I doubt you need that much bandwidth or reliability. Training is “on site”, you just need to upload training material once, then download the trained model.
wow, one year back, I had made a prediction to a friend that this is the direction that Starlink will head in. I was thinking it would proceed like this:
1. provide internet.
2. provide CDN.
3. Edge Compute.
4. Full-on cloud.
These guys see to be focussing on what is basically offline processing (AI training).
More like, these guys will be focused on parting VCs from their money.
Datacenters in space makes no sense at all. Even ignoring the huge cost of sending hardware there in the first place, cooling is a massive issue in space. No medium to sink heat into means the only way to cool anything is by running water through giant infrared radiators. Not ideal when cooling is the largest bottleneck in scaling datacenters. Note that they would also have to dissipate the large amounts heat their datacenter satellite gets from being exposed to the Sun.
Also disregard the cost it takes to send a technician for maintenance, of updating hardware, etc.
Can you please not post comments like this? Thoughtful criticism is welcome, of course, but this sort of thing isn't. Besides breaking the site guidelines, it takes threads in less interesting directions and evokes even worse comments from others. We're trying to avoid that here.
"Don't be snarky."
"Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something."
"Don't be curmudgeonly. Thoughtful criticism is fine, but please don't be rigidly or generically negative."
Really, on second look, snark still feels justified here. The issue is with TFA. There is little room for a thoughtful comment in response to something transparent.
Some type of submissions will invariably not result in very deep discussion, when the topic itself is so shallow.
We need you (I don't mean you personally, of course, but all commenters here) to follow the site guidelines regardless of how bad an article is or you feel it is.
Someone else being wrong or some other post being bad isn't a reason to make things worse. Doing so just creates a downward spiral, which it's all too easy to fall into.
Tbh your moderation is normally very restrained and even handed so was a bit surprising to see you take down several borderline overly snarky comments in a row (that just so happen to be directed against VC investors or YC founders).
To me those comments seemed over the line, not borderline! I'd post the same replies wherever I saw comments like that, regardless of who or what they're about.
Stationkeeping is not free, satellite monitoring is not free, and any replacement to any component is now a multi-year, at least 1+ million dollar affair (or most likely a complete replacement, since not many satellites have done in-situ repairs).
Not an expert in this area, but I think that that "just" is hiding a lot of complexity. Plus you also need some remotely operated robots to mount the replacement.
> “We still don’t appreciate the energy needs of this technology… there’s no way to get there without a breakthrough… we need fusion or we need radically cheaper solar plus storage or something” -Sam Altman
It's kind of depressing that the only way to make this tech better is to feed it more energy. (And apparently now to send it to space)
Just spitballing here, but what if you built it on Earth, and then used the savings to build a second one on the opposite side of Earth? Now you have equivalently continuous power via solar array and also, as a bonus, air.
Well I guess it's a good thing for them they actually did the math. Please show me where they sell you electricity at ~$0.002/kWh where it's politically stable enough to build super expensive datacenters.
It’s is on earth as well using solar and batteries. What is likely to get cheaper faster? Solar and batteries? Or lifting datacenters to space? The world is almost at the point of deploying 1TW/year of solar, and batteries are catching up. No space required. There aren't a lot of VC investment opportunities speeding the rate of solar and battery deployments though.
The argument probably is that battery advances require not yet existing tech via new chemistry etc while what they are proposing is basically just integrating tech that already exists
Power in needs to equal heat out, and that isn't easy in space. They, deceptively, claim that their novel solution is radiative cooling. Relying on radiation for cooling in space is the problem statement! Convective (as on Earth) is significantly more effective.
I'm not one of those idiots who would claim that "we should focus on terrestrial problems instead of space," but this idea seems to have only downsides.
Massive radiators. The ISS has radiators that have a dissipation capacity of about 3m^2/kW. If we use that number, we'd need a 3000m^2 radiator per megawatt, which is the scale they're talking about. This could theoretically be brought down, but not even by an order of magnitude.
I wonder how much cooling the solar panels alone would need, when operating at that scale.
You cool the fluid by flowing it through the radiator. The radiator emits heat radiation into space and cools down the fluid. As long as the fluid is hotter than the equilibrium temperature of the radiator (determined by radiator, space and sun radiation), it will emit more energy than it receives and cool down the fluid.
Passive cooling refers to "passive radiative cooling"[0]. This is a well established technique, but I have doubts on how well it will scale with the heat generated by computation.
Radiative cooling works by exploiting the fact that hot objects emit electromagnetic radiation (glow), and hot means everything above absolute zero. The glow carries away energy which cools down the object. One complication is that each glowy object is also going to be absorbing glow from other objects. While the sun, earth, and moon all emit large amounts of glow (again, heat radiation), empty space is around 2.7 Kelvin, which is very cold and has little glow. So the radiative coolers typically need to have line of sight to empty space, which allows them to emit more energy than they absorb.
This is exactly right, and an important fact is that there is a limited bandwidth for heat radiation. So essentially they need to create a giant lightbulb...
> Additionally, deep space is cold, which is accurate in that the "effective" ambient temperature is around -270°C, corresponding to the temperature of the cosmic microwave background.
There's a lot of bad information in their document too. This -270C temperature is ambient space, i.e. deep space. You may experience this when you're in the shadow of Earth or on the dark side of the moon but you're going to switch that negative sign to a positive when you're facing the sun... Which is clearly something they want to do considering that they are talking about solar power. Which means they have to deal with HEATING as well! I don't see any information about this in the document.
> he mass of radiation shielding scales linearly with the container surface area, whereas the compute per container scales with the volume
This is also a weird statement designed to be deceptive. Your radiation shielding is a shell enclosing some volume.
> Therefore the mass of shielding needed per compute unit decreases linearly with container size.
