For real. It's an enormous problem solved only with 1) sheer scale, and 2) Science Fiction.<p>Both of those are expensive as hell, by the way.<p>Cooling via radiation follows Stefan–Boltzmann: P = εσAT⁴. Let's assume a good surface (emissivity ~0.9) at 300 K (27 °C) at 400 W per square meter per side. A flat panel radiating from both faces into deep space gets 800 W/m, not including the losses from, say, the Sun, or from IR coming off the Earth. Now, input power. Sunlight in orbit ~1,360 W/m², assume ~22% cell efficiency, we got 300 W/m². So each 1 MW compute, 3,300 m² of solar panel and minimum 1,200–1,500 m² of radiator.<p>In case ya didn't know - 1 MW is <i>tiny</i> from a present-day-datacenter perspective. It's like 8 racks. So we're talking orbital megastructures here, many many many square kilometers, and this is with all the best case assumptions, and magic radiator panels that never see the sun, or the earth, or the moon.<p>This is just the basic numbers here, by the way. There's a garbage truck full of other unsolvable problems if you poke your head in there.<p>Aside from the "Avoid Regulations" aspect, and the "Everything That Burns Deorbiting is Depreciation" aka "The Starlink Trick", I'm not sure what the hell the draw is.
Cooling in space is hard but not impossible - while current (IMHO stupid without advanced in space infra) space data center projects work with a couple MW, many advanced space propulsion concepts might have to reject hundreds of MW if not a couple GW.<p>For that you might need more advanced stuff like liquid droplet radiators (<a href="https://en.wikipedia.org/wiki/Liquid_droplet_radiator" rel="nofollow">https://en.wikipedia.org/wiki/Liquid_droplet_radiator</a>), heat sinks & pulsed operation. Still, it should be eventually doable.<p>As for space data centers - I think the main issue is the complete lack of in space infrastructure for resource mining, processing and manufacturing & maintenance. It is kinda like building your first practical steam locomotive & the deciding to build directly an airliner. No suitable materils, experience, work force, material sources, etc.<p>We eventually went from locomotives to airliner, in an incremental manner & expanding the supporting infrastructure to support the ever more ambitious projects.
It might be eventually doable, as an experiment or as a flex, sure. But it's never going to come close to being cost-reasonable versus the equivalent infrastructure here on earth.
The security advantages are enormous since access to space is so tightly restricted and controlled, compared to the access potential of a land based data center. Only risk in space is maybe we start WWIII with china and the US directly trading blows. On the ground, any insurgent group can disable your infrastructure. Nothing is truly safe on the surface of the earth. Anyone can strap a bomb on a drone now. See examples from the currently active wars.
According to Wikipedia this reduces weight and not the required area. Also AI said the same thing, but I can't trust in it this blindly. So, how smaller would be the required surface area?
> Now, input power. Sunlight in orbit ~1,360 W/m², assume ~22% cell efficiency, we got 300 W/m². So each 1 MW compute, 3,300 m² of solar panel and minimum 1,200–1,500 m² of radiator.<p>We need > 2x more solar panels than we need radiators. Doesn't this imply radiation isn't really the limiting factor here?
Glad you are on the case before these companies foolishly waste their money sending datacenters to space.
Is it possible for one side of panel to be used for solar power and other side for radiating heat?
Getting this all up into orbit it obviously the hard part, but if you're already building so much solar capacity the cooling actually doesn't seem unreasonable?
Radiating 1MW at 500K (227C) with a 0.4MW heat pump takes about 200 m^2 flat sheet surface. Inputs - solar+nuclear for double fun. So - quite feasible.
Moves 1 MW of heat with 0.4 MW of work? I.e. 2.5 COP {coefficient of performance). That's insane, and I mean that in a good way. Could you dig me up a cite for that?<p>That's thumping the Carnot limit: [[T_cold / (T_hot − T_cold)]].<p>2.5, while rejecting at 500 K, cold side's at least 357 K (eeehhhhhhh 84 °C) . . . and that's an <i>absolutely perfect</i> Carnot machine. At 50% Carnot -- a pretty good heat pump, real world performance is 40-60 -- cold side's at 417 K (144 °C). 417k, feeding your GPU coolant loops.
