still lose heat to radiation, but it is minimal when compared to convection on earth. Your body would eventually freeze due to heat loss from radiation, but it would take a while.
More specifically the vacuum would cause every liquid you have to boil and you would dessicate before you eventually freeze. A form of mummification. Regardless, you're in for a very bad time.
More specifically the vacuum would cause every liquid you have to boil and you would dessicate before you eventually freeze. A form of mummification.
Actually, it's the opposite! Most particles in space are moving very quickly, so their average "temperature" (kinetic energy) is quite high. There's just not many of them.
Space is cold in the sense that the average temperature of the few molecules floating around are close to absolute 0.
Actually, it's the opposite! Most particles in space are moving very quickly, so their average "temperature" (kinetic energy) is quite high. There's just not many of them.
Oh thank you for the correction! I thought that temperature was always an average of all particles and that space temperature was very low. So is mostly empty space considered to be a high...
Oh thank you for the correction! I thought that temperature was always an average of all particles and that space temperature was very low.
So is mostly empty space considered to be a high temperature since the few particles have high kinetic energy, or is temperature measured in a way that the temperature would still be close to absolute zero?
The temperature of the matter in space is high, but since there's not much matter the thermal energy is low. You might be thinking about the heat of the radiation in space - there is ambient light...
The temperature of the matter in space is high, but since there's not much matter the thermal energy is low.
You might be thinking about the heat of the radiation in space - there is ambient light coming from the cosmic microwave background, but it's only a few °K. As such radiators always trend toward that temperature (except in the presence of other sources of hotter radiation like planets and starlight).
For someone like Elon, advocating for space data centers is a way to justify the SpaceX purchase of xAI (which includes Twitter), since that company is burning money. SpaceX is primarily a...
For someone like Elon, advocating for space data centers is a way to justify the SpaceX purchase of xAI (which includes Twitter), since that company is burning money.
SpaceX is primarily a government contractor. I suspect this is all simply a ploy to get U.S. taxpayers to subsidize the development of large language models.
The idea is that if you put solar panels in space, they produce consistent power all the time, no dealing with night time, cloudiness etc. Plus the atmosphere absorbs a lot of the sun's energy so...
The idea is that if you put solar panels in space, they produce consistent power all the time, no dealing with night time, cloudiness etc. Plus the atmosphere absorbs a lot of the sun's energy so if the panels are in space you don't have to deal with that.
That said it costs like $10,000/kg to put things in space (maybe less these days) so the world in which it makes economic sense to build space data centers looks a lot different than the world we live in now.
I feel like we'd be seeing a lot more offshore datacenter barges in international waters if that were a significant driver, though? Those would have a whole stack of their own logistical problems,...
I feel like we'd be seeing a lot more offshore datacenter barges in international waters if that were a significant driver, though? Those would have a whole stack of their own logistical problems, but I'd bet you can build them with 2x-3x redundancy (physically and legally) and still come out hundreds of millions cheaper than launching them into space. Let's say one ship flagged in Liberia and moored off the coast of the Philippines, one flagged in Vanuatu and moored near Mexico, one flaged in the Marshall Islands and moored off India - you're gonna cause a decade of legal gridlock trying to shut down even one of those, let alone all three.
Until the US Navy and Coast Guard decide to chase down your ship and board it due to "irregularities," then confiscate the ship! I agree that lack of regulation isn't the goal here. But even if it...
Until the US Navy and Coast Guard decide to chase down your ship and board it due to "irregularities," then confiscate the ship!
I agree that lack of regulation isn't the goal here. But even if it was, like someone else pointed out, might > law. Quite often.
Oh yeah for sure, and not just state level actors - good old fashioned speedboats-and-AK47s piracy would be a legitimate concern too. One of the few genuine advantages of space vs ocean is...
Oh yeah for sure, and not just state level actors - good old fashioned speedboats-and-AK47s piracy would be a legitimate concern too. One of the few genuine advantages of space vs ocean is limiting the number of organisations/governments who’d be physically capable of interfering, actually!
I will say that I think you’d need to really piss off the US for them to go full legally-questionable naval confiscation, though, and I get the impression that the game for a lot of billionaires’ dealings is plausible deniability and a legal fig leaf rather than a direct middle finger to the government that’d prompt a targeted response. You’re giving yourself the ability to bury the plaintiff in paperwork and convince the US courts that it’s not their problem or jurisdiction when someone sues you, more than poking the bear and going against the US itself, I think.
But yeah, edge cases and practicalities aside I’m not suggesting that extrajudicial seafaring datacenters are actually a good idea - I just think we’d see them as a stepping stone to orbital ones if the actual demand were there.
Regulation doesn't depend on the physical location of a server, it depends on the physical location of a headquarters or citizenship or whatever. As long as spaceX can be reached by a court order,...
Regulation doesn't depend on the physical location of a server, it depends on the physical location of a headquarters or citizenship or whatever. As long as spaceX can be reached by a court order, it can be regulated.
Literally the first sentence of SpaceX's statement is
SpaceX has asked US regulators to approve a new satellite system
That's not really how regulation works. If the company has a US presence or does business with US citizens, it can be regulated. Those data centers in space still need to actually be communicated...
That's not really how regulation works. If the company has a US presence or does business with US citizens, it can be regulated.
Those data centers in space still need to actually be communicated with and run by a company. It's not like you could run a pirate data center and the company running it would be untouchable by US law just because their servers aren't physically reachable.
Yeah I understand that argument in theory, but without knowing the math, I assumed the cost wouldn't be worth it unless you also had easy cooling. Thanks!
Yeah I understand that argument in theory, but without knowing the math, I assumed the cost wouldn't be worth it unless you also had easy cooling. Thanks!
To take off my rage hat and put on my optimistic futurist hat, here are some potential benefits of space-based data centres (assuming you somehow solved cooling): Effectively infinite solar power,...
To take off my rage hat and put on my optimistic futurist hat, here are some potential benefits of space-based data centres (assuming you somehow solved cooling):
Effectively infinite solar power, so you can run forever with low cost, hypothetically eventually covering the heavy cost of launch and deployment through those savings.
No environmental damage on Earth from the heat emissions and real estate being taken up.
Security from weather and other turbulent events, as you only have to worry the solar wind (oh hey that's actually pretty major the further out from Earth you get but I'm being positive here...)
The one I'm most interested in: deployment of edge-computing solutions. Rather than massive data centres, I think there's actually a lot of very, very interesting use cases you could have from making a space "cloud" centre that provides compute power to remote assets on the Moon and farther. People forget that even the Moon has notable delays - a quick google got me 2.7s to send a radio or light message to and from the Moon. This problem gets much worse when you start to take into account things like:
Line of sight (what do you do when the Moon's blocking your comms point with Earth?)
Bandwidth (we actually can't just beam terabytes of data back to Earth easily - and that could be a huge problem for deploying most AI/ML systems in space)
So with an edge-computing solution - call it a data centre orbiting the Moon - you could enable a lot more autonomous analysis and navigation solutions on the Moon and further. You could also parse space-telescope data coming in before it hits the choke points like the Deep Space Network which is already at capacity. Anything we can do to try and make it easier to send deep-space data back home will be a huge improvemnet.
So! Not all bad news here. And as mentioned elsewhere, if they can solve the cooling problem, that has massive, massive potential benefits on Earth and for other space applications.
Your #2 is overlooking the emissions from the rocket that gets it all up there. The rocket equation can tell us exactly how much methane or kerosene it will take to deliver it. All that combustion...
Your #2 is overlooking the emissions from the rocket that gets it all up there. The rocket equation can tell us exactly how much methane or kerosene it will take to deliver it. All that combustion happens in our atmosphere. Same when the satellites 'burn up' on re-entry.
Yes for sure, I was trying to be specific about only positives for this but definitely agree. For any space option to be truly considered environmentally friendly, launch emissions and burn ups...
Yes for sure, I was trying to be specific about only positives for this but definitely agree. For any space option to be truly considered environmentally friendly, launch emissions and burn ups should be considered in the equation.
I wasn’t going to post because this thread is a little old and my comment is low effort, but someone else bumped it so I will too 😛. I saw a tweet speculating that it is very much for the Less...
I wasn’t going to post because this thread is a little old and my comment is low effort, but someone else bumped it so I will too 😛.
I saw a tweet speculating that it is very much for the Less rules & regulations reason. And that if you are an AI believer and think we’re scaling toward AGI, and you worry that regulations will potentially come down on building data centers (to slow that AI progress), then being able to build in space gives you more options to this pursuit.
Anyways I thought it was an interesting possibility however skeptical we might be of it.
Yeah there definitely is some level of less regulation when it comes to space not that regulation and laws have stopped tech billionaires and AI enthusiasts to this point.
Yeah there definitely is some level of less regulation when it comes to space not that regulation and laws have stopped tech billionaires and AI enthusiasts to this point.
When I read it I was just like huh, if there’s a future where both AI delivers and elon/xAI win out, it would probably be that one. OpenAI, anthropic and google would all have to have their hands...
When I read it I was just like huh, if there’s a future where both AI delivers and elon/xAI win out, it would probably be that one. OpenAI, anthropic and google would all have to have their hands tied for xAI to win.
I both agree with the ridiculousness of downplaying how big a barrier the heat dissipation issue is, and hope some of that stupidly large pool of AI bubble money gets dumped into exactly that...
I both agree with the ridiculousness of downplaying how big a barrier the heat dissipation issue is, and hope some of that stupidly large pool of AI bubble money gets dumped into exactly that research.
A massive improvement in heat-dump / heat-pump radiator tech for space-based infrastructure is one of the stepping stones we need for larger (orders of magnitude larger) space stations and, I hope, real around-our-solar-system space travel and industrialization outside the gravity well.
Assuming we can get to that point before wiping ourselves out of course.
Agree completely with this. Small benefit is at least this is going to involve a larger set of interesting problems to solve than just "more data centre and GPUs pls".
Agree completely with this. Small benefit is at least this is going to involve a larger set of interesting problems to solve than just "more data centre and GPUs pls".
My first reaction when hearing tech billionaires talk about sprawling space based infrastructure was "Oh no. They really believe space is cold because of ice." Isn't this high school physics....
My first reaction when hearing tech billionaires talk about sprawling space based infrastructure was "Oh no. They really believe space is cold because of ice."
Isn't this high school physics. Matter states are relative to the atmosphere. Remove atmosphere and there is just dense blocks of solid stuff or clouds of gas stuff. Even if they somehow have a convoy of ships ferrying chunks of ice from around the solar system to vaporize against their machines, what then. Unless they plan to pour all that energy, god knows what gasses and moisture right back at earth. That may be ecologically worse than having these things on the ground?
I did some back of the envelope numbers on this when the topic came up a month or two ago. The short version is that the gap in scale from what we have now in space-based cooling systems to what...
I did some back of the envelope numbers on this when the topic came up a month or two ago. The short version is that the gap in scale from what we have now in space-based cooling systems to what we’d need is basically equivalent to going from a Prius to a supertanker. And that needs to be done in a way that’s financially competitive compared to just… not launching them into space, and using big-ass fans for cooling instead.