They clearly do not understand the mass volume relationship here. Density (ρ) is mass (m) divided by volume (V).
m = ρV.
Let's simplify and assume we're using a sphere since this is the most efficient, giving V = 4/3r^3. Your shield is going to be approximately constant density since you need to shield from all directions (can optimize by using other things in your system).
m ∝ ρr^3
I'm not sure what here is decreasing nor what is a linear relationship. To adjust this to a shell you just need to consider the thickness so you can do Δr = r_outer - r_inner and that doesn't take away the cubic relationship.
FWIW, i think their description for the radiation shielding is fine. Your analysis is off. If we assume the spherical case, the mass of the shielding is proportional to surface area, not the volume[0]. You might be confusing general radiation shielding and thermal shielding. Thermal shielding is easier because you can point things towards the sun, earth, and moon.
I am more concerned about heat dissipation, which should scale with surface area, but heat generation scales with compute volume.
[0]:
shell thickness, t
compute radius, r
shell volume is (r+t)^3 - r^3 = 3 r^2 t + 3 r t^2 + t^3 = O(r^2)
shielding/compute is O(r^2)/O(r^3) = O(1/r), ie their linear decrease
Followup question, wouldn't nearly any cooling solution that works in space also work on the ground? Radiative cooling is the most basic/common cooling solution on the ground, the main challenge is just figuring out how to to move heat from the component to the radiator, which I don't think is solved by simply putting it in space?
> Radiative cooling is the most basic/common cooling solution on the ground
Thats tricky. I know the heat exchange components are called radiators but most of the heat they give off is by convection not radiation. (At least here on the ground.) I heard 80%-20% rule of thumb.
But you are right in the broad strokes. Cooling is not easier in space. Mostly because you have no convective heat transfer.
Oh right, that makes sense. So the argument is that comparing a 50C GPU+radiator in a 20C room vs a 50C GPU+radiator in 0K space, the one in space will dissipate more heat via radiation than the one on the ground? As you say, I'd expect that air cooling is much better than EM radiation, but I guess there is some basis for claiming the possibility that cooling in space is somehow better than on the ground, however unlikely.
Is radiative cooling the most common on Earth? I don't think so. Most terrestial "radiators" actually work with convection, ie moving relatively cold air across hot metal fins, which doesn't work in space.
I think other have already corrected you, but radiative cooling is probably the least common on the ground and the only viable option in space.
I can help explain why. On earth, we are surrounded by stuff. Radiative cooling relies on thermal radiation leaving an object. Crucially, it also requires the object to absorb less thermal radiation than it emits. On earth we are surrounded by stuff, including air, that emits thermal radiation. There is a window of wavelengths, called the atmospheric window[0], that will allow parts of the thermal radiation out into space, rather than returned back. Imagine shining a flashlight on tinted glass, the light will get through depending on the color. If the light gets through, it has escaped. If not, the light is returned and heats up your surroundings again.
Also on earth the other methods (conduction, convection, and phase changes) are more effective. The earth can be used as a very big heat sink. On a spaceship or satellite, you don't have the extra mass to store the energy, so radiative is the only option.
I very much agree here. This is clearly nothing more than a scam meant to separate investors from their money. I read the "white paper" and very few of their suppositions actually hold water, as many others have detailed in this thread. It seems like what this kind of "startup" does is pick two or three trending words, put them together and make up a story that would be believable to people who don't dig too deep. These guys chose the words "AI" and "space" and figured out how to stick them together. Speaking of that, anyone want to invest in my new quantum blockchain technology?
> Their design calls for a cluster of shipping container-style boxes packed with high-speed AI chips. These would be anchored at the centre of a 16 sq km array of solar panels generating up to five gigawatts of power — about 25 per cent more than Drax, Britain’s biggest power station. The mammoth structure would circle the Earth in “sun synchronous” orbit so that it is never in shade
Their whitepaper clearly demonstrates a profound lack of knowledge of thermal engineering. E.g. heat pumps are described as magical things.
They are literally planning to feed the radiators using a coolant like water and sensible heat at 35 degC to 5 degC. At 5 GW, you then need to be pumping 60 000 liters of water per second.
That's like a tenth of the Sacramento river, going through a 16 sq km array in space and hoping that nothing leaks.
Even if they somehow figure out all these problems. how do you manage a space based data centre? do you have rotating staff living there? or are they just praying that nothing ever goes wrong??? Isn't radiation a massive problem in space? i would expect consumer grade hardware to be constantly flipping bits accidentally that shouldn't have flipped.
I was wondering if these server racks in space would need to be specifically designed for enough radiative cooling. Apparently the answer is yes: the radiators would be expansive and placed on the reverse side of the solar panels.
Starcloud is developing a lightweight deployable radiator design with a very
large area - by far the largest radiators deployed in space - radiating primarily towards deep space, which has an average temperature of about 2.7 Kelvin or -270°C. The radiators can be positioned in-line with the solar
arrays as shown in Figure 3, with one side exposed to sunlight.
…
Figure 3. A data center in Sun Synchronous Orbit, showing a 4km x 4km deployed solar array and radiators.
Definitely an out of this world idea. I wonder if their micro datacenter is going to be self-sufficient power wise using only solar energy? And how would they address the hardware failures that are likely when you train large language models at scale?
Given the purported cost benefits in their whitepaper [1], hardware failures might be an irrelevant rounding error. They suggest something to the tune of 100x cheaper.
The white paper back of the envelope calculations show a 4km x 4km solar panels and radiators are required for a 5 GW datacenter. I am not sure how the authors were not cracking up while writing that white paper.
I've been saying for a long time that we should consider remote areas for building datacenters for batch processing.
At first I thought the poles (of the planet) might be good. The cooling is basically free. But the energy and internet connectivity would be a problem. At the poles you can really only get solar about three months a year, and even then you need a lot of panels. Most of Antarctica is powered diesel because of this.