And if you look at SpaceX's Starmind sats, they will have a 160 m^2 liquid radiator for 175kw/250kw peak compute.
Do we even _have_ semiconductors that can work at 220C? And if you're thinking about using some kind of refrigeration cycle, its efficiency is going to be bad.
1) The chips don't reach 220C. The 220C is the temperature at the hot end of the heat pump. The chips are on the cold end of the heat pump.<p>2) The International Space Station has used a dual-loop ammonia/water-based heat pump to cool the station temperatures. It's been in place for several decades. Heat pumps are a proven technology.<p>Other satellites have also used heat pumps, such as SES-17 in geostationary orbit <a href="https://www.esa.int/Applications/Connectivity_and_Secure_Communications/Very_High_Throughput_Satellite_ready_to_pump_heat" rel="nofollow">https://www.esa.int/Applications/Connectivity_and_Secure_Com...</a>
Wouldn't this be solved like similar problems on earth by making small structures with large surface areas?
I don't think so. Large surface area helps with convective cooling I think by increasing the surface area that participates in heat exchange with the air (or other thermally conducting material), radiative cooling wouldn't benefit from this because you can't concentrate light beyond the source that it's emitted from (etendue).<p>Though I do wonder if it would be possible to have some kind of internal heat pump driven by electrical power to juice up the temperature of the radiators to increase the power being radiated away? E.g., run a heat pump to increase the temperature of a working fluid and then run high temperature radiators? I think it would work and I don't immediately see that it would violate the laws of thermodynamics? (this is ignoring all practically, I'm sure the engineering would be devilishly hard, although if you're already shooting for the moon you might as well throw in some artificial gravity to boot, it's not like the robots get motion sickness)
Don’t you get 4 faces to radiate away, assuming a long rectangular tube.
Great read, thanks for sharing. I am interested in reading some more about the other unsolvable problems that exist in this space, do you have any recommendations that you wouldn't mind pointing me at? It would be greatly appreciated, and thank you :)
It would be a shame if a rock came out of nowhere and hit that many square kilometers stucture<p>Luckily, there are almost no rock in space.
there are two arguments for it.<p>one is marketing.<p>the other is that you could make tiny datacenters and flood the sky with them. in effect, not datacenters at all, but some kind of dataswarm coordinating at literal lightspeed via lasers.<p>they'd still be wildly expensive to deploy, and probably litter the orbit zone with fast-moving debris.
Your "other" makes no sense. It doesn't matter if you make a few big or a lot smaller, in space you will still need the same space for the same amount of megawatt. Or did you miss the scale of parent's post ? Because in that dream scenario of "let's ignore all the issues except that" and "the earth and the sun don't have any impact", it's still 3 THOUSANDS square meters for a MW of 8 racks.<p>You want to go smaller and go one rack only sure, it's still hundreds of square meters. Check the size of current orbital structure for a point of reference, you can't dwarf those and call it a "dataswarm of tiny datacenters flooding the sky".
From the article itself, sounds like it’s an open problem that they are experimenting with:<p>“ We’re working on a number of different approaches for this, including a combination of heat pipes and radiators to cool the chips. So far, our team has tested the technology in a thermal vacuum chamber that simulates both the thermal and vacuum environment in space. We’ll see how our new TPU cooling system works in space and refine our designs as we learn more.”
Can't they harness the heat to generate more electricity?
The article mentions that:<p>> The biggest challenge was how to cool the A.I. chips, which heat up when they perform calculations and process information. Fans, which typically help dissipate the heat, do not work in space. So the Silicon Valley company instead developed a cooling system that uses layers of conductive material to expel the heat into space.<p>> The bottom layer is made up of Google’s A.I. chips, which sit on a green motherboard. The next layer consists of “thermal interface material,” a pale green putty that comes in sheets like Fruit Roll-Ups and connects the chips to layers of aluminum and copper, radiating heat away from the motherboard. Finally, there is a radiator panel, which projects heat into space.<p>> The chips can operate for about 15 minutes in space before needing to be shut down so they can cool off, said Travis Beals, Google’s senior director of product management for Project Suncatcher.