I've been surprised and disappointed by the reception the space-based datacenters push has received. We should be encouraging these companies to invest heavily in solar power and advanced cooling...
I've been surprised and disappointed by the reception the space-based datacenters push has received. We should be encouraging these companies to invest heavily in solar power and advanced cooling R&D. Breakthroughs there would be helpful for all kinds of other technologies. If they're going to spend untold billions, this seems like a far more prosocial place for it to go than continuing to build terrestrial datacenters using the same methods as always.
I also think it's a bit hubristic to assume that they're idiots who can't see that it's infeasible and we've got it all figured out. Obviously it's possible that Google, SpaceX et al are mistaken and will fail, but "lol they don't know it's hot in space" doesn't seem a particularly likely reason for that, you know? Speaking for myself, I would've told you, 15 years ago, that there was zero chance SpaceX could succeed. I have since been humbled, and I am wiser for it.
I don't agree with the idea that we should support billionaires with their less destructive ideas so they won't continue to act on their more destructive ones (even as they still do it.). They...
Exemplary
I don't agree with the idea that we should support billionaires with their less destructive ideas so they won't continue to act on their more destructive ones (even as they still do it.).
They don't need my support and appeasement and should be regulated into non-destructive actions regardless of whether I'm nice to them about it. I'm not cheering on the Loop because it could have worked out and surely the next one will be better.
I honestly don't want space travel in from these hyper capitalists. I grew up on Star Trek, not The Expanse. Tech developed along the way is great. That doesn't mean it's worth whatever else they do in the meantime. And it doesn't mean it'll be accessible to anyone else depending on patent laws.
However you feel about terrestrial datacenters has virtually nothing to do with whether or not companies can/should/will invest in space-based datacenters. You can regulate terrestrial datacenters...
However you feel about terrestrial datacenters has virtually nothing to do with whether or not companies can/should/will invest in space-based datacenters. You can regulate terrestrial datacenters out of existence and that's not the same thing as encouraging R&D into something else. Mostly what would happen is datacenter owners would move to countries with less regulation, more slave labor, and dirtier power.
Billionaires bad. AI bad. I get it. We can either nudge things to make the real world a better place or we can take a principled stand on not 'appeasing' billionaires and end up worse off. And appeasing here doesn't mean giving them tax breaks or something, it literally only means saying "neat, good luck."
Elon Musk doesn't give a fuck if I respond to his plans with "fuck off or "neat, good luck." Pretending that if I don't make nice I'm somehow contributing to their unethical decisions is absolute...
Elon Musk doesn't give a fuck if I respond to his plans with "fuck off or "neat, good luck."
Pretending that if I don't make nice I'm somehow contributing to their unethical decisions is absolute nonsense. I'm sure if I said nicer things about him he'd stop being a white supremacist too.
This isn't even about a principled stand, it's that this argument is based on nothing realistic. They're not going to stop doing the worst things while doing something that might be less bad and might pay off.... For them.
From your previous post You're saying that what I say will make a difference, either be nice ("neat good luck") and make it better or be discouraging/mean and "end up worse off." In your initial...
From your previous post
We can either nudge things to make the real world a better place or we can take a principled stand on not 'appeasing' billionaires and end up worse off. And appeasing here doesn't mean giving them tax breaks or something, it literally only means saying "neat, good luck."
You're saying that what I say will make a difference, either be nice ("neat good luck") and make it better or be discouraging/mean and "end up worse off."
In your initial post you said we should be encouraging them to be more pro-social. In your reply you said they'll just go use slave labor if we regulate them.
I'm saying there is absolutely no difference in what Musk does based on how we respond here. "We" don't have "nudge" power here. I'm responding to the things you said not talking past you.
ETA: and society is not doing nothing to discourage him so I don't think that's relevant
The hard working engineers at SpaceX are far from idiots. The owner of the company they work for has always been a charismatic rich idiot, that actually smart people sucked up to so he would...
The hard working engineers at SpaceX are far from idiots. The owner of the company they work for has always been a charismatic rich idiot, that actually smart people sucked up to so he would provide capital for their business ideas, and use his charisma to get more capital from other rich people. Now, he has surrounded himself with yes men for so long that he deeply believes the sycophancy.
Being a charismatic figurehead providing capital to actually smart people worked for a long time, but as Elon has gotten in deeper thinking he is the smart one, his businesses have deteriorated. Twitter and Tesla are stagnant, and SpaceX is now saddled with covering the losses at Twitter and the Elon part of the AI bubble spending. It is not hubristic to watch the pattern of behavior and conclude Elon, and other billionaires who believe they are smart only because they surround themselves with sycophants to tell them that, are idiots.
Respectfully, I think it would be the peak of hubris to look at Elon Musk and conclude that he's an idiot who's only been successful because people give him money. It's a peculiarly liberal vice...
It is not hubristic to watch the pattern of behavior and conclude Elon, and other billionaires who believe they are smart only because they surround themselves with sycophants to tell them that, are idiots.
Respectfully, I think it would be the peak of hubris to look at Elon Musk and conclude that he's an idiot who's only been successful because people give him money.
It's a peculiarly liberal vice to conclude that anyone who sucks must be a moron - look at the dialogue surrounding US president George W. Bush, for instance. But you can suck and even be crazy - clearly something's not right with Musk - without being an idiot. He has built multiple successful businesses in historically difficult sectors. The fact that several are now stagnant does not change that fundamental accomplishment.
I don't see how one can regard the conclusion that the man who built SpaceX, whether he be an engineer or no, must actually be a fool who has a worse grasp of whether a space enterprise is technologically feasible than the average internet forum user as anything other than the peak of hubris.
Even if it were accurate that Elon's only talent is steering capital to other, smarter people, that would still be a huge deal.
I think people get really hung up on "Elon is not an engineer". I've read interviews with engineers who have worked with him and it seemed very clear he has a very strong grasp of engineering...
I think people get really hung up on "Elon is not an engineer". I've read interviews with engineers who have worked with him and it seemed very clear he has a very strong grasp of engineering concepts, not the details of the execution.
Basically enough knowledge to pick engineering goals and pick the right people to do the advanced maths, metallurgy, etc to make it happen. And enough knowledge to cut losses if a goal proves beyond reach.
I think there is also an assumption that he is an arsehole 24/7, which would be a challenge. A friend of a friend worked for one of his companies (many years ago) at a level where they would occasionally get 3am phone calls from him (timezones are hard, apparently) and they got on with him well enough for work purposes at least.
If you take a look at Elon from a sober perspective, you can see that he really is a very smart guy. He picks things up quickly, he has good ideas, and he's much more technically inclined than the...
If you take a look at Elon from a sober perspective, you can see that he really is a very smart guy. He picks things up quickly, he has good ideas, and he's much more technically inclined than the vast majority of business executives.
His problem is that he's drank his own kool-aid. Being very smart doesn't mean you're good at everything. Even the worlds foremost expert on self driving cars isn't also good at rocketry, or social media moderation, or business strategy, or fucking path of exile 2.
The difference between Elon and most actual experts is that most people who are very very good at something will admit they're not good at everything.
Elon seems to have this opinion of himself that he's just straight up better than every other human that's ever existed. He views himself as a guy that can go into a field and immediately grasp all of the problems with that field and solve them where experts who have devoted their lives to those problems have failed.
He thinks of himself as a comic book genius, like Tony Stark or Mister Fantastic.
People like that don't exist though. They never have. If you devote your life to one field and you're naturally smart, you can become an expert on the cutting edge of that field. You don't get to just do that with everything you touch though.
He thinks he's the exception to that rule, he's surrounded himself with people who affirm that belief, and he has legions of crypto bro/hyper capitalist/right wing extremist fans online that buy into that myth.
I always am a little annoyed when people cast him as an idiot though, because he's pretty clearly not a stupid guy based on objective standards, only compared to the image of himself he's created.
^ for what it's worth, my take is that Elon Musk is spectacularly good at appearing to know what he's talking about (especially to investors, or technical-but-unrelated folks), and quickly...
^ for what it's worth, my take is that Elon Musk is spectacularly good at appearing to know what he's talking about (especially to investors, or technical-but-unrelated folks), and quickly dismissing his many failures.
If you hang around competent people for long enough, you can roughly figure out what to say in order to sound proficient. Pepper in some jargon, remember a few choice arguments, and presto -- you have the persona of a genius in your back pocket, ready to be deployed on demand. If you've ever worked in a large corporation -- or heard a tech CEO open their mouths -- you'd notice how shallow their knowledge is, but how truly extraordinary they are at selling those few inches as fathomless depths. It's a make or break skill when your job is to gain an investor's confidence.
My impression is that Elon surrounded himself with brilliant minds that've spent lifetimes mastering their crafts, and then laundered their collective takes in order to seem competent to a wide audience. Absent some evidence otherwise, the only thing he is uniquely capable of is assembling a cult of personality, then turning it into investment capital. So I guess I wouldn't call him an idiot per se, but my alternative would be to use a series of adjectives that'd see me banned quite rapidly 😅
I don't think anyone is suggesting that the people who work at these companies are idiots. I think it's a lot more likely that they think that many of their investors are idiots, and that far...
I don't think anyone is suggesting that the people who work at these companies are idiots.
I think it's a lot more likely that they think that many of their investors are idiots, and that far fetched, exciting, sci fi ideas are a good way to get those investors to fork over more money.
The engineers working at these companies are well aware of the physical challenges of doing this. The business strategy people are well aware that it makes no economic sense.
Their executives are just scratching for a new angle to make headlines, and maybe some engineers will be tasked to bash their heads against the problem, and they may even launch something, but they're not going to make it economically viable without absolutely earth shattering breakthroughs in propulsion physics. They know that too.
But the main companies making headlines - SpaceX, Google - aren't taking money from investors. They're funding this themselves. Voyager (the company in the original link) already went public too -...
think it's a lot more likely that they think that many of their investors are idiots, and that far fetched, exciting, sci fi ideas are a good way to get those investors to fork over more money.
But the main companies making headlines - SpaceX, Google - aren't taking money from investors. They're funding this themselves. Voyager (the company in the original link) already went public too - and frankly I don't think the CEO would be on the record as saying there are still major unresolved challenges if he was just trying to trick idiot investors into forking over more money.
The engineers working at these companies are well aware of the physical challenges of doing this. The business strategy people are well aware that it makes no economic sense. Their executives are just scratching for a new angle to make headlines, and maybe some engineers will be tasked to bash their heads against the problem, and they may even launch something, but they're not going to make it economically viable without absolutely earth shattering breakthroughs in propulsion physics. They know that too.
I suppose this is possible. There are multiple reputable, established companies going for this, though. By Occam's razor, if nothing else, it should be at least as likely that Google et al. believe they have a shot of pulling it off than that it's all one big grift... A grift that we, random people on the Internet, can figure out trivially but which will somehow hoodwink investors.