So the next thought was space. At the time, launching to space was way too costly for it to ever make sense. But now, with much cheaper launches, space is accessible.
Power seems easily solved. You can get lots of free energy from the sun with some modest panels. But to do that requires an odd orbit where you wouldn't be over the same spot on earth, which could make internet access difficult. Or you can go geostationary over a powerful ground station, but then you'd need some really big batteries for all the time you aren't in the sun.
But cooling is a huge problem. Space is cold, but there is no medium to transfer the heat away from the hot objects. I think this will be the biggest sticking point, unless they came up with an innovative solution.
I have no clue about space technology but many comments point the difficulty to cool anything in space. If Starcloud had an innovative solution to this problem, why on Earth (sic) focus on data centers when they could help the entire space industry? It does not smell good.
>But cooling is a huge problem. Space is cold, but there is no medium to transfer the heat away from the hot objects. I think this will be the biggest sticking point, unless they came up with an innovative solution.
Their main tech breakthrough would have to be in this area otherwise the company is worthless imo.
It's possible to do all of this with current technology. Just... Why? The cost would be exorbitant; even with really clever deployment tech, the launch costs are gonna be dominated by solar panels and radiators.
This is a super cool idea and seems like perfect investor-bait. That's about where it ends.
Apocalyptic scenarios where terrestrial communication methods going back over a century are no longer feasible, but we can still readily talk to space? And maintain/replace the stuff we have up there?
Like the commenter, your error is thinking the hedge is for you. And what's more—you're assuming the scenario must be total destruction, there's a gradation of disruption where this hedge remains feasible and even vital.
Genuinely most "AI" DCs are spending less than 9KW on cooling for every 100KW of servers. If you were that bothered about getting that to zero, you could literally sink them into the ocean, build a heat network so the town can take the heat for free or use any of a dozen more established and practical ways to do that.
It's a bit demoralizing how many suggestions in this thread would have significant environmental effects beyond what large scale AI training already has.
I'm talking about the above proposals (albiet hypothetical) to either cover a pole of our planet in solar and other ocean based proposals--not solar in general.
It's a bit demoralizing people talk about AI training as if it were even 1/100th the environmental impact of the personal automobile or frequent airplane trips
I don't think they can bend the laws of physics though. Vacuum means the only way to dissipate heat is through thermal radiation, hence the huge infrared radiators.
I'd imagine the bandwidth limitations are physical and similar for Starlink or other satellites either now or very soon. I was replying to the LoS concerns, I agree overall this doesn't work, but I don't think it's because of LoS problems.
> As conduction and convection to the environment are not available in space, this means the data center will require radiators capable of radiatively dissipating gigawatts of thermal load. To achieve this, Starcloud is developing a lightweight deployable radiator design with a very large area - by far the largest radiators deployed in space - radiating primarily towards deep space...
They claim they can radiate "633.08 W / m^2". At that rate, they're looking at square kilometers of radiators to dissipate gigawatts of thermal load, perhaps hectares of radiators.
They also claim that they can "dramatically increase" heat dissipation with heat pumps.
So, there you have it: "all you have to do" is deploy a few hectares of radiators in space, combined with heat pumps that can dissipate gigawatts of thermal load with no maintenance at all over a lifetime of decades.
This seems like the sort of "not technically impossible" problem that can attract a large amount of VC funding, as VCs buy lottery tickets that the problem can be solved.
An object of that size in orbit seems like it'd run into problems developing sizable holes due to space junk and whathaveyou. There's probably some maintenance...
It’s only a grift if they know they can’t solve the cooling issue and they falsify data around their proposed solution and they publicly embarrass their investors a la Theranos.
Outside of that, accepting money and saying “I will simply solve the enormous problem with my idea by solving it” is not only normal, but actively encouraged and rewarded in the VC sphere. Suggesting that that way of operating is anything short of the standard that should be aspired to is actually seen as derisive and offensive on here and can get you labeled as gauche or combative.
I'd argue that some of the assumptions made in the whitepaper are so egregiously optimistic that they cross the line into grifting, but it's impossible to know the true intentions of the founders.
For one, the cost they ascribe to the space bound solar array being only $2 million for 40 MW is pretty out there.
Or we could build a large vacuum chamber here on Earth and put a data center in it, if the goal is to make cooling as difficult as possible. "My data center is too hot! It's burning me!" "Put it in a giant thermos, then you won't feel it anymore."
> They also claim that they can "dramatically increase" heat dissipation with heat pumps.
Right, great idea. Start with the heat where you don't want it -- in the chip -- and pump it out to where it can't go anywhere. Then you can recirculate the medium back and have slightly older heat that you can mix with the new heat! It'll be a heat party!
It's just like a terrestrial heat pump, where you pump the heat out to where you have a huge environmental sink to transfer the heat to. In space, you have something like a hundred thousand hydrogen atoms per cubic meter to take up the heat. A HUNDRED THOUSAND! That's a bigly number, it must work out. We can always make those atoms go really, really fast!
Obviously use the heat pumps to concentrate the thermal energy up to 2700k, then conduct it along a bunch of tungsten filaments, now it's the world's biggest incandescent lightbulb on top of being the first datacenter in space. Maybe get it up to 4000k for a more modern lighting look. Guess we're gonna assume the dark forest hypothesis is false.
This is the best idea to come out of this whole scheme. Space solar panels are super cheap and efficient? Prove it! Launch them and transmit the energy down.
This is orders of magnitude easier than the original proposal -- and yet still nonsensical.
Beaming power down to Earth from space-based solar collectors is a concept that's been around for a while.
"Dr. Glaser is best known as the inventor of the Solar Power Satellite concept, which he first presented in the journal Science for November 22, 1968 (“Power from the Sun: It’s Future”). In 1973 he was granted a U.S. patent on the Solar Power Satellite to supply power from space for use on the Earth."
