Economics of Orbital vs. Terrestrial Data Centers
127 comments
·December 15, 2025jofer
What really worries me is that I keep hearing "cooling is cheap and easy in space!" in a lot of these conversations, and it couldn't be farther from the truth. Cooling is _really_ hard and can't use efficient (i.e. advection-based air or water cooling) approaches and are limited to dramatically less efficient radiative cooling. It doesn't matter that space is cold because cooling is damned hard in a vacuum.
The article makes this point, but it's relatively far in and I felt it was worth making again.
With that said, my employer now appears to be in this business, so I guess if there's money there, we can build the satellites. (Note: opinions my own) I just don't see how it makes sense from a practical technical perspective.
Space is a much harder place to run datacenters.
yabones
Yeah, I don't see a way to get around the fact that space is a fabulous insulator. That's precisely how expensive insulated drink containers work so well.
If it was just about cooling and power availability, you'd think people would be running giant solar+compute barges in international waters, but nobody is doing that. Even the "seasteading" guys from last decade.
These proposals, if serious, are just to avoid planning permission and land ownership difficulties. If unserious, it's simply to get attention. And we're talking about it, aren't we?
pavon
Cooling isn't anymore difficult than power generation. For example, on the ISS solar panels generate up to 75 W/m², while the EATCS radiators can dissipate about 150 W/m².
Solar panels have improved more than cooling technology since ISS was deployed, but the two are still on the same order of magnitude.
fanf2
This article assumes that no extra mass is needed for cooling, i.e. that cooling is free. The list of model assumptions includes:
• No additional mass for liquid cooling loop infrastructure; likely needed but not included
• Thermal: only solar array area used as radiator; no dedicated radiator mass assumed
wmf
None of it is easy but neither is cooling impossible as many people are saying.
cmgbhm
Doing like an 8xh200 server (https://docs.nvidia.com/dgx/dgxh100-user-guide/introduction-...) is 10.2kW.
Let’s say you need 50m^2 solar panels to run it, then just a ton of surface area to dissipate. I’d love to be proven wrong but space data centers just seem like large 2d impact targets.
wmf
Yeah, you need 50m^2 of solar panels and 50m^2 of radiators. I don't see why one is that much more difficult than the other.
renewiltord
For some decades now I’ve heard the debunk many times more than the bunk. The real urban myth appears to be any appreciable fraction of people believe the myth.
terminalshort
But space isn't actually cold, or at least not space near Earth. It's about 10 C. And that's only about a 10 C less than room temperature, so a human habitable structure in near earth space won't radiate very much heat. But heat radiated is O(Tobject^4 - Tbackground^4), and a computer can operate up to around 90C (I think) so that is actually a very big difference here. Back of the envelope, a data center at 90C will radiate about 10x the heat that a space station at 20C will. With the massive caveat that I don't know what the constant is here, it could actually be easy to keep a datacenter cool even though it is hard to keep a space station cool.
modeless
The temperature that you raise to the fourth power is not Celsius, it's Kelvin. Otherwise things at -200 C would radiate more heat than things at 100 C. Also the temperature of space is ~3 K (cosmic microwave background), not 10 C.
zozbot234
Will these space-based data centers run on rad-hard silicon (which is dog slow compared to anything on Earth) or just silently accept wrong results, hardware lockups and permanent failure due to the harsh space environment? Will they cool that hardware with special über-expensive high-temperature Peltiers that heat the radiators up to visible incandescence so that the heat can be shed with any efficiency? There's zillions of those issues. The whole idea is just bonkers.
kragen
At Satellogic, we famously flew mostly just regular cellphone parts on orbit. We did have higher rates of various kinds of failures than is usual on Earth, but hardware failure can generally be masked by software redundancy.
wmf
rad-hard silicon ... or just silently accept wrong results, hardware lockups and permanent failure
Somehow I don't think those are the only options. AFAIK Starlink is using a lot of non-rad-hard silicon already.
enderfusion
The radiation effects on the silicon solar cells is often underestimated, it's not just the GPUs!
danpalmer
Starlink is however operating at ~500km where radiation is less of a concern, but where the lifetime of a satellite is only 2-3 years.