As someone who's spent years working in various tech start-up circles and currently works in "new space", let me tell you: you have far too much faith in investor rigor. The whole VC-game has...
As someone who's spent years working in various tech start-up circles and currently works in "new space", let me tell you: you have far too much faith in investor rigor. The whole VC-game has absolutely nothing to do with real results or good engineering. It has everything to do with selling hype-cycles, building interest quickly, and exiting before the bubble pops.
A normal VC-round, even in the Space sector, aims to exit with at least 10x return on their investment within 5 years. @papasquat is mostly right, except I'd rephrase it that they don't think their investors are idiots. They think their investors are ethic-less sharks who are hunting for the next big story that they can push into the broader media and government to generate more hype. That's what this is: a lot of soul-less but smart hype-master CEOs seeing the writing on the wall for covering the planet in ever more data centres (see: increasing protests, balking at costs, trillions in promised efficiency gains not arising), and are looking for a way to pivot so they can keep the government and institutional investor gravy-train rolling without admitting that they might have a fundamental problem.
And yes, a side-benefit of this is that a good amount of the work being done by honest engineers and scientists will hopefully be released into the broader ecosystem and that'd be laudable. But the increasing disconnect from reality in these pitches drives me insane.
I do not assume they're idiots. I assume they're manipulating people who don't understand physics to get government subsidies. We do have it all figured out. It's basic physics and economics. It's...
I also think it's a bit hubristic to assume that they're idiots who can't see that it's infeasible and we've got it all figured out.
I do not assume they're idiots. I assume they're manipulating people who don't understand physics to get government subsidies.
We do have it all figured out. It's basic physics and economics. It's the same reason the Hyperloop and the Mars Mission WERE scams, and many experts in the field rightly called as such. Space X doesn't violate physics and economics out of the gate on a fundamental level, and that was discussed heavily when its project was proposed as well.
Can’t the heat be used for thrust, power generation, manufacturing, or life support? The problem seems to be due to limited infrastructure that demands heat in space. Heat is energy. It’s easy to...
Can’t the heat be used for thrust, power generation, manufacturing, or life support? The problem seems to be due to limited infrastructure that demands heat in space. Heat is energy. It’s easy to store heat in space.
You've received other replies about this, but the concept you're looking for is entropy. Heat is energy, but it's the most useless kind of energy because it's also high entropy; to get useful work...
Can’t the heat be used for thrust, power generation, manufacturing, or life support?
You've received other replies about this, but the concept you're looking for is entropy. Heat is energy, but it's the most useless kind of energy because it's also high entropy; to get useful work out of it we need to be able to transfer it from a hot region (high entropy) to a cold region (low entropy).
Unfortunately, that transfer is the very thing that's difficult in space. Since there's no direct contact with the environment to cool via conduction or convection, heat has to be radiated outwards – essentially by ensuring that the device's infrared glow points away from the sun. That's not very efficient, particularly when the solar panels have to be pointed at the sun to provide power.
Does it have to be radiated outward, or is that what we currently do?* It seems like if you can get it into contact with something colder than the heat, it could be used for a beneficial process....
Unfortunately, that transfer is the very thing that's difficult in space. Since there's no direct contact with the environment to cool via conduction or convection, heat has to be radiated outwards – essentially by ensuring that the device's infrared glow points away from the sun. That's not very efficient, particularly when the solar panels have to be pointed at the sun to provide power.
Does it have to be radiated outward, or is that what we currently do?* It seems like if you can get it into contact with something colder than the heat, it could be used for a beneficial process. If we bring ore from asteroids into a facility - couldn't we use heat to process that material into something more useful internal to the facility itself? Couldn't we channel the heat into a sink and then transfer that to something else - maybe the moon? Transfer the heat into the moon?
I'm not an expert. Just poking at it. We have plenty of orbital bodies that don't seem to have issues managing heat.
*I understand that ultimately it all results in radiating outward, no matter how many steps in between.
Yes. The temperature scale for electronics is relatively narrow in the thermodynamic sense of things: the GPUs will want to operate somewhere between 0C and 100C. If you want to use the heat for...
Does it have to be radiated outward, or is that what we currently do?*
Yes.
The temperature scale for electronics is relatively narrow in the thermodynamic sense of things: the GPUs will want to operate somewhere between 0C and 100C. If you want to use the heat for other directly useful things like "ore processing", then you'd need to develop spaceborne heat pumps (probably based on gas compression?) that can handle that kind of temperature difference at high efficiency. As far as I know this is essentially unobtanium, especially since at ore-melting temperatures you'd have the problem of the hot side of your heat pump wanting to melt itself.
If we did have that kind of heat pump technology, then we'd also just use it for radiators! The efficiency of thermal radiation scales with T^4, so doubling the absolute temperature of the material (from 0C ≈ 273K to 300C ≈ 573K) would result in a 16-fold change to the outgoing radiation.
Note that we also have a perpetual motion problem if we're not careful. If we go the other way, sinking heat from a 1000C source into a 100C target we could extract energy with (1 - 373K/1273K = 70%) efficiency because of the Carnot cycle, so our peak theoretical performance at moving heat is 1/(1-70%) = 140%. Applying 1W of energy to this system would move 1.4W of heat energy, so to sink X megawatts of GPU-related power we'd need an additional 2.5X megawatts to power the heat pumps (and move their own waste heat).
Mostover, if we did melt ore in space, what then? We'd still need to radiate that heat from the molten/refined ore in order to get the useful product (e.g. cast metal) out the other end, so we've only complicated the radiation problem.
Couldn't we channel the heat into a sink and then transfer that to something else - maybe the moon? Transfer the heat into the moon?
That's basically what we do on Earth with air cooling: we channel the heat into the air (or water, with open cycle cooling) and then transfer it into the rest of the planet. This necessarily involves a steady stream of material flowing in and out, such that doing this in space is not really viable.
Let's put numbers to this. Suppose that one of our spaceborne data centres uses 1GW of electrical energy (random Google results give me values up to 5GW, so I think I'm in the ballpark here). Let's ignore the heat pump problems from above and pretend we already have it at 1000C and can dump this into incoming iron with no losses (e.g. via some kind of perfect counterflow system).
The specific heat of iron is about 0.45J/g/K (taking 0.45J to heat 1g of iron by 1K/°C). A Joule is a watt-second, so that's 0.45 Ws/g/K, or 0.45 kW s / kg / K, or 0.45MW s / kg / 1000 K, which is the temperature difference we have to work with. We've already said that the space data centre has to dump 1GW of heat energy, so we can find the mass flow rate with some basic algebra:
1 GW = X (kg / s) * 0.45 MW s / kg → X = 2200 kg/second
That is, the space data centre would need to have 2 tons of iron flowing through it per second to use the incoming material as a heatsink. That's not a satellite, it's an industrial conveyor belt with extra steps. We in fact do this kind of thing on Earth, but here we call it a steel smelter.
I’m not talking about the data center being the industrial facility to use the heat. I’m talking about a modular system of heat sinks that can be shuttled away from the data center like a “heat...
I’m not talking about the data center being the industrial facility to use the heat. I’m talking about a modular system of heat sinks that can be shuttled away from the data center like a “heat battery pack” and replaced with a cooled heat sink. Move the “energy” somewhere it can be turned into value instead of waste.
If not value, dispose of it more efficiently. IE a facility on the moon or with a large amount of radiators.
Smelting iron is 700-1500 C. It is my understanding that ceramics and other materials can enclose temperatures like that consistently. Inside the heat sinks you’d need a material to actually store the heat - molten salt? Maybe Silicon? I don’t know. We have a lot of research earth side on storing heat for concentrated solar applications. Example: https://1414degrees.com.au/sibox/
You could, but the reason why heat is hard to deal with in space is because it's so empty. You could use it to smelt ore... Kinda, but that would require concentrating the heat, which requires...
You could, but the reason why heat is hard to deal with in space is because it's so empty. You could use it to smelt ore... Kinda, but that would require concentrating the heat, which requires even more energy, which generates more heat.
You could transfer it to the moon, but that would require the data center to be on the moon, not in space, and the moon is way harder to get to than orbit.
You could also just build heat pumps that cool down the areas you want to cool down, and heat up the radiators to super high temperatures, which makes them better at radiating heat, or use that heat to boil water to be used as steam for thrust or whatever.
The real issue is that there's no technical reason why you couldn't put a data center in space. We have the technology to do that right now if we really really wanted to.
There's no economic reason to do it though. Every aspect of it would be more expensive than putting them on earth, and not a little more expensive, like... tens of thousands of times more expensive, not to mention risky and dangerous.
The technical problems are implementation details that could be solved with enough resources. I don't see how the economic problems are solvable without some sci fi space launch technology.
If you have modular heat sinks then you don't need the data center on the moon, you could shift heat to the moon. Would a data center in space allow you to have less compute on the vehicles...
You could transfer it to the moon, but that would require the data center to be on the moon, not in space, and the moon is way harder to get to than orbit.
If you have modular heat sinks then you don't need the data center on the moon, you could shift heat to the moon.
There's no economic reason to do it though. Every aspect of it would be more expensive than putting them on earth, and not a little more expensive, like... tens of thousands of times more expensive, not to mention risky and dangerous.
Would a data center in space allow you to have less compute on the vehicles themselves by offloading the work?
That would require regularly going up and down a gravity well, which would make the whole thing require an incredible amount of energy. Space vehicles don't have a ton of compute on them; at least...
If you have modular heat sinks then you don't need the data center on the moon, you could shift heat to the moon.
That would require regularly going up and down a gravity well, which would make the whole thing require an incredible amount of energy.
Would a data center in space allow you to have less compute on the vehicles themselves by offloading the work?
Space vehicles don't have a ton of compute on them; at least not compared to AI workloads. There's no need for them to offload any of it. The types of orbital calculations done by the Apollo guidance computer can be done thousands of times faster by a mid range smartphone. If there ever were a need to offload any of it though, it would be a heck of a lot easier to just have that compute on earth.
I agree with most of your points, but I enjoy the discussion so I'm going to continue. Space elevator on the moon to solve this problem? I believe a lot of research has actually been done on that...
I agree with most of your points, but I enjoy the discussion so I'm going to continue.
That would require regularly going up and down a gravity well, which would make the whole thing require an incredible amount of energy.
Space elevator on the moon to solve this problem? I believe a lot of research has actually been done on that topic.
If there ever were a need to offload any of it though, it would be a heck of a lot easier to just have that compute on earth.
Isn't there an issue with transmission cutting out between Earth and vehicles during re-entry and/or ascent? I believe SpaceX has had to swap over to satellite based feeds for signal during some of their launches. I guess if you need real time compute sensitive to that level of signal interruption you wouldn't want it happening remotely anyways.
Any thoughts on compute demands increasing in space in the near future?
During re-entry, vehicles generate a plasma sheath as the friction between the atmosphere and the hull get so hot that they ionize that matter. One of the interesting properties of plasma is that...