One thing that always struck me was that you wouldn't want to be living near the "collectors". A very small angular error in beaming could result in being literally microwaved.
> "A very small angular error in beaming could result in being literally microwaved."
One of the SimCity games had this as an occasional disaster event. You had to make sure your ground collector stations weren't too close to the rest of the city or risk setting your buildings on fire.
My thermodynamics is rusty, but if they could concentrate it, that would mean that it's part of the "free energy" that could be used to power the data center itself, and given that it's impossible in practice to have perfect efficiency, there will always be excess heat that cannot be concentrated (entropy going up).
The whitepaper shows a 4km x 4km solar array, which is 1600 hectares (3200 International Space Stations). Would assume the array they're proposing would be cheaper since its structurally more homogenous, but $480 trillion dollars is a whole lot of money.
> So, there you have it: "all you have to do" is deploy a few hectares of radiators in space, combined with heat pumps that can dissipate gigawatts of thermal load…
Starcloud’s whitepaper suggests a 4 km × 4 km radiator. For comparison, the James Web Space Telescope has a sunshield measuring 21 m × 14 m and the International Space Station measures 109 m × 73 m.
Heat pumps could dramatically impact performance by increasing the temperature of the radiators. The hotter they are, the more power they can dissipate per unit time & area.
Doubling the radiator temperature would give you 16x more radiated power.
> they're looking at square kilometers of radiators to dissipate gigawatts of thermal load
Presumably they'll put them behind the 4km2 solar panels!
I mean this is a ridiculous concept. We've never put anything remotely that size into space. To argue that this would be cheaper than putting something underwater or in the middle of nowhere is crazy. I'd rather deal with salt than deal with radiation.
> I've been saying for a long time that we should consider remote areas for building datacenters for batch processing.
FWIW there's a reason that Sweden has a bunch of datacenters in the north that are peanuts compared to hosting in Virginia.
They're "poorly" connected (by virtue of being a bit out of the way), but the free cooling and power from renewables make them extremely attractive. There was a time where they were the favourite of crypto-miners for the same reason as they would be attractive to AI training farms.
As for the meat of the paper. Anyone with a passing understanding of space will be quick to point out that:
A) Heat is a problem in space, it's either way-way-way to hot (IE; you're in the path of the Sun) or it's way-way-way too cold (IE; you're out of the sun) and the shift between the two means you need to build for both. You also can't dissipate heat as there's no air to take the heat away.
B) Power is not so abundant and solar panels degrade; a huge amount of satellite building is essentially managing a decline in the capability of hardware. That's part of why there are so many up there.
C) Getting reasonably sized hardware up there is beyond improbable, though I'll grant you that most of the weight in a computer is the cooling components and chassis.
D) Cosmic Rays. No electromagnetic barrier from earth and extremely tight lithographies. I mean... there's a reason NASA is still using CPU's measured in the megahertz range.
AFAIK someone (Mars Ingenuity helicopter team) discovered that some chips handle them much better than others, so they just test a bunch and keep resistant ones.
Perhaps I'm missing something, but if the only energy they get is coming from the sun, then they only need to dissipate that same amount of heat (minus whatever energy was needed for beaming data down to Earth).
That’s not how it works. With conservation of energy, all the energy coming in to power the computers has to be emitted somehow. Powering computers doesn’t get rid of the energy, it just makes it unusable and converts it into heat.
In principle heat can be converted to work if there are two bodies at different temperature. In space heat can only be irradiated away since there’s no matter available for conduction, so I would think that eventually all of your machinery would thermalize with the environment and the boundary conditions. While there should be a temperature gradient between the side exposed to the sun and the one opposite to it, I’m not sure how much of this is actually harvestable.
Right and that's why the heat is the problem, in space.
But if the collected heat comes from a large area of solar-cells, and is then focused on the small area of a computer or graphics-card, that computer might melt.
- "But to do that requires an odd orbit where you wouldn't be over the same spot on earth, which could make internet access difficult".
Surely you'd want to use satellite constellations as relays? There's thousands of those satellites in line-of-sight all the time.
It's strictly superior on pure geometry anyway (I think). You have a finite channel capacity between your satellite and your ground station; but different satellites, in non-overlapping microwave spots, are in separate spatial channels.
Integrate compute units + Starlink into solar panels so people can buy them to earn money/tokens. Much cheaper than tangling massive power lines around the planet.
Regarding Internet connectivity regardless of the orbit or location, something like YC co Bifrost Orbital (https://bifrostorbital.com/), might be an option.
Cooling at the south pole is not free- it's actually hard work to keep data centers cool at the pole. My friend helped run the IceCube data center at the station and got to fly down and stay there for a week, basically to insert debian CDs and press buttons.
> Or you can go geostationary over a powerful ground station, but then you'd need some really big batteries for all the time you aren't in the sun.
Geostationary satellites only go into Earth's shadow on around 20 days on each side of an equinox. That leave 280+ days each year where they are in sun all day. Maybe that's enough to be worth it?
Or if you do need to keep the things working even on those ~80 days a year when they do spend part of the day in shadow maybe they could be powered by energy beamed in from those not in shadow? You'd put a bunch in geostationary orbits spread out evenly so that each is close enough to its neighbors for power beaming.
I wonder if something crazy might work? Could you actually connect adjacent satellites by an actual physical power cable, which would also be in geostationary orbit?
I'd guess you'd actually need two conductors in your cable, carrying current in opposite direction to cancel out interactions with Earth's magnetic field so the system doesn't get pushed out of its orbit (which would probably be bad).
There would probably be gravitational interactions like with the Moon that might also make it hard to keep everything in place, but maybe by purposefully sending different currents in opposite directions on some of the links you could purposefully use interactions with the Earth's magnetic field to move the cable back where you wanted?