No solution, but that is the crux of the problem. They probably need to make a radiator that 1000x smaller and lighter.<p>TPU: 100,000+ watts/square-meter<p>Radiator: ~300 watts/square-meter<p><a href="https://youtu.be/ktdbUIZKeSE?t=76" rel="nofollow">https://youtu.be/ktdbUIZKeSE?t=76</a>
I watched a video with Elon musk the other day and he was very confident saying it's already a solved problem and they already do it with Starlink to some degree. He was baffled that this debate keeps coming up.
He is free to launch his own space GPU if he is so confident it is profitable*.<p>I will take it as a given that the Google engineers know what they are doing, and their first version can only run for 15 minutes before it needs to shut down to cool.<p>*Profitability requiring that other Musk controlled companies do not pay for the service at elevated rates in a classic self dealing scheme.
> He is free to launch his own space GPU if he is so confident it is profitable*.<p>Are you not aware that's exactly what SpaceX is doing?? <a href="https://www.spacex.com/spacexai/starmind" rel="nofollow">https://www.spacex.com/spacexai/starmind</a> Sure, it's now SpaceXAI or whatever, but how is that any different than Google sending up sats and selling the compute via Gemini?
That guy is baffled we're not all driving around in cybertrucks talking to mecha-hitler. Why would i care?
That's the neat part: you don't (ergo, this is yet another marketing crap)
We don't. This is all cover for the militarisation of space, there's no real benefit that'd be ever economical to put a DC up in space when you could build one on the ground. The whole narrative exists to allow google to tap into the Golden Dome / Space force bucket of pork that's basically SDI II.<p>You don't need civilian scale compute in space, but you absolutely can put miliary application up there and get filthy rich, like Elon, who doesn't care if it ever works, because he's getting paid as taxi service to shoot stuff up.
This is the most interesting perspective I've heard on this topic, which otherwise always converges on the same political dismissals or heat dissipation arguments (the latter are fascinating, but going in circles by now).<p>Sounds very probable: gives a plausible reasons for sending a lot of infra up, absolutely doesn't have to be profitable or even effective, puts the focus on AI instead of the MIC, and is in line with the kind of contracts Big Tech runs after.<p>To me this model explains a lot of why so many big companies seem to be investing into what every expert I've heard says goes against basic physics. The only thing missing for it to be more than an interesting idea is why China and the EU are not fighting this.
Please stop with this argument. This is one of the easiest problems to solve. Heat dissipation requires a radiator, which is just a dumb hunk of metal. All you need is a cheap way to launch mass to orbit, which is exactly what SpaceX (and Chinese companies) are doing.<p>The only valid argument against data centers in space is the economic cost. If the per megatoken price for space datacenters is higher than terrestrial, then this won't work.<p>But the price for terrestrial datacenters keeps going up and technology keeps dropping the price of space-based.
That is not a fake argument, but real physics. Yes, you can design out X kw can be dissipated by this much radiators, but that adds an enormous quantity of mass, more than the solar panels that feed it.<p>If you are trying to generate a profit, every extra kg puts you further in the red. It still costs over $1000/kg of mass to get into orbit.
So is it a physics problem or an economic problem? Sounds to me like you are acknowledging that it is just an economic problem. If it cost $1/kg to get to orbit then this wouldn't be a problem, right? Just make a bigger radiator.<p>But if it's an economic argument, then you need to do the actual math. How big is the radiator? How low could the price to orbit go? How much can you charge per million tokens in 2030?<p>The reason this is a fake argument is because the validity depends on the math, and nobody advancing the "you can't cool stuff in space" argument is actually doing the math.
The economics is constrained by physics. AI in space is not viable if it costs $1000/kg to launch to space. Starship promises to cut that down to $100/kg (more if you believe Elon, but most don't), but that's still not competitive.<p>Sure, things would be different if the cost was $1/kg, but short of somebody building a space elevator that's just not going to happen.