During re-entry, vehicles generate a plasma sheath as the friction between the atmosphere and the hull get so hot that they ionize that matter. One of the interesting properties of plasma is that it's electrically conductive, and radio waves are absorbed by electricly conductive materials, so the craft generates it's own temporary faraday cage that makes radio communication extremely difficult. A data center being in space wouldn't really help with that. The craft would still need to use a radio link to communicate.
Space elevator on the moon to solve this problem? I believe a lot of research has actually been done on that topic.
Also ridiculously expensive. Remember, these aren't technical problems. They're economic problems. Until you can make launching, powering, cooling, maintaining, and communicating with a massive data center in space cheaper than buying a few dozen of acres of barren land, slapping down a cheap metal building with a grid connection and AC units and filling it with servers, the idea doesn't make any sense to do.
That's going to be extremely difficult to do unless either A. Land gets so expensive that maybe .0001% of the population is able to own a single family home or B. Space travel gets so cheap that the average family can decide to take a trip to LEO on a whim on any given weekend.
I don't see either of those things happening within 300 years, let alone 10.
I agree that 10 years is unlikely but I don’t know if 300 is accurate. If I look specifically at US naval vessels over the last 100 years they got 2x longer, 3x heavier, went from steam turbines...
I agree that 10 years is unlikely but I don’t know if 300 is accurate. If I look specifically at US naval vessels over the last 100 years they got 2x longer, 3x heavier, went from steam turbines and oil boilers to nuclear reactors.
Airplanes barely existed 100 years ago, they are common now. Likewise computers (as we commonly think of them) have existed for less than a hundred years.
I’m more optimistic that we will make significant progress in space during my lifetime.
The difference is that the reason naval vessels or computers or whatever other technology got better and faster was because of improved engineering. Our understanding of physics didn't actually...
The difference is that the reason naval vessels or computers or whatever other technology got better and faster was because of improved engineering. Our understanding of physics didn't actually need to change to make those things happen. It expanded certainly, but it didn't require a fundementaly new understanding of how the universe works. Nuclear reactors are a notable exception there, but the nuclear bomb and nuclear power were only possible because of 50+ years of radioactivity research and new fundemental theories that took decades to develop.
A similar, but even larger series of breakthroughs would need to happen to make space travel that cheap. It's possible, sure, but it's also entirely plausible that it just straight up isn't possible, and that launching something into orbit will always take a ridiculous amount of energy, and thus launching things into space to harness the power of sunlight in space literally never makes sense to do.
The fact is that we're not close to any sort of antigravity drive that could change that equation. There's nothing even on the horizon. There's not even the theoretical underpinnings that would make something like that possible.
The heat we're talking about is the waste energy that's left over after the solar panels have captured incoming solar energy, converted it to electricity as efficiently as they're able to, and...
The heat we're talking about is the waste energy that's left over after the solar panels have captured incoming solar energy, converted it to electricity as efficiently as they're able to, and used that electricity for the computers to do work.
Each step of the process has inefficiencies, where energy is wasted heating up the equipment rather than being harnessed to do something productive. On the ground, it's relatively easy to get rid of that waste heat by just blowing out the hot air and pulling in cool air, but in a vacuum you're basically stuck with an almost-closed system that keeps getting hotter and hotter due to those inefficiencies the longer it runs.
You can get rid of heat in a vacuum, but you have to radiate it, which is a whole lot more difficult than using conduction or convection - doing that at datacenter scale while keeping the systems lightweight enough that the launch costs don't make the whole thing utterly prohibitive is the unsolved problem here.
I was talking about waste heat as well. I'm not an expert, just interested in the topic. Heat is energy. If heat is plentiful and cheap in space, it seems like an economic opportunity to find a...
I was talking about waste heat as well. I'm not an expert, just interested in the topic. Heat is energy. If heat is plentiful and cheap in space, it seems like an economic opportunity to find a way to turn that energy into something valuable. This happens a lot with manufacturing processes where we turn something that used to be wasted into something valuable.
Your instinct is a good one! The tricky part about doing it in space is that you're not just dealing with the difficulties of a vacuum environment where heat doesn't really go anywhere, you're...
Your instinct is a good one! The tricky part about doing it in space is that you're not just dealing with the difficulties of a vacuum environment where heat doesn't really go anywhere, you're balancing that with the realities of a $1,500/kg launch cost too.
You're always going to need some level of radiative cooling system (can't beat thermodynamics in the end, however good your reclamation systems get, so there will always be some waste heat to dump), so the fixed costs there are already paid - in terms of actual system cost and in terms of having to launch it. If you add, say, thermoelectric heat reclamation then it might allow you to make the surface area of the cooling system smaller, but you've now got the fixed cost and launch weight of an additional new system to account for. It only makes sense if you can get the weight per watt reclaimed lower than the additional weight per watt of making your existing cooling system bigger, basically.*
If it were a scientific mission with specific requirements, or just a technical research platform to see if we can viably add those kind of systems, I'd be all for it - I'd actually love to see the kind of advancements we could make in balancing and optimising those competing concerns! But for an allegedly profit-making project, we're kinda already giving them the benefit of the doubt when we talk about the cooling system size, because that's actually the most viable way to minimise weight, and therefore launch costs, with current tech.
It's also why a lot of us are so skeptical: datacenters, which are a notoriously power-dense and cooling-heavy use case in general, just seem inherently unsuited to the environment. In a space elevator future with $0.10/kg cost to orbit the calculations change pretty dramatically, but if you're looking that far off you need to account for things like nuclear fusion too. Right now it's like starting a business growing coconuts in Antarctica, or building a golf course in the middle of the Sahara - we probably could do it, even with current technology, but it's going to be wildly expensive and difficult for a result that's no better than doing it the easy way.
*You absolutely need to account for energy input costs too, but the incoming solar is free and abundant, and panels are relatively cheap and lightweight to launch, so that's rarely a breaking concern for earth-orbital systems. It gets trickier if you're doing scientific missions further from the sun...
Are the launch costs as relevant for infrastructure already in space? If I launch some sort of base specifically designed to collect and process/radiate waste heat from other facilities for a fee...
Are the launch costs as relevant for infrastructure already in space? If I launch some sort of base specifically designed to collect and process/radiate waste heat from other facilities for a fee for example.
Other facilities would need to already have their own ability to radiate waste heat already to avoid frying themselves until your secondary solution came around, so you would be losing the main...
Other facilities would need to already have their own ability to radiate waste heat already to avoid frying themselves until your secondary solution came around, so you would be losing the main advantage of having some secondary system to repurpose waste heat in the first place (even if the tech is there, and I'm dubious that it is for something like that).
I'm pretty sure we can make modular heat sinks that could be swapped. I agree it's a bit of a chicken/egg problem. We've run into this on earth as well with things like EV infrastructure and other...
I'm pretty sure we can make modular heat sinks that could be swapped. I agree it's a bit of a chicken/egg problem. We've run into this on earth as well with things like EV infrastructure and other technologies that require some sort of secondary thing to exist to enable them to be more efficient / cheaper.
No. Thrust requires a propellant, not heat. No, generating energy from heat requires both a hot and a comparatively very cold region, so if anything you would need more cooling, plus an entire...
Can’t the heat be used for thrust,
No. Thrust requires a propellant, not heat.
power generation,
No, generating energy from heat requires both a hot and a comparatively very cold region, so if anything you would need more cooling, plus an entire turbine apparatus, plus "using" the heat to generate power wouldn't actually make it go anywhere, it would still be hot as hell
manufacturing,
I don't understand how heat would be relevant to manufacturing besides the energy angle I've already addressed
or life support?
No. Unmanaged heat would continue to build up until it cooked the circuitry to death, let alone any life on board. Exact opposite of life support.
It’s easy to store heat in space.
Can you elaborate on what you mean by this? I can think of some ways to store heat but none of them would actually make a spacebound facility colder.
If I'm going into XKCD mode... if we interpret 'heat' as energy, then thrust can be generated with energy alone in the absence of reaction mass (propellant) so long as you have terrifyingly large...
No. Thrust requires a propellant, not heat.
If I'm going into XKCD mode... if we interpret 'heat' as energy, then thrust can be generated with energy alone in the absence of reaction mass (propellant) so long as you have terrifyingly large quantities of energy.
Safety note. If you do have those energy levels at your disposal and want to use it to exert thrust just to say you did, please avoid thrusting directly away from earth as the secondary effects would be... impolite.
No, generating energy from heat requires both a hot and a comparatively very cold region, so if anything you would need more cooling, plus an entire turbine apparatus, plus "using" the heat to generate power wouldn't actually make it go anywhere, it would still be hot as hell
Energy generation sources that don’t use turbines and rely on heat have been used in space since 1961
Ah, I wasn't aware of that technology. The drawback of that is that you have to expend propellant constantly to cool the place down. It's not exactly the same as "using the heat as propellant when...
Multiple propulsion systems are based on heating a propellant, not just igniting it directly, e.g. solar thermal rocket,
Ah, I wasn't aware of that technology. The drawback of that is that you have to expend propellant constantly to cool the place down. It's not exactly the same as "using the heat as propellant when you need thrust".
Energy generation sources that don’t use turbines and rely on heat have been used in space since 1961
How did I forget about the power source on the voyager probes... Ok that doesn't require a turbine but still needs a hot region and cool region to work, which means you still need to get rid of the heat somehow, which will be building up as you use it because thermocouples don't cool things down.
Preamble: I am not an expert in any of this. Just interested in it. You can use anisotropic radiation of waste heat to provide a tiny amount of thrust. See: Support for the thermal origin of the...
Preamble: I am not an expert in any of this. Just interested in it.
I was also able to find EQUULEUS which appears to use the waste heat from communications equipment in part of it's propulsion systems.
No, generating energy from heat requires both a hot and a comparatively very cold region, so if anything you would need more cooling, plus an entire turbine apparatus, plus "using" the heat to generate power wouldn't actually make it go anywhere, it would still be hot as hell
Doing some basic research on this I seemed to hit three topics:
Radioisotope thermoelectric generator (RTG)
Stirling Engines
Brayton Cycles
I'm not sure if these are explicitly using waste heat, but many of them consume the heat generated by some other power generation process and recover that to generate more electricity.
Though you are just moving the forms of energy around with this. IE if you charge a battery, it will eventually release most of that energy back out as heat anyways. It's just more useful to have electricity instead of heat.
I don't understand how heat would be relevant to manufacturing besides the energy angle I've already addressed
Smelting? Welding? Refining? This is the topic I have done the least research into. We use heat to do these things on earth. If we mine other planets and/or asteroids it may make more sense for a lot of the refining and manufacturing of those raw materials to happen in space before moving back into the atmosphere. Especially if waste heat is a cheap and plentiful source of energy.
Can you elaborate on what you mean by this? I can think of some ways to store heat but none of them would actually make a spacebound facility colder.
Pump the heat into a sink or storage device (molten salt? etc?). Another vessel swaps the sink with a cold sink and transports the heat to somewhere it can be used for something beneficial.
I think the problem with this idea is that waste heat is usually not hot enough for industrial applications. And because the system will always try to move towards equilibrium, if you have enough...