If the satellites are connected by cables then maybe they could actually be slightly higher than geostationary but moving faster than circular orbital speed at that altitude so there is a net outward force from that, which could be countered by tension in the power cables to force them into a circular path that is still geostationary.
Aside from the obvious cooling issues people have already mentioned, isn't cosmic radiation also very unkind to modern ultra dense silicon? AIUI they tend to use really old silicon processes in space stuff for that reason, and even then they have to build in redundant compute to mitigate logic errors that probably wouldn't happen on Earth.
> They tend to use really old silicon processes in space stuff for that reason.
To be fair that's mostly part "if it works don't change it" and part "that's how we've always done it". SpX uses newer hardware w/ traditional OSs (linux) w/ lots of redundancy.
It sounds like their vision for space-based data centers presupposes nearly-free energy costs, delivered via a colossal solar farm made possible by falling launch costs.
Temporarily putting aside (extremely fair) feasibility questions around those two pre-requisites, data centers are a not-bad choice for things to do with unlimited space energy.
Aluminum smelting or growing food are the two I’d think of otherwise, and neither of those can have inputs/outputs beamed to a global network of high-bandwidth satellites.
Solar energy isn’t that much more efficient in Earth orbit than on Earth - maybe twice as efficient. That sounds nice, but you’re saving half of your solar panel cost while massively increasing every other cost.
The one benefit is being able to be in a synchronous orbit with the sun, so you don’t have to contend with night. However, that’s just another ~doubling of efficiency, which I think still nowhere near makes up for the additional costs.
I would imagine you could launch a new rack, dump the old one, and connect the new one to the existing solar / cooling array. Hopefully with some sort of re-entry and recycling plan for the old one. The sheer size the arrays are going to need to be feel like they are going to be the more important part of it.
Unless they've figured out some impressive cooling tech, which I would expect would be worth more than the rest of their company combined, then this is pretty much DoA. "More efficient cooling architecture taking advantage of higher ΔT in space" would indeed be useful if you had a nice medium to radiate into. It turns out that thermal radiation is incredibly poor into the vacuum of space lol.
Space (with a sunshade) is a nearly perfect medium into which to radiate heat, in the sense that there’s nothing better.
But I agree with your general point. At 100°C, you can radiate about 1kW/m^2. That’s 1000m^2 of radiator per MW of datacenter, assuming you can operate with the radiator at 100°C. You can fudge this a bit with a heat pump (to run the radiator hotter, paying a linear-ish power penalty and gaining a fourth-power radiation benefit), but that’s expensive and that power isn’t free.
Here on Earth, you can cool by conduction or evaporation, which isn’t an option in space.
> Space (with a sunshade) is a nearly perfect medium into which to radiate heat, in the sense that there’s nothing better.
There is radiation but zero convection. As anyone with an oven or PC will tell you, even a very tiny fan makes a big difference in the ability to dump heat. We're not putting our PCs into vacuum chambers for a good reason. A small fan in your oven not only makes for more consistent heating in your food, but it requires less power
Scott Manley has published a video a few months ago explaining why putting data centers in space is an absolutely terrible idea. Lumen Orbit, the company mentioned, is a former name of Starcloud.
Very ambitious but it seems futile if you’re not building the rockets yourself. Personally I’m more bullish on figuring out how to use analog chips to train models.
- You can't build 40MW of solar panels for $2M, even with theoretical maximum efficiency. You can't even build the cabling and regulators at that price.
- You need battery storage -- not as your backup -- but as primary source. It is going to cost more than $2M. Batteries are heavy. They are going to cost a lot to launch. This is not even solved on the ground yet.
- You need a heat transport medium to move heat into your massive radiator. Either you use water or you use air or you use heatpipes (metal). You have to pay for the cost and weight and launch expense. This is probably half the weight of the rack and I haven't bothered to do the math about how you transport heat into a 500 foot solar sail.
- Let's not even talk about how you need to colocate multiple other racks for compute and storage. There aren't any 1TBps orbital link technologies.
- Rad shielding? It doesn't work, but I'll let this slide; it seems like the least problematic part of the proposal.
- 15 year lifetime? GPUs are obsolete after 12 months.
I don't want to be the guy who shoots stuff down just for fun, but this doesn't even pass the sniff test. Maybe you can get 10x cheaper power and cooling in space. Still doesn't work.
Also: repairs. Every time I read someone’s story about large-scale ML training, a bunch of it is about identifying failing or flaky equipment and fixing it. That’s not so easy in space.
Nonsense, it's right there in the acronym: Space Reliability Engineer (or I guess one could also just leave "Site" as is, since space is for sure a site). That PagerDuty rotation is gonna be hell
Temperature isn't a power source; heat flowing across a temperature gradient can be. But that brings us back to the first problem - how to make it flow.
For continuous you need to either go for a polar orbit or go very far in space. Most launch centers & providers are not well situation for polar orbits because its not a common use case, so you need to sacrifice launch mass. The same goes for far away orbits - you need to sacrifice launch mass to go further. Also if you are far then you get latency issues.
So it skews the economics pretty harshly. I think OP is right - you need good batteries somehow.
I think the proposal suggested an orbit where the solar panels are always in sun and always properly aligned and always clean due to space gophers.
But more seriously, GPU loads are super spiky. Ground-based power grids and generators and batteries have trouble keeping up with them. You can go from 1MW idle to 50MW full power in 10ms. Unbuffered solar cells are right out.
> "GPU loads are super spiky... You can go from 1MW idle to 50MW full power in 10ms."
That sounds like something that could be addressed in software, if necessary? Cap/throttle the GPUs according to the available power, and ramp power up/down gradually if spikiness is the issue.
324 comments
[ 3.0 ms ] story [ 292 ms ] thread1. provide internet. 2. provide CDN. 3. Edge Compute. 4. Full-on cloud.