I won't believe less than $100/kg until I see it. I agree with you on that.<p>But are you sure that $100/kg is not competitive? I don't remember all the math, but even their initial AI1 design[1] would throw a lot of profit, if you can sell at the price they offered to Anthropic.<p>I suspect the price of manufacturing the satellite, plus chips, is the dominating factor, not necessarily the launch costs.<p>[1] <a href="https://www.spacex.com/spacexai/starmind" rel="nofollow">https://www.spacex.com/spacexai/starmind</a>
It's a fundamental physics problem. You need to have huge radiating surfaces.<p>A 1GW datacenter with chips operating at 100C (which is probably doable) will need a radiator that has a surface of one square _kilometer_, and this is with all the favorable assumptions. Realistically you'll need about 2x of that.<p>If you want your DCs to be on a 1000km orbit (for reasonable ping times), you'll be able to _resolve_ these satellites with a naked eye!<p>Sorry. But this idea is fundamentally unworkable.
The current plan is for 250 kW peak with 160 m^2 radiators. It will weigh about 4 tons so you can pack 25 on one flight of Starship.<p>Deploy 4,000 and you're at 1 GW. That's 160 launches.<p>BTW: SpaceX has already manufactured and launched 10,000 Starlink satellites and Falcon 9 launches about 150 times per year. None of this seems unworkable.
Can we solve the fertilizer price problem by hauling it with airplanes? Yes, we can! It's easy, just load the potash fertilizer into an airplane and unload it directly into the traincars. I even designed a neat conveyor belt system to speed up unloading!<p>Look at the numbers. 1kW of nuclear power capacity on Earth is around $2000, and that's 24/7 guaranteed power. So a 250kW cluster needs $500000 to cover its power demand with near 100% reliability and with some ongoing cost.<p>And if we're OK with some interruptions, then we can use solar+wind at around $100000 and with essentially no ongoing cost. If we assume the absolutely best projected launch cost of $100 per kg (vs the current one of ~$800), that's just 1 ton of material in space!<p>So you're off by 2-3 orders of magnitude in cost. And this kind of "it's unprofitable" is actually a fundamental issue.
> This is one of the easiest problems to solve. /../ All you need is a cheap way to launch mass to orbit.<p>So... not easy? I don't suppose we're at a point where you could reasonably send a large-enough radiator for a multi-gigawatt cluster into space?
Not only that - it would be totally insane to launch something heavy & at the same time fragile from earth (under a lot of vibrations & heavy g-loading).<p>This is all a weird speedrun or race. If something we should be working on setting up resource mining from the Moon & asteroids, materials processing on the Lunar surface & in orbit, simple manufacturing in space, etc.<p>Instead some people think we can jump straight to a computronium Dyson swarm. :P
Now you've moved the goal posts. It's no longer "you can't cool stuff in space"; now it's "we can't launch a 10-gW compute cluster by the end of the year."<p>You don't need to send a multi-gigawatt satellite to space. You just need to launch a few thousand 250 kW satellites. That's not against the laws of physics.