Smelting? Welding? Refining? This is the topic I have done the least research into. We use heat to do these things on earth. If we mine other planets and/or asteroids it may make more sense for a lot of the refining and manufacturing of those raw materials to happen in space before moving back into the atmosphere. Especially if waste heat is a cheap and plentiful source of energy.
I think the problem with this idea is that waste heat is usually not hot enough for industrial applications. And because the system will always try to move towards equilibrium, if you have enough heat for industrial applications, eventually your whole satellite/space base will be that temperature, and any electronics will have failed long before you get there. The system is always trying to average the temperature across the whole satellite.
There’s an interesting XKCD What-If? about a hairdryer sealed in a box, and what to expect, which touches on equilibrium of energy input and external temperature assuming it’s in an atmosphere which is convecting away the heat. But it also looks at the simplified relationship between energy input and heat, because all the energy you pump into the system has to go somewhere. So if it’s not radiated away, then it’s sticking around.
With things like heat pumps, you can move it around a bit, so you can make one room hotter while you make the other room colder. Theoretically (but I don’t think we have the current technology to achieve) you can get to the kinds of extremes where you have some parts of your satellite cool enough to run a data centre, and the other end hot enough to run a foundry. But you have to constantly be adding energy into the system just to run these heat pumps, because thermodynamics is always trying to bring the whole satellite towards the average temperature. And heat pumps that we have in houses will dramatically lose efficiency when trying to operate outside the narrow ideal band of temperatures they’re designed for, so you would definitely need to be using different technology to enable foundry-to-cool room levels of heat pump effectiveness.
I guess in my head I don’t envision a foundry attached to a data center. I envision heat sinks being moved from a data center to a foundry. Two completely different facilities.
I guess in my head I don’t envision a foundry attached to a data center. I envision heat sinks being moved from a data center to a foundry. Two completely different facilities.
Heat sinks like that would fall under the category of requiring us to make something else very cold for the purpose of heat transfer in an environment that is already about as extremely...
Heat sinks like that would fall under the category of requiring us to make something else very cold for the purpose of heat transfer in an environment that is already about as extremely inhospitable for heat transfer as it can get.
In that case I think you might have made the situation even more complicated, because the combo foundry + data centre has a place to put the hot and a place to put the cold, so the problem is only...
In that case I think you might have made the situation even more complicated, because the combo foundry + data centre has a place to put the hot and a place to put the cold, so the problem is only how to effectively move heat around where you need it as you need it.
But having them isolated, with stored heat bricks to be shipped off to the other facility, means you’re already working with the challenge of moving heat around, but with the added complexity of trying to capture industrial levels of heat in a relatively small object that somehow doesn’t melt the pods/shuttle that ferries back and forth between the facilities, and also now your heat needs to be quantised into “how much useful heat (or cold) can this brick hold without being damaged” rather than just moving heat around continuously.
The driving force behind space based data centres is power consumption, right? In that there's a tension between (1) the growth curve that the tech sector desires for their LLM training data...
The driving force behind space based data centres is power consumption, right? In that there's a tension between (1) the growth curve that the tech sector desires for their LLM training data centres, and (2) various countries' desires (and abilities) to build out a power grid to support it.
So, uh, dumb question, but why aren't any of these tech CEOs talking about space-based solar power? I mean, if we're going to shoot for the stars anyways, why not do so with a giant space-borne microwave laser?
I'll just paste my comment from elsewhere in this thread - basically the issue is that's a harder sell. Satellites beaming laser-power to the surface sounds more intuitively challenging than...
I'll just paste my comment from elsewhere in this thread - basically the issue is that's a harder sell. Satellites beaming laser-power to the surface sounds more intuitively challenging than cooling in very cold (but awful heat transmissivity) space.
There are a lot of well-funded space based solar power/power beaming initiatives going on as well, though.
(I believe your markdown link accidentally inverted the ], btw) Fair enough! I also agree that proposing space-based data centres is absurd at present, but I was mostly curious as to why they'd...
(I believe your markdown link accidentally inverted the ], btw)
Fair enough! I also agree that proposing space-based data centres is absurd at present, but I was mostly curious as to why they'd draw the line at lasing earth from space.
In practical terms, it doesn't make sense to do. Granted, it makes more sense than space based data centers, but it's still pretty pointless. The link says that the sun is 144% as intense in space...
In practical terms, it doesn't make sense to do. Granted, it makes more sense than space based data centers, but it's still pretty pointless.
The link says that the sun is 144% as intense in space as on the ground, which is great. The problem is that another great thing about solar panels is that they're very, very cheap, and getting cheaper every year. It's just insane how cheap they are. They're easy to build and don't require a ton of energy in their manufacture. What isn't cheap is lifting them into space.
So if they generate 1.44 times as much power in space, we could just build twice as many, or three times as many to cover for the fact that it's night time on earth sometimes. It would still be way, way cheaper to do that than to launch them into space.
If ground area is an issue (it isn't, theres plenty of desolate land on earth) you can put them in giant desert regions, or the oceans, both of which are way more practical than putting them into space.
I think everyone knows that, and the real reason why CEOs are talking about AI data centers is because AI is the thing that gets capital excited right now though.
Makes sense! As noted in another comment, I was moreso curious why the line had been drawn at building data centres in orbit, since both proposals seem equivalently credulous (naively). Oh 100%...
I think everyone knows that, and the real reason why CEOs are talking about AI data centers is because AI is the thing that gets capital excited right now though.
Makes sense! As noted in another comment, I was moreso curious why the line had been drawn at building data centres in orbit, since both proposals seem equivalently credulous (naively).
So if they generate 1.44 times as much power in space, we could just build twice as many, or three times as many to cover for the fact that it's night time on earth sometimes. It would still be way, way cheaper to do that than to launch them into space.
If ground area is an issue (it isn't, theres plenty of desolate land on earth) you can put them in giant desert regions, or the oceans, both of which are way more practical than putting them into space.
Oh 100% agreed. I figure that there are still some environmental concerns to tackle in those -- afaik deserts aren't completely devoid of life, and even the the open expanse of the ocean harbours photosynthetic life which could be affected by blanketing the oceans with solar panels. But those concerns aren't insurmountable, and are absolutely more feasible than launching said materials into an incredibly hostile environment.
Continuing my rant about how Space-based data centres need to address the massive hurdle of cooling and almost no one is saying anything about it, here's the CEO of Voyager Technologies pointing it out too. They were one of the first to say they were going to work on Space-based Data Centres, and are also building a lot of neat space hardware, largely focused on space stations.
As an ignorant person, I assumed the main benefit of space data centres would be easy cooling. Learning that that is basically the opposite makes me wonder what benefit there is to space data centres? Less rules? Less chance of citizens protesting? If Elon is involved I assume it's mostly lies and grift.
It makes no logical sense. Full stop. With the hyper loop you could say “well it it was possible it would be faster even though we don’t need that” as SOME level of reason.
This just sounds like something that might make sense to the layman and 100% does not.
It is easier to get power on earth. Easier to cool. Easier to maintain. Easier to build.
You would need to be out of space on the surface AND in the ocean for this to make sense or be at the point where orbital lift was trivial (skyhook, space elevator, whatever) for this to even start making sense
Okay, that's kind of what I assumed. I should have thought more logically about it. My brain was just like, space is cold, data centres need cooling, so obviously that's why they are suddenly obsessed with it.
Space is cold in the sense that the average temperature of the few molecules floating around are close to absolute 0.
The main mechanism for one cold thing to cool another thing (water cooling a data center or wind cooling your skin), heat actually needs to be transferred between molecules in a process called convection.
There are so few molecules in space that the heat doesn't have anywhere to go. If you were instantly teleported to space without a suit it wouldn't actually feel that cold, you would just die from lack of oxygen.
Things in space still lose heat to radiation, but it is minimal when compared to convection on earth. Your body would eventually freeze due to heat loss from radiation, but it would take a while.
More specifically the vacuum would cause every liquid you have to boil and you would dessicate before you eventually freeze. A form of mummification.
Regardless, you're in for a very bad time.
Actually, it's the opposite! Most particles in space are moving very quickly, so their average "temperature" (kinetic energy) is quite high. There's just not many of them.
Oh thank you for the correction! I thought that temperature was always an average of all particles and that space temperature was very low.
So is mostly empty space considered to be a high temperature since the few particles have high kinetic energy, or is temperature measured in a way that the temperature would still be close to absolute zero?
The temperature of the matter in space is high, but since there's not much matter the thermal energy is low.
You might be thinking about the heat of the radiation in space - there is ambient light coming from the cosmic microwave background, but it's only a few °K. As such radiators always trend toward that temperature (except in the presence of other sources of hotter radiation like planets and starlight).
For someone like Elon, advocating for space data centers is a way to justify the SpaceX purchase of xAI (which includes Twitter), since that company is burning money.
SpaceX is primarily a government contractor. I suspect this is all simply a ploy to get U.S. taxpayers to subsidize the development of large language models.
The idea is that if you put solar panels in space, they produce consistent power all the time, no dealing with night time, cloudiness etc. Plus the atmosphere absorbs a lot of the sun's energy so if the panels are in space you don't have to deal with that.
That said it costs like $10,000/kg to put things in space (maybe less these days) so the world in which it makes economic sense to build space data centers looks a lot different than the world we live in now.
I think the real idea is they can't regulate you in space.
I feel like we'd be seeing a lot more offshore datacenter barges in international waters if that were a significant driver, though? Those would have a whole stack of their own logistical problems, but I'd bet you can build them with 2x-3x redundancy (physically and legally) and still come out hundreds of millions cheaper than launching them into space. Let's say one ship flagged in Liberia and moored off the coast of the Philippines, one flagged in Vanuatu and moored near Mexico, one flaged in the Marshall Islands and moored off India - you're gonna cause a decade of legal gridlock trying to shut down even one of those, let alone all three.
Until the US Navy and Coast Guard decide to chase down your ship and board it due to "irregularities," then confiscate the ship!
I agree that lack of regulation isn't the goal here. But even if it was, like someone else pointed out, might > law. Quite often.
Oh yeah for sure, and not just state level actors - good old fashioned speedboats-and-AK47s piracy would be a legitimate concern too. One of the few genuine advantages of space vs ocean is limiting the number of organisations/governments who’d be physically capable of interfering, actually!
I will say that I think you’d need to really piss off the US for them to go full legally-questionable naval confiscation, though, and I get the impression that the game for a lot of billionaires’ dealings is plausible deniability and a legal fig leaf rather than a direct middle finger to the government that’d prompt a targeted response. You’re giving yourself the ability to bury the plaintiff in paperwork and convince the US courts that it’s not their problem or jurisdiction when someone sues you, more than poking the bear and going against the US itself, I think.
But yeah, edge cases and practicalities aside I’m not suggesting that extrajudicial seafaring datacenters are actually a good idea - I just think we’d see them as a stepping stone to orbital ones if the actual demand were there.
They can, though. It's trivial for US or China to shoot your datacenter down with a missile.
I wouldn't call that regulation.