These guys see to be focussing on what is basically offline processing (AI training).
Datacenters in space makes no sense at all. Even ignoring the huge cost of sending hardware there in the first place, cooling is a massive issue in space. No medium to sink heat into means the only way to cool anything is by running water through giant infrared radiators. Not ideal when cooling is the largest bottleneck in scaling datacenters. Note that they would also have to dissipate the large amounts heat their datacenter satellite gets from being exposed to the Sun.
Also disregard the cost it takes to send a technician for maintenance, of updating hardware, etc.
This won't happen. If a satellite fails they will just write it off. Maintenance would be more expensive than depreciation
https://news.ycombinator.com/newsguidelines.html
"Don't be snarky."
"Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something."
"Don't be curmudgeonly. Thoughtful criticism is fine, but please don't be rigidly or generically negative."
https://news.ycombinator.com/newsguidelines.html
Some type of submissions will invariably not result in very deep discussion, when the topic itself is so shallow.
Someone else being wrong or some other post being bad isn't a reason to make things worse. Doing so just creates a downward spiral, which it's all too easy to fall into.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...
[0] - https://starcloudinc.github.io/wp.pdf
It's kind of depressing that the only way to make this tech better is to feed it more energy. (And apparently now to send it to space)
It’s is on earth as well using solar and batteries. What is likely to get cheaper faster? Solar and batteries? Or lifting datacenters to space? The world is almost at the point of deploying 1TW/year of solar, and batteries are catching up. No space required. There aren't a lot of VC investment opportunities speeding the rate of solar and battery deployments though.
https://news.ycombinator.com/item?id=43624308
I'm not one of those idiots who would claim that "we should focus on terrestrial problems instead of space," but this idea seems to have only downsides.
The big radiators on the ISS can only dump a few server racks worth of heat.
https://en.wikipedia.org/wiki/Spacecraft_thermal_control?wpr...
> Most spacecraft radiators reject between 100 and 350 W of internally generated electronics waste heat per square meter.
I wonder how much cooling the solar panels alone would need, when operating at that scale.
Radiative cooling works by exploiting the fact that hot objects emit electromagnetic radiation (glow), and hot means everything above absolute zero. The glow carries away energy which cools down the object. One complication is that each glowy object is also going to be absorbing glow from other objects. While the sun, earth, and moon all emit large amounts of glow (again, heat radiation), empty space is around 2.7 Kelvin, which is very cold and has little glow. So the radiative coolers typically need to have line of sight to empty space, which allows them to emit more energy than they absorb.
[0] https://en.wikipedia.org/wiki/Radiative_cooling
m = ρV.
Let's simplify and assume we're using a sphere since this is the most efficient, giving V = 4/3r^3. Your shield is going to be approximately constant density since you need to shield from all directions (can optimize by using other things in your system).
m ∝ ρr^3
I'm not sure what here is decreasing nor what is a linear relationship. To adjust this to a shell you just need to consider the thickness so you can do Δr = r_outer - r_inner and that doesn't take away the cubic relationship.
https://en.wikipedia.org/wiki/Thermal_radiation#Characterist...
https://en.wikipedia.org/wiki/Black-body_radiation
https://www.nasa.gov/smallsat-institute/sst-soa/thermal-cont...
https://ocw.mit.edu/courses/16-851-satellite-engineering-fal...
I am more concerned about heat dissipation, which should scale with surface area, but heat generation scales with compute volume.
[0]:
shell thickness, t
compute radius, r
shell volume is (r+t)^3 - r^3 = 3 r^2 t + 3 r t^2 + t^3 = O(r^2)
shielding/compute is O(r^2)/O(r^3) = O(1/r), ie their linear decrease
Your thickness is defined by an inner radius and outer.
Thats tricky. I know the heat exchange components are called radiators but most of the heat they give off is by convection not radiation. (At least here on the ground.) I heard 80%-20% rule of thumb.
But you are right in the broad strokes. Cooling is not easier in space. Mostly because you have no convective heat transfer.
I can help explain why. On earth, we are surrounded by stuff. Radiative cooling relies on thermal radiation leaving an object. Crucially, it also requires the object to absorb less thermal radiation than it emits. On earth we are surrounded by stuff, including air, that emits thermal radiation. There is a window of wavelengths, called the atmospheric window[0], that will allow parts of the thermal radiation out into space, rather than returned back. Imagine shining a flashlight on tinted glass, the light will get through depending on the color. If the light gets through, it has escaped. If not, the light is returned and heats up your surroundings again.
Also on earth the other methods (conduction, convection, and phase changes) are more effective. The earth can be used as a very big heat sink. On a spaceship or satellite, you don't have the extra mass to store the energy, so radiative is the only option.
[0] https://en.wikipedia.org/wiki/Atmospheric_window
huh? I was under the impression that cooling in space is an absolute nightmare since radiating heat into vacuum is super hard?
Even the comparatively small and decidedly H100-free ISS needed giant radiators
https://en.wikipedia.org/wiki/External_Active_Thermal_Contro...
I will call it MartianCloud.
One NVIDIA DGX SuperPOD consumes 10 kW which would be ~500 square feet of solar panels and ~100 square feet of radiator area.
> Their design calls for a cluster of shipping container-style boxes packed with high-speed AI chips. These would be anchored at the centre of a 16 sq km array of solar panels generating up to five gigawatts of power — about 25 per cent more than Drax, Britain’s biggest power station. The mammoth structure would circle the Earth in “sun synchronous” orbit so that it is never in shade
This plan seems about as realistic as Bluthton though. https://www.reddit.com/r/arresteddevelopment/comments/1gtyvv...
They are literally planning to feed the radiators using a coolant like water and sensible heat at 35 degC to 5 degC. At 5 GW, you then need to be pumping 60 000 liters of water per second.