> Please stop with this argument. This is one of the easiest problems to solve. Heat dissipation requires a radiator, which is just a dumb hunk of metal.<p>Convective radiation does not happen in space and this challenge is far more significant than your comment implies. Rather than "a dumb hunk of metal", radiators for spacecraft are often made of ceramics and carbon laminates with higher IR emissivity than convective radiators made of simple metals.<p>From the article you're commenting on:<p>> The chips can operate for about 15 minutes in space before needing to be shut down so they can cool off, said Travis Beals, Google’s senior director of product management for Project Suncatcher.<p>The Thermal Control section on NASA's Small Spacecraft documentation center is quite an interesting read for this subject: <a href="https://www.nasa.gov/smallsat-institute/sst-soa/thermal-control/" rel="nofollow">https://www.nasa.gov/smallsat-institute/sst-soa/thermal-cont...</a><p>ML Infrastructure comes with some pros (larger emissive footprint) and cons (exponentially larger TDP) compared to the concerns there, but if you aren't familiar with the challenges of heat dissipation in space, please give it a read.<p>There's also a pretty interesting pop-sci article on cooling the Webb telescope, since it needs to be <i>especially</i> cold for its purpose. Not <i>directly</i> related, but may give insight into both challenges and solutions as well as well. <a href="https://science.nasa.gov/mission/webb/science-overview/science-explainers/how-does-webb-stay-cold/" rel="nofollow">https://science.nasa.gov/mission/webb/science-overview/scien...</a>
Emissivity is one factor, but it is dwarfed by the T^4 term. Sure, maybe if you use exotic materials you can get from 0.9 to 0.95 emissivity, but why bother? Just run the radiators a little hotter.<p>The equation is:<p><pre><code> A ~ (1000 P) / (2 e k T^4)
</code></pre>
Where<p><pre><code> A is the radiator area in square meters
P is the power in kilowatts
e is emissivity (usually 0.9)
k is the constant 5.67e-8
</code></pre>
P and T are the dominating factors. Don't worry about emissivity.
Emissivity <i>is</i> an important factor here because as I said, and as the sources I linked for you to reference clearly stated <i>convective radiation is not taking place in space</i>.<p>Both Google and NASA are worried about this for a reason, if you think they are wrong, you should offer your assistance to them, rather than debating me.
It will never be cheaper to put compute into orbit. And costs for AI are dropping like a rock here on Earth.
and yet notice how the cooling video / section was the only one they didn't have a solution for... just saying "radiator" doesn't make sense - the radiator heats up too. it's how you get rid of the heat, not where you put it.
Huh? Radiators are known technology. They have them on ISS; they have them on every Starlink satellite. This isn't like warp drive or antigravity.<p>Moreover, you can easily calculate how big of a radiator you need for a given power level and temperature. You can use the Stefan–Boltzmann law:<p><pre><code> A ~ (1000 P) / (2 e k T^4)
</code></pre>
Where<p><pre><code> A is the radiator area in square meters
P is the power in kilowatts
e is emissivity (usually 0.9)
k is the constant 5.67e-8
</code></pre>
For a 1 kW test like Googles, you just need 1 square meter of radiators (assuming two-sided).<p>For SpaceX's 175 kW satellites, they will need ~170 square meters of radiators.<p>None of this is impossible or even difficult to calculate. That's why I think this is the laziest argument against space data centers. There are so many other more reasonable arguments (like whether they will be economically competitive) but people love to latch on to this one for some reason.
To your last paragraph: when opposing something, it makes sense to use the laziest argument first. Only if that doesn't achieve your goals, would you move towards less lazy arguments. I forget what the term for this is, it's generally criticized ("you should just put your strongest arguments first").
sure, the physics is solved: we know how radiators work, and we can calculate the area needed, blah blah. but we haven't put sustained-AI-computer systems into space yet - heat is a genuine concern, and personally I'm curious if they are developing something <i>beyond</i> simple radiators. if a ~1m^2 radiator works, great. but if the current test can only run the TPUs for ~15-minute bursts before it has to stop and dump heat, this issue isn't exactly "solved" in my book.
I don't understand your argument. Sounds like you're saying, "In theory it should work, but what if there are space pixies that keep rebooting the TPUs? What do we do then?"<p>The 15-minute figure is what they currently designed for because they have mass constraints. If they wanted a different figure (like forever) they could do it with a larger radiator.<p>As you said, the physics is solved! We know exactly how much heat a surface radiates in space--there is literally an equation for it. We know how to cool stuff in space. Will the price be low enough to make a profit? That's the real question. But stop worrying about cooling in space.
> For SpaceX's 175 kW satellites<p>That's like a quarter of a rack of modern AI hardware. Modern AI datacenters are now in the multi-GW range.
Okay, so you agree the cooling problem has a solution.<p>Now the argument is, what, you can't launch that many satellites?
Is Kessler syndrome priced into cost? Or is that just like, someone else's problem?