Regulation doesn't depend on the physical location of a server, it depends on the physical location of a headquarters or citizenship or whatever. As long as spaceX can be reached by a court order, it can be regulated.
Literally the first sentence of SpaceX's statement is
It means laws can be enforced, if there were laws.
I would think this is not really a move they would do unless it's very serious due to space debris.
That's not really how regulation works. If the company has a US presence or does business with US citizens, it can be regulated.
Those data centers in space still need to actually be communicated with and run by a company. It's not like you could run a pirate data center and the company running it would be untouchable by US law just because their servers aren't physically reachable.
Yeah I understand that argument in theory, but without knowing the math, I assumed the cost wouldn't be worth it unless you also had easy cooling. Thanks!
To take off my rage hat and put on my optimistic futurist hat, here are some potential benefits of space-based data centres (assuming you somehow solved cooling):
So with an edge-computing solution - call it a data centre orbiting the Moon - you could enable a lot more autonomous analysis and navigation solutions on the Moon and further. You could also parse space-telescope data coming in before it hits the choke points like the Deep Space Network which is already at capacity. Anything we can do to try and make it easier to send deep-space data back home will be a huge improvemnet.
So! Not all bad news here. And as mentioned elsewhere, if they can solve the cooling problem, that has massive, massive potential benefits on Earth and for other space applications.
Your #2 is overlooking the emissions from the rocket that gets it all up there. The rocket equation can tell us exactly how much methane or kerosene it will take to deliver it. All that combustion happens in our atmosphere. Same when the satellites 'burn up' on re-entry.
Yes for sure, I was trying to be specific about only positives for this but definitely agree. For any space option to be truly considered environmentally friendly, launch emissions and burn ups should be considered in the equation.
I wasn’t going to post because this thread is a little old and my comment is low effort, but someone else bumped it so I will too 😛.
I saw a tweet speculating that it is very much for the Less rules & regulations reason. And that if you are an AI believer and think we’re scaling toward AGI, and you worry that regulations will potentially come down on building data centers (to slow that AI progress), then being able to build in space gives you more options to this pursuit.
Anyways I thought it was an interesting possibility however skeptical we might be of it.
Yeah there definitely is some level of less regulation when it comes to space not that regulation and laws have stopped tech billionaires and AI enthusiasts to this point.
When I read it I was just like huh, if there’s a future where both AI delivers and elon/xAI win out, it would probably be that one. OpenAI, anthropic and google would all have to have their hands tied for xAI to win.
I both agree with the ridiculousness of downplaying how big a barrier the heat dissipation issue is, and hope some of that stupidly large pool of AI bubble money gets dumped into exactly that research.
A massive improvement in heat-dump / heat-pump radiator tech for space-based infrastructure is one of the stepping stones we need for larger (orders of magnitude larger) space stations and, I hope, real around-our-solar-system space travel and industrialization outside the gravity well.
Assuming we can get to that point before wiping ourselves out of course.
Agree completely with this. Small benefit is at least this is going to involve a larger set of interesting problems to solve than just "more data centre and GPUs pls".
My first reaction when hearing tech billionaires talk about sprawling space based infrastructure was "Oh no. They really believe space is cold because of ice."
Isn't this high school physics. Matter states are relative to the atmosphere. Remove atmosphere and there is just dense blocks of solid stuff or clouds of gas stuff. Even if they somehow have a convoy of ships ferrying chunks of ice from around the solar system to vaporize against their machines, what then. Unless they plan to pour all that energy, god knows what gasses and moisture right back at earth. That may be ecologically worse than having these things on the ground?
I did some back of the envelope numbers on this when the topic came up a month or two ago. The short version is that the gap in scale from what we have now in space-based cooling systems to what we’d need is basically equivalent to going from a Prius to a supertanker. And that needs to be done in a way that’s financially competitive compared to just… not launching them into space, and using big-ass fans for cooling instead.
Lol, I love the reference you made there to the old system of scientific patronage... But the motive is hubris instead of altruism.
I've been surprised and disappointed by the reception the space-based datacenters push has received. We should be encouraging these companies to invest heavily in solar power and advanced cooling R&D. Breakthroughs there would be helpful for all kinds of other technologies. If they're going to spend untold billions, this seems like a far more prosocial place for it to go than continuing to build terrestrial datacenters using the same methods as always.
I also think it's a bit hubristic to assume that they're idiots who can't see that it's infeasible and we've got it all figured out. Obviously it's possible that Google, SpaceX et al are mistaken and will fail, but "lol they don't know it's hot in space" doesn't seem a particularly likely reason for that, you know? Speaking for myself, I would've told you, 15 years ago, that there was zero chance SpaceX could succeed. I have since been humbled, and I am wiser for it.
I don't agree with the idea that we should support billionaires with their less destructive ideas so they won't continue to act on their more destructive ones (even as they still do it.).
They don't need my support and appeasement and should be regulated into non-destructive actions regardless of whether I'm nice to them about it. I'm not cheering on the Loop because it could have worked out and surely the next one will be better.
I honestly don't want space travel in from these hyper capitalists. I grew up on Star Trek, not The Expanse. Tech developed along the way is great. That doesn't mean it's worth whatever else they do in the meantime. And it doesn't mean it'll be accessible to anyone else depending on patent laws.
However you feel about terrestrial datacenters has virtually nothing to do with whether or not companies can/should/will invest in space-based datacenters. You can regulate terrestrial datacenters out of existence and that's not the same thing as encouraging R&D into something else. Mostly what would happen is datacenter owners would move to countries with less regulation, more slave labor, and dirtier power.
Billionaires bad. AI bad. I get it. We can either nudge things to make the real world a better place or we can take a principled stand on not 'appeasing' billionaires and end up worse off. And appeasing here doesn't mean giving them tax breaks or something, it literally only means saying "neat, good luck."
Elon Musk doesn't give a fuck if I respond to his plans with "fuck off or "neat, good luck."
Pretending that if I don't make nice I'm somehow contributing to their unethical decisions is absolute nonsense. I'm sure if I said nicer things about him he'd stop being a white supremacist too.
This isn't even about a principled stand, it's that this argument is based on nothing realistic. They're not going to stop doing the worst things while doing something that might be less bad and might pay off.... For them.
Clearly we're talking past one another.
From your previous post
You're saying that what I say will make a difference, either be nice ("neat good luck") and make it better or be discouraging/mean and "end up worse off."
In your initial post you said we should be encouraging them to be more pro-social. In your reply you said they'll just go use slave labor if we regulate them.
I'm saying there is absolutely no difference in what Musk does based on how we respond here. "We" don't have "nudge" power here. I'm responding to the things you said not talking past you.
ETA: and society is not doing nothing to discourage him so I don't think that's relevant
The hard working engineers at SpaceX are far from idiots. The owner of the company they work for has always been a charismatic rich idiot, that actually smart people sucked up to so he would provide capital for their business ideas, and use his charisma to get more capital from other rich people. Now, he has surrounded himself with yes men for so long that he deeply believes the sycophancy.
Being a charismatic figurehead providing capital to actually smart people worked for a long time, but as Elon has gotten in deeper thinking he is the smart one, his businesses have deteriorated. Twitter and Tesla are stagnant, and SpaceX is now saddled with covering the losses at Twitter and the Elon part of the AI bubble spending. It is not hubristic to watch the pattern of behavior and conclude Elon, and other billionaires who believe they are smart only because they surround themselves with sycophants to tell them that, are idiots.
Respectfully, I think it would be the peak of hubris to look at Elon Musk and conclude that he's an idiot who's only been successful because people give him money.
It's a peculiarly liberal vice to conclude that anyone who sucks must be a moron - look at the dialogue surrounding US president George W. Bush, for instance. But you can suck and even be crazy - clearly something's not right with Musk - without being an idiot. He has built multiple successful businesses in historically difficult sectors. The fact that several are now stagnant does not change that fundamental accomplishment.
I don't see how one can regard the conclusion that the man who built SpaceX, whether he be an engineer or no, must actually be a fool who has a worse grasp of whether a space enterprise is technologically feasible than the average internet forum user as anything other than the peak of hubris.
Even if it were accurate that Elon's only talent is steering capital to other, smarter people, that would still be a huge deal.
I think people get really hung up on "Elon is not an engineer". I've read interviews with engineers who have worked with him and it seemed very clear he has a very strong grasp of engineering concepts, not the details of the execution.
Basically enough knowledge to pick engineering goals and pick the right people to do the advanced maths, metallurgy, etc to make it happen. And enough knowledge to cut losses if a goal proves beyond reach.
I think there is also an assumption that he is an arsehole 24/7, which would be a challenge. A friend of a friend worked for one of his companies (many years ago) at a level where they would occasionally get 3am phone calls from him (timezones are hard, apparently) and they got on with him well enough for work purposes at least.
If you take a look at Elon from a sober perspective, you can see that he really is a very smart guy. He picks things up quickly, he has good ideas, and he's much more technically inclined than the vast majority of business executives.
His problem is that he's drank his own kool-aid. Being very smart doesn't mean you're good at everything. Even the worlds foremost expert on self driving cars isn't also good at rocketry, or social media moderation, or business strategy, or fucking path of exile 2.
The difference between Elon and most actual experts is that most people who are very very good at something will admit they're not good at everything.
Elon seems to have this opinion of himself that he's just straight up better than every other human that's ever existed. He views himself as a guy that can go into a field and immediately grasp all of the problems with that field and solve them where experts who have devoted their lives to those problems have failed.
He thinks of himself as a comic book genius, like Tony Stark or Mister Fantastic.
People like that don't exist though. They never have. If you devote your life to one field and you're naturally smart, you can become an expert on the cutting edge of that field. You don't get to just do that with everything you touch though.
He thinks he's the exception to that rule, he's surrounded himself with people who affirm that belief, and he has legions of crypto bro/hyper capitalist/right wing extremist fans online that buy into that myth.
I always am a little annoyed when people cast him as an idiot though, because he's pretty clearly not a stupid guy based on objective standards, only compared to the image of himself he's created.
^ for what it's worth, my take is that Elon Musk is spectacularly good at appearing to know what he's talking about (especially to investors, or technical-but-unrelated folks), and quickly dismissing his many failures.
If you hang around competent people for long enough, you can roughly figure out what to say in order to sound proficient. Pepper in some jargon, remember a few choice arguments, and presto -- you have the persona of a genius in your back pocket, ready to be deployed on demand. If you've ever worked in a large corporation -- or heard a tech CEO open their mouths -- you'd notice how shallow their knowledge is, but how truly extraordinary they are at selling those few inches as fathomless depths. It's a make or break skill when your job is to gain an investor's confidence.
My impression is that Elon surrounded himself with brilliant minds that've spent lifetimes mastering their crafts, and then laundered their collective takes in order to seem competent to a wide audience. Absent some evidence otherwise, the only thing he is uniquely capable of is assembling a cult of personality, then turning it into investment capital. So I guess I wouldn't call him an idiot per se, but my alternative would be to use a series of adjectives that'd see me banned quite rapidly 😅
You have just described a classic Fachidiot.