That's like a tenth of the Sacramento river, going through a 16 sq km array in space and hoping that nothing leaks.
Then they just need to worry about the extra friction heat generated by moving that much liquid.
Starcloud is developing a lightweight deployable radiator design with a very large area - by far the largest radiators deployed in space - radiating primarily towards deep space, which has an average temperature of about 2.7 Kelvin or -270°C. The radiators can be positioned in-line with the solar arrays as shown in Figure 3, with one side exposed to sunlight.
…
Figure 3. A data center in Sun Synchronous Orbit, showing a 4km x 4km deployed solar array and radiators.
https://www.starcloud.com/wp
[1] https://starcloudinc.github.io/wp.pdf
At first I thought the poles (of the planet) might be good. The cooling is basically free. But the energy and internet connectivity would be a problem. At the poles you can really only get solar about three months a year, and even then you need a lot of panels. Most of Antarctica is powered diesel because of this.
So the next thought was space. At the time, launching to space was way too costly for it to ever make sense. But now, with much cheaper launches, space is accessible.
Power seems easily solved. You can get lots of free energy from the sun with some modest panels. But to do that requires an odd orbit where you wouldn't be over the same spot on earth, which could make internet access difficult. Or you can go geostationary over a powerful ground station, but then you'd need some really big batteries for all the time you aren't in the sun.
But cooling is a huge problem. Space is cold, but there is no medium to transfer the heat away from the hot objects. I think this will be the biggest sticking point, unless they came up with an innovative solution.
Their main tech breakthrough would have to be in this area otherwise the company is worthless imo.
This is a super cool idea and seems like perfect investor-bait. That's about where it ends.
Something like 2/3rds of sunshine is already being absorbed by oceans. How much solar power do humans harvest? A billionth?
Routing through starlink should have direct LoS at all times.
(probably not)
> As conduction and convection to the environment are not available in space, this means the data center will require radiators capable of radiatively dissipating gigawatts of thermal load. To achieve this, Starcloud is developing a lightweight deployable radiator design with a very large area - by far the largest radiators deployed in space - radiating primarily towards deep space...
They claim they can radiate "633.08 W / m^2". At that rate, they're looking at square kilometers of radiators to dissipate gigawatts of thermal load, perhaps hectares of radiators.
They also claim that they can "dramatically increase" heat dissipation with heat pumps.
So, there you have it: "all you have to do" is deploy a few hectares of radiators in space, combined with heat pumps that can dissipate gigawatts of thermal load with no maintenance at all over a lifetime of decades.
This seems like the sort of "not technically impossible" problem that can attract a large amount of VC funding, as VCs buy lottery tickets that the problem can be solved.
https://news.ycombinator.com/newsguidelines.html
You have to find trillions of dollars of future launches to justify current valuations.
Outside of that, accepting money and saying “I will simply solve the enormous problem with my idea by solving it” is not only normal, but actively encouraged and rewarded in the VC sphere. Suggesting that that way of operating is anything short of the standard that should be aspired to is actually seen as derisive and offensive on here and can get you labeled as gauche or combative.
For one, the cost they ascribe to the space bound solar array being only $2 million for 40 MW is pretty out there.
> They also claim that they can "dramatically increase" heat dissipation with heat pumps.
Right, great idea. Start with the heat where you don't want it -- in the chip -- and pump it out to where it can't go anywhere. Then you can recirculate the medium back and have slightly older heat that you can mix with the new heat! It'll be a heat party!
It's just like a terrestrial heat pump, where you pump the heat out to where you have a huge environmental sink to transfer the heat to. In space, you have something like a hundred thousand hydrogen atoms per cubic meter to take up the heat. A HUNDRED THOUSAND! That's a bigly number, it must work out. We can always make those atoms go really, really fast!
Did an AI invent this whole scheme?
This is orders of magnitude easier than the original proposal -- and yet still nonsensical.
... could we instead beam the energy down to a data center on the sea floor?
"Dr. Glaser is best known as the inventor of the Solar Power Satellite concept, which he first presented in the journal Science for November 22, 1968 (“Power from the Sun: It’s Future”). In 1973 he was granted a U.S. patent on the Solar Power Satellite to supply power from space for use on the Earth."
One thing that always struck me was that you wouldn't want to be living near the "collectors". A very small angular error in beaming could result in being literally microwaved.
https://nss.org/in-memoriam-peter-e-glaser-1923-2014/
One of the SimCity games had this as an occasional disaster event. You had to make sure your ground collector stations weren't too close to the rest of the city or risk setting your buildings on fire.
https://www.youtube.com/watch?v=4CrTJEY8zNs
Such fun.
From my AI data centre project??
Get out of here, we’re only interested in getting rich, we don’t actually care about doing something _useful_ for people.
The whitepaper shows a 4km x 4km solar array, which is 1600 hectares (3200 International Space Stations). Would assume the array they're proposing would be cheaper since its structurally more homogenous, but $480 trillion dollars is a whole lot of money.
Starcloud’s whitepaper suggests a 4 km × 4 km radiator. For comparison, the James Web Space Telescope has a sunshield measuring 21 m × 14 m and the International Space Station measures 109 m × 73 m.
Webb took a long time because this stuff is very, very challenging. One of its primary engineering challenges was… cooling!
Doubling the radiator temperature would give you 16x more radiated power.
I mean this is a ridiculous concept. We've never put anything remotely that size into space. To argue that this would be cheaper than putting something underwater or in the middle of nowhere is crazy. I'd rather deal with salt than deal with radiation.
FWIW there's a reason that Sweden has a bunch of datacenters in the north that are peanuts compared to hosting in Virginia.
They're "poorly" connected (by virtue of being a bit out of the way), but the free cooling and power from renewables make them extremely attractive. There was a time where they were the favourite of crypto-miners for the same reason as they would be attractive to AI training farms.