The unit economics of orbital GPUs suggest that we'll need to run them for much longer than that. This is actually one of the few good points of orbital data centers, normally older hardware is cycled out because it's not economic to run anymore due to power efficiency improvements, but if your power is "free" and you've already got sufficient solar power onboard for the compute, you can just keep running old compute as long as you can keep the satellite up there.
wmf
I think they last 2-3 years after they run out of argon fuel, so more like 7-8 years total. It looks like some Starlinks from Nov 2019 are still operational.
johnsmith1840
My understanding is non rad hardened method get around this by basically doubling or some multiple of repeating calculations and chexking data often.
Random errors will occur you just need to be checking fast enough to fix and update that bad bit flip.
I am sure there's all sorts of fun algorithms in this space but I am under the impression there is SOME tax to doing this. What is the tax? Is it 10% ir 60% I have no idea would love to know!
marcosdumay
Why make a GW datacenter on the ground if you can make two and pay to launch them into space?
danpalmer
There's more than that, it's possible to get permanent hardware damage from radiation at smaller (modern standard) process sizes.
notahacker
Your other options of fault tolerance typically achieved by doing everything at least twice and being willing to reboot (and accepting attrition from total ionizing radiation) or lots of shielding are fine for building functioning space hardware but suboptimal for building datacentre business models...
tekno45
they throw those satellites to a fiery doom on a regular cadence.
Fomite
The LLMs they hope to have in those data centers already silently accept wrong results.
turtletontine
> Will…
I think “won’t”. I could be wrong of course, but I imagine efforts to put servers into orbit will die before anything is launched. It’s just a bad idea. Maybe a few grifters will make bank taking suckers’ money before it becomes common knowledge that this is stupid, but I will be genuinely surprised if real servers with GPUs are launched.
I don’t mean to be facetious here. But saying “will” is treating it as inevitable that this will happen, which is how the grifters win.
JumpCrisscross
At this scale could you do shielding?
inejge
There are ways in which shielding in space can do harm: really energetic particles get trapped and produce a shower of daughter particles and rays over a greater area. So you'd need even more shielding. Or you accept that such things will happen and use rad-hard parts, redundancy etc. When you have the whole atmosphere above, it's much less of a concern.
Besides, that's even more mass to be lofted. Pushing the economics further into the ludicrous end.
mjhay
Orbital data centers are impractical for a lot of reasons (to put it mildly) but radiation shielding isn’t one of them. Proportionally less shielding is needed as one scales up, due to lower surface/volume ratios.
ted_dunning
Sure. At the cost of lofting that shielding from the ground and taking the economics from 500x to 2000x crazy.
boddu
Really interesting article. Curious to see the ChatGPT link attached
But when I click on it, I get this error.
Failed to load shared conversation. Request is not allowed. Please try again later. (403, 9aebe525df75165e-BLR)
Ataraxic
Scott Manley had a video about this last year.
https://www.youtube.com/watch?v=d-YcVLq98Ew
The short of it is that cooling is likely the biggest problem, given you will need to pump the heat to the backside and radiate it away, and the amount of mass you will need to dedicate to cooling works against deployments and increases the cost per unit significantly. Not to mention, the idea of these huge deployments runs into potential space debris issues.
Whenever one of these ventures actually manages to launch a proof of concept, I think we'll be able to quickly discern if there is actually a near-future here.
daemonologist
> This is all to say that the current discourse is increasingly bothering me due to the lack of rigor; people are using back-of-the-envelope math, doing a terrible job of it, and only confirming whatever conclusion they already want. Calculating radiation and the cost of goods is not difficult. Run the numbers.
> References: Gemini, Gemini, ChatGPT, ChatGPT, Gemini, ChatGPT, Gemini, ChatGPT, Grok, Gemini (There are sub-references from these services in the GitHub.)
I think, if you're going to make statements like this - especially from a position of expertise, you should be personally verifying the numbers and citing their sources directly. What good is asking the reader to trust an AI on your behalf? They should trust you.
(To be clear, I suspect the conclusions drawn are still correct.)
sigmar
>This isn't about talent. It's about integration... Vertical integration isn't a nice-to-have. It's the whole ballgame.
I'm going to assume there's tons of logical errors and oversights in the math, considering the author