I don't think anyone is suggesting that the people who work at these companies are idiots.
I think it's a lot more likely that they think that many of their investors are idiots, and that far fetched, exciting, sci fi ideas are a good way to get those investors to fork over more money.
The engineers working at these companies are well aware of the physical challenges of doing this. The business strategy people are well aware that it makes no economic sense.
Their executives are just scratching for a new angle to make headlines, and maybe some engineers will be tasked to bash their heads against the problem, and they may even launch something, but they're not going to make it economically viable without absolutely earth shattering breakthroughs in propulsion physics. They know that too.
But the main companies making headlines - SpaceX, Google - aren't taking money from investors. They're funding this themselves. Voyager (the company in the original link) already went public too - and frankly I don't think the CEO would be on the record as saying there are still major unresolved challenges if he was just trying to trick idiot investors into forking over more money.
I suppose this is possible. There are multiple reputable, established companies going for this, though. By Occam's razor, if nothing else, it should be at least as likely that Google et al. believe they have a shot of pulling it off than that it's all one big grift... A grift that we, random people on the Internet, can figure out trivially but which will somehow hoodwink investors.
As someone who's spent years working in various tech start-up circles and currently works in "new space", let me tell you: you have far too much faith in investor rigor. The whole VC-game has absolutely nothing to do with real results or good engineering. It has everything to do with selling hype-cycles, building interest quickly, and exiting before the bubble pops.
A normal VC-round, even in the Space sector, aims to exit with at least 10x return on their investment within 5 years. @papasquat is mostly right, except I'd rephrase it that they don't think their investors are idiots. They think their investors are ethic-less sharks who are hunting for the next big story that they can push into the broader media and government to generate more hype. That's what this is: a lot of soul-less but smart hype-master CEOs seeing the writing on the wall for covering the planet in ever more data centres (see: increasing protests, balking at costs, trillions in promised efficiency gains not arising), and are looking for a way to pivot so they can keep the government and institutional investor gravy-train rolling without admitting that they might have a fundamental problem.
And yes, a side-benefit of this is that a good amount of the work being done by honest engineers and scientists will hopefully be released into the broader ecosystem and that'd be laudable. But the increasing disconnect from reality in these pitches drives me insane.
I do not assume they're idiots. I assume they're manipulating people who don't understand physics to get government subsidies.
We do have it all figured out. It's basic physics and economics. It's the same reason the Hyperloop and the Mars Mission WERE scams, and many experts in the field rightly called as such. Space X doesn't violate physics and economics out of the gate on a fundamental level, and that was discussed heavily when its project was proposed as well.
Can’t the heat be used for thrust, power generation, manufacturing, or life support? The problem seems to be due to limited infrastructure that demands heat in space. Heat is energy. It’s easy to store heat in space.
You've received other replies about this, but the concept you're looking for is entropy. Heat is energy, but it's the most useless kind of energy because it's also high entropy; to get useful work out of it we need to be able to transfer it from a hot region (high entropy) to a cold region (low entropy).
Unfortunately, that transfer is the very thing that's difficult in space. Since there's no direct contact with the environment to cool via conduction or convection, heat has to be radiated outwards – essentially by ensuring that the device's infrared glow points away from the sun. That's not very efficient, particularly when the solar panels have to be pointed at the sun to provide power.
Does it have to be radiated outward, or is that what we currently do?* It seems like if you can get it into contact with something colder than the heat, it could be used for a beneficial process. If we bring ore from asteroids into a facility - couldn't we use heat to process that material into something more useful internal to the facility itself? Couldn't we channel the heat into a sink and then transfer that to something else - maybe the moon? Transfer the heat into the moon?
I'm not an expert. Just poking at it. We have plenty of orbital bodies that don't seem to have issues managing heat.
*I understand that ultimately it all results in radiating outward, no matter how many steps in between.
Yes.
The temperature scale for electronics is relatively narrow in the thermodynamic sense of things: the GPUs will want to operate somewhere between 0C and 100C. If you want to use the heat for other directly useful things like "ore processing", then you'd need to develop spaceborne heat pumps (probably based on gas compression?) that can handle that kind of temperature difference at high efficiency. As far as I know this is essentially unobtanium, especially since at ore-melting temperatures you'd have the problem of the hot side of your heat pump wanting to melt itself.
If we did have that kind of heat pump technology, then we'd also just use it for radiators! The efficiency of thermal radiation scales with T^4, so doubling the absolute temperature of the material (from 0C ≈ 273K to 300C ≈ 573K) would result in a 16-fold change to the outgoing radiation.
Note that we also have a perpetual motion problem if we're not careful. If we go the other way, sinking heat from a 1000C source into a 100C target we could extract energy with (1 - 373K/1273K = 70%) efficiency because of the Carnot cycle, so our peak theoretical performance at moving heat is 1/(1-70%) = 140%. Applying 1W of energy to this system would move 1.4W of heat energy, so to sink X megawatts of GPU-related power we'd need an additional 2.5X megawatts to power the heat pumps (and move their own waste heat).
Mostover, if we did melt ore in space, what then? We'd still need to radiate that heat from the molten/refined ore in order to get the useful product (e.g. cast metal) out the other end, so we've only complicated the radiation problem.
That's basically what we do on Earth with air cooling: we channel the heat into the air (or water, with open cycle cooling) and then transfer it into the rest of the planet. This necessarily involves a steady stream of material flowing in and out, such that doing this in space is not really viable.
Let's put numbers to this. Suppose that one of our spaceborne data centres uses 1GW of electrical energy (random Google results give me values up to 5GW, so I think I'm in the ballpark here). Let's ignore the heat pump problems from above and pretend we already have it at 1000C and can dump this into incoming iron with no losses (e.g. via some kind of perfect counterflow system).
The specific heat of iron is about 0.45J/g/K (taking 0.45J to heat 1g of iron by 1K/°C). A Joule is a watt-second, so that's 0.45 Ws/g/K, or 0.45 kW s / kg / K, or 0.45MW s / kg / 1000 K, which is the temperature difference we have to work with. We've already said that the space data centre has to dump 1GW of heat energy, so we can find the mass flow rate with some basic algebra:
1 GW = X (kg / s) * 0.45 MW s / kg → X = 2200 kg/second
That is, the space data centre would need to have 2 tons of iron flowing through it per second to use the incoming material as a heatsink. That's not a satellite, it's an industrial conveyor belt with extra steps. We in fact do this kind of thing on Earth, but here we call it a steel smelter.
I’m not talking about the data center being the industrial facility to use the heat. I’m talking about a modular system of heat sinks that can be shuttled away from the data center like a “heat battery pack” and replaced with a cooled heat sink. Move the “energy” somewhere it can be turned into value instead of waste.
If not value, dispose of it more efficiently. IE a facility on the moon or with a large amount of radiators.
Smelting iron is 700-1500 C. It is my understanding that ceramics and other materials can enclose temperatures like that consistently. Inside the heat sinks you’d need a material to actually store the heat - molten salt? Maybe Silicon? I don’t know. We have a lot of research earth side on storing heat for concentrated solar applications. Example: https://1414degrees.com.au/sibox/
I appreciate your well thought out response.
You could, but the reason why heat is hard to deal with in space is because it's so empty. You could use it to smelt ore... Kinda, but that would require concentrating the heat, which requires even more energy, which generates more heat.
You could transfer it to the moon, but that would require the data center to be on the moon, not in space, and the moon is way harder to get to than orbit.
You could also just build heat pumps that cool down the areas you want to cool down, and heat up the radiators to super high temperatures, which makes them better at radiating heat, or use that heat to boil water to be used as steam for thrust or whatever.
The real issue is that there's no technical reason why you couldn't put a data center in space. We have the technology to do that right now if we really really wanted to.
There's no economic reason to do it though. Every aspect of it would be more expensive than putting them on earth, and not a little more expensive, like... tens of thousands of times more expensive, not to mention risky and dangerous.
The technical problems are implementation details that could be solved with enough resources. I don't see how the economic problems are solvable without some sci fi space launch technology.
If you have modular heat sinks then you don't need the data center on the moon, you could shift heat to the moon.
Would a data center in space allow you to have less compute on the vehicles themselves by offloading the work?
That would require regularly going up and down a gravity well, which would make the whole thing require an incredible amount of energy.
Space vehicles don't have a ton of compute on them; at least not compared to AI workloads. There's no need for them to offload any of it. The types of orbital calculations done by the Apollo guidance computer can be done thousands of times faster by a mid range smartphone. If there ever were a need to offload any of it though, it would be a heck of a lot easier to just have that compute on earth.
I agree with most of your points, but I enjoy the discussion so I'm going to continue.
Space elevator on the moon to solve this problem? I believe a lot of research has actually been done on that topic.
Isn't there an issue with transmission cutting out between Earth and vehicles during re-entry and/or ascent? I believe SpaceX has had to swap over to satellite based feeds for signal during some of their launches. I guess if you need real time compute sensitive to that level of signal interruption you wouldn't want it happening remotely anyways.
Any thoughts on compute demands increasing in space in the near future?
During re-entry, vehicles generate a plasma sheath as the friction between the atmosphere and the hull get so hot that they ionize that matter. One of the interesting properties of plasma is that it's electrically conductive, and radio waves are absorbed by electricly conductive materials, so the craft generates it's own temporary faraday cage that makes radio communication extremely difficult. A data center being in space wouldn't really help with that. The craft would still need to use a radio link to communicate.
Also ridiculously expensive. Remember, these aren't technical problems. They're economic problems. Until you can make launching, powering, cooling, maintaining, and communicating with a massive data center in space cheaper than buying a few dozen of acres of barren land, slapping down a cheap metal building with a grid connection and AC units and filling it with servers, the idea doesn't make any sense to do.
That's going to be extremely difficult to do unless either A. Land gets so expensive that maybe .0001% of the population is able to own a single family home or B. Space travel gets so cheap that the average family can decide to take a trip to LEO on a whim on any given weekend.
I don't see either of those things happening within 300 years, let alone 10.
I agree that 10 years is unlikely but I don’t know if 300 is accurate. If I look specifically at US naval vessels over the last 100 years they got 2x longer, 3x heavier, went from steam turbines and oil boilers to nuclear reactors.
Airplanes barely existed 100 years ago, they are common now. Likewise computers (as we commonly think of them) have existed for less than a hundred years.
I’m more optimistic that we will make significant progress in space during my lifetime.
The difference is that the reason naval vessels or computers or whatever other technology got better and faster was because of improved engineering. Our understanding of physics didn't actually need to change to make those things happen. It expanded certainly, but it didn't require a fundementaly new understanding of how the universe works. Nuclear reactors are a notable exception there, but the nuclear bomb and nuclear power were only possible because of 50+ years of radioactivity research and new fundemental theories that took decades to develop.
A similar, but even larger series of breakthroughs would need to happen to make space travel that cheap. It's possible, sure, but it's also entirely plausible that it just straight up isn't possible, and that launching something into orbit will always take a ridiculous amount of energy, and thus launching things into space to harness the power of sunlight in space literally never makes sense to do.