Fortlax has some I believe; https://www.fortlax.se
-----
As for the meat of the paper. Anyone with a passing understanding of space will be quick to point out that:
A) Heat is a problem in space, it's either way-way-way to hot (IE; you're in the path of the Sun) or it's way-way-way too cold (IE; you're out of the sun) and the shift between the two means you need to build for both. You also can't dissipate heat as there's no air to take the heat away.
B) Power is not so abundant and solar panels degrade; a huge amount of satellite building is essentially managing a decline in the capability of hardware. That's part of why there are so many up there.
C) Getting reasonably sized hardware up there is beyond improbable, though I'll grant you that most of the weight in a computer is the cooling components and chassis.
D) Cosmic Rays. No electromagnetic barrier from earth and extremely tight lithographies. I mean... there's a reason NASA is still using CPU's measured in the megahertz range.
AFAIK someone (Mars Ingenuity helicopter team) discovered that some chips handle them much better than others, so they just test a bunch and keep resistant ones.
But if the collected heat comes from a large area of solar-cells, and is then focused on the small area of a computer or graphics-card, that computer might melt.
Yes, let's go ahead and finish melting the ice caps, great idea
Surely you'd want to use satellite constellations as relays? There's thousands of those satellites in line-of-sight all the time.
It's strictly superior on pure geometry anyway (I think). You have a finite channel capacity between your satellite and your ground station; but different satellites, in non-overlapping microwave spots, are in separate spatial channels.
Geostationary satellites only go into Earth's shadow on around 20 days on each side of an equinox. That leave 280+ days each year where they are in sun all day. Maybe that's enough to be worth it?
Or if you do need to keep the things working even on those ~80 days a year when they do spend part of the day in shadow maybe they could be powered by energy beamed in from those not in shadow? You'd put a bunch in geostationary orbits spread out evenly so that each is close enough to its neighbors for power beaming.
I wonder if something crazy might work? Could you actually connect adjacent satellites by an actual physical power cable, which would also be in geostationary orbit?
I'd guess you'd actually need two conductors in your cable, carrying current in opposite direction to cancel out interactions with Earth's magnetic field so the system doesn't get pushed out of its orbit (which would probably be bad).
There would probably be gravitational interactions like with the Moon that might also make it hard to keep everything in place, but maybe by purposefully sending different currents in opposite directions on some of the links you could purposefully use interactions with the Earth's magnetic field to move the cable back where you wanted?
If the satellites are connected by cables then maybe they could actually be slightly higher than geostationary but moving faster than circular orbital speed at that altitude so there is a net outward force from that, which could be countered by tension in the power cables to force them into a circular path that is still geostationary.
To be fair that's mostly part "if it works don't change it" and part "that's how we've always done it". SpX uses newer hardware w/ traditional OSs (linux) w/ lots of redundancy.
I can't imagine running bleeding edge GPUs in a particle accelerator and getting reasonable results.
Temporarily putting aside (extremely fair) feasibility questions around those two pre-requisites, data centers are a not-bad choice for things to do with unlimited space energy.
Aluminum smelting or growing food are the two I’d think of otherwise, and neither of those can have inputs/outputs beamed to a global network of high-bandwidth satellites.
But I agree with your general point. At 100°C, you can radiate about 1kW/m^2. That’s 1000m^2 of radiator per MW of datacenter, assuming you can operate with the radiator at 100°C. You can fudge this a bit with a heat pump (to run the radiator hotter, paying a linear-ish power penalty and gaining a fourth-power radiation benefit), but that’s expensive and that power isn’t free.
Here on Earth, you can cool by conduction or evaporation, which isn’t an option in space.
https://www.youtube.com/watch?v=d-YcVLq98Ew
- You can't build 40MW of solar panels for $2M, even with theoretical maximum efficiency. You can't even build the cabling and regulators at that price.
- You need battery storage -- not as your backup -- but as primary source. It is going to cost more than $2M. Batteries are heavy. They are going to cost a lot to launch. This is not even solved on the ground yet.
- You need a heat transport medium to move heat into your massive radiator. Either you use water or you use air or you use heatpipes (metal). You have to pay for the cost and weight and launch expense. This is probably half the weight of the rack and I haven't bothered to do the math about how you transport heat into a 500 foot solar sail.
- Let's not even talk about how you need to colocate multiple other racks for compute and storage. There aren't any 1TBps orbital link technologies.
- Rad shielding? It doesn't work, but I'll let this slide; it seems like the least problematic part of the proposal.
- 15 year lifetime? GPUs are obsolete after 12 months.
I don't want to be the guy who shoots stuff down just for fun, but this doesn't even pass the sniff test. Maybe you can get 10x cheaper power and cooling in space. Still doesn't work.
The biggest problem is software. The CUDA stack is not maintained forever and certainly less than 15 years.
... couldn't you just merge both problems into a solution - your radiators ARE you power source
How so? Is it not possible to position the satellite in an orbit that keeps it in perpetual sunlight?
Not arguing with your overall point - this company looks ridiculous.
So it skews the economics pretty harshly. I think OP is right - you need good batteries somehow.
But more seriously, GPU loads are super spiky. Ground-based power grids and generators and batteries have trouble keeping up with them. You can go from 1MW idle to 50MW full power in 10ms. Unbuffered solar cells are right out.
That sounds like something that could be addressed in software, if necessary? Cap/throttle the GPUs according to the available power, and ramp power up/down gradually if spikiness is the issue.
Just that it tends to involve heavy AF materials like water
I guess you need connections too, and maybe a previous exit.
This idea in particular doesn't make any sense... Currently. Maybe in a decade or so with better technology.
Although the prospect of polluting the stars itself with a bunch of computers generating AI slips... We paved Paradise to put up a parking lot