The fact is that we're not close to any sort of antigravity drive that could change that equation. There's nothing even on the horizon. There's not even the theoretical underpinnings that would make something like that possible.
The heat we're talking about is the waste energy that's left over after the solar panels have captured incoming solar energy, converted it to electricity as efficiently as they're able to, and used that electricity for the computers to do work.
Each step of the process has inefficiencies, where energy is wasted heating up the equipment rather than being harnessed to do something productive. On the ground, it's relatively easy to get rid of that waste heat by just blowing out the hot air and pulling in cool air, but in a vacuum you're basically stuck with an almost-closed system that keeps getting hotter and hotter due to those inefficiencies the longer it runs.
You can get rid of heat in a vacuum, but you have to radiate it, which is a whole lot more difficult than using conduction or convection - doing that at datacenter scale while keeping the systems lightweight enough that the launch costs don't make the whole thing utterly prohibitive is the unsolved problem here.
I was talking about waste heat as well. I'm not an expert, just interested in the topic. Heat is energy. If heat is plentiful and cheap in space, it seems like an economic opportunity to find a way to turn that energy into something valuable. This happens a lot with manufacturing processes where we turn something that used to be wasted into something valuable.
Your instinct is a good one! The tricky part about doing it in space is that you're not just dealing with the difficulties of a vacuum environment where heat doesn't really go anywhere, you're balancing that with the realities of a $1,500/kg launch cost too.
You're always going to need some level of radiative cooling system (can't beat thermodynamics in the end, however good your reclamation systems get, so there will always be some waste heat to dump), so the fixed costs there are already paid - in terms of actual system cost and in terms of having to launch it. If you add, say, thermoelectric heat reclamation then it might allow you to make the surface area of the cooling system smaller, but you've now got the fixed cost and launch weight of an additional new system to account for. It only makes sense if you can get the weight per watt reclaimed lower than the additional weight per watt of making your existing cooling system bigger, basically.*
If it were a scientific mission with specific requirements, or just a technical research platform to see if we can viably add those kind of systems, I'd be all for it - I'd actually love to see the kind of advancements we could make in balancing and optimising those competing concerns! But for an allegedly profit-making project, we're kinda already giving them the benefit of the doubt when we talk about the cooling system size, because that's actually the most viable way to minimise weight, and therefore launch costs, with current tech.
It's also why a lot of us are so skeptical: datacenters, which are a notoriously power-dense and cooling-heavy use case in general, just seem inherently unsuited to the environment. In a space elevator future with $0.10/kg cost to orbit the calculations change pretty dramatically, but if you're looking that far off you need to account for things like nuclear fusion too. Right now it's like starting a business growing coconuts in Antarctica, or building a golf course in the middle of the Sahara - we probably could do it, even with current technology, but it's going to be wildly expensive and difficult for a result that's no better than doing it the easy way.
*You absolutely need to account for energy input costs too, but the incoming solar is free and abundant, and panels are relatively cheap and lightweight to launch, so that's rarely a breaking concern for earth-orbital systems. It gets trickier if you're doing scientific missions further from the sun...
Are the launch costs as relevant for infrastructure already in space? If I launch some sort of base specifically designed to collect and process/radiate waste heat from other facilities for a fee for example.
Other facilities would need to already have their own ability to radiate waste heat already to avoid frying themselves until your secondary solution came around, so you would be losing the main advantage of having some secondary system to repurpose waste heat in the first place (even if the tech is there, and I'm dubious that it is for something like that).
I'm pretty sure we can make modular heat sinks that could be swapped. I agree it's a bit of a chicken/egg problem. We've run into this on earth as well with things like EV infrastructure and other technologies that require some sort of secondary thing to exist to enable them to be more efficient / cheaper.
No. Thrust requires a propellant, not heat.
No, generating energy from heat requires both a hot and a comparatively very cold region, so if anything you would need more cooling, plus an entire turbine apparatus, plus "using" the heat to generate power wouldn't actually make it go anywhere, it would still be hot as hell
I don't understand how heat would be relevant to manufacturing besides the energy angle I've already addressed
No. Unmanaged heat would continue to build up until it cooked the circuitry to death, let alone any life on board. Exact opposite of life support.
Can you elaborate on what you mean by this? I can think of some ways to store heat but none of them would actually make a spacebound facility colder.
If I'm going into XKCD mode... if we interpret 'heat' as energy, then thrust can be generated with energy alone in the absence of reaction mass (propellant) so long as you have terrifyingly large quantities of energy.
Safety note. If you do have those energy levels at your disposal and want to use it to exert thrust just to say you did, please avoid thrusting directly away from earth as the secondary effects would be... impolite.
Multiple propulsion systems are based on heating a propellant, not just igniting it directly, e.g. solar thermal rocket, [nuclear propulsion](https://www.nasa.gov/space-technology-mission-directorate/tdm/space-nuclear-propulsion/#:~:text=NASA's%20Space%20Nuclear%20Propulsion%20(SNP,across%20the%20outer%20solar%20system.).
Energy generation sources that don’t use turbines and rely on heat have been used in space since 1961
Ah, I wasn't aware of that technology. The drawback of that is that you have to expend propellant constantly to cool the place down. It's not exactly the same as "using the heat as propellant when you need thrust".
How did I forget about the power source on the voyager probes... Ok that doesn't require a turbine but still needs a hot region and cool region to work, which means you still need to get rid of the heat somehow, which will be building up as you use it because thermocouples don't cool things down.
Preamble: I am not an expert in any of this. Just interested in it.
You can use anisotropic radiation of waste heat to provide a tiny amount of thrust. See: Support for the thermal origin of the Pioneer anomaly.
I was also able to find EQUULEUS which appears to use the waste heat from communications equipment in part of it's propulsion systems.
Doing some basic research on this I seemed to hit three topics:
Radioisotope thermoelectric generator (RTG)
Stirling Engines
Brayton Cycles
I'm not sure if these are explicitly using waste heat, but many of them consume the heat generated by some other power generation process and recover that to generate more electricity.
Though you are just moving the forms of energy around with this. IE if you charge a battery, it will eventually release most of that energy back out as heat anyways. It's just more useful to have electricity instead of heat.
Smelting? Welding? Refining? This is the topic I have done the least research into. We use heat to do these things on earth. If we mine other planets and/or asteroids it may make more sense for a lot of the refining and manufacturing of those raw materials to happen in space before moving back into the atmosphere. Especially if waste heat is a cheap and plentiful source of energy.
Pump the heat into a sink or storage device (molten salt? etc?). Another vessel swaps the sink with a cold sink and transports the heat to somewhere it can be used for something beneficial.
I think the problem with this idea is that waste heat is usually not hot enough for industrial applications. And because the system will always try to move towards equilibrium, if you have enough heat for industrial applications, eventually your whole satellite/space base will be that temperature, and any electronics will have failed long before you get there. The system is always trying to average the temperature across the whole satellite.
There’s an interesting XKCD What-If? about a hairdryer sealed in a box, and what to expect, which touches on equilibrium of energy input and external temperature assuming it’s in an atmosphere which is convecting away the heat. But it also looks at the simplified relationship between energy input and heat, because all the energy you pump into the system has to go somewhere. So if it’s not radiated away, then it’s sticking around.
With things like heat pumps, you can move it around a bit, so you can make one room hotter while you make the other room colder. Theoretically (but I don’t think we have the current technology to achieve) you can get to the kinds of extremes where you have some parts of your satellite cool enough to run a data centre, and the other end hot enough to run a foundry. But you have to constantly be adding energy into the system just to run these heat pumps, because thermodynamics is always trying to bring the whole satellite towards the average temperature. And heat pumps that we have in houses will dramatically lose efficiency when trying to operate outside the narrow ideal band of temperatures they’re designed for, so you would definitely need to be using different technology to enable foundry-to-cool room levels of heat pump effectiveness.
I guess in my head I don’t envision a foundry attached to a data center. I envision heat sinks being moved from a data center to a foundry. Two completely different facilities.
Heat sinks like that would fall under the category of requiring us to make something else very cold for the purpose of heat transfer in an environment that is already about as extremely inhospitable for heat transfer as it can get.
In that case I think you might have made the situation even more complicated, because the combo foundry + data centre has a place to put the hot and a place to put the cold, so the problem is only how to effectively move heat around where you need it as you need it.
But having them isolated, with stored heat bricks to be shipped off to the other facility, means you’re already working with the challenge of moving heat around, but with the added complexity of trying to capture industrial levels of heat in a relatively small object that somehow doesn’t melt the pods/shuttle that ferries back and forth between the facilities, and also now your heat needs to be quantised into “how much useful heat (or cold) can this brick hold without being damaged” rather than just moving heat around continuously.
The driving force behind space based data centres is power consumption, right? In that there's a tension between (1) the growth curve that the tech sector desires for their LLM training data centres, and (2) various countries' desires (and abilities) to build out a power grid to support it.
So, uh, dumb question, but why aren't any of these tech CEOs talking about space-based solar power? I mean, if we're going to shoot for the stars anyways, why not do so with a giant space-borne microwave laser?
I'll just paste my comment from elsewhere in this thread - basically the issue is that's a harder sell. Satellites beaming laser-power to the surface sounds more intuitively challenging than cooling in very cold (but awful heat transmissivity) space.
There are a lot of well-funded space based solar power/power beaming initiatives going on as well, though.
(I believe your markdown link accidentally inverted the ], btw)
Fair enough! I also agree that proposing space-based data centres is absurd at present, but I was mostly curious as to why they'd draw the line at lasing earth from space.
Fixed! Thanks. I always mess those up, especially on mobile.
In practical terms, it doesn't make sense to do. Granted, it makes more sense than space based data centers, but it's still pretty pointless.
The link says that the sun is 144% as intense in space as on the ground, which is great. The problem is that another great thing about solar panels is that they're very, very cheap, and getting cheaper every year. It's just insane how cheap they are. They're easy to build and don't require a ton of energy in their manufacture. What isn't cheap is lifting them into space.
So if they generate 1.44 times as much power in space, we could just build twice as many, or three times as many to cover for the fact that it's night time on earth sometimes. It would still be way, way cheaper to do that than to launch them into space.
If ground area is an issue (it isn't, theres plenty of desolate land on earth) you can put them in giant desert regions, or the oceans, both of which are way more practical than putting them into space.
I think everyone knows that, and the real reason why CEOs are talking about AI data centers is because AI is the thing that gets capital excited right now though.
Makes sense! As noted in another comment, I was moreso curious why the line had been drawn at building data centres in orbit, since both proposals seem equivalently credulous (naively).
Oh 100% agreed. I figure that there are still some environmental concerns to tackle in those -- afaik deserts aren't completely devoid of life, and even the the open expanse of the ocean harbours photosynthetic life which could be affected by blanketing the oceans with solar panels. But those concerns aren't insurmountable, and are absolutely more feasible than launching said materials into an incredibly hostile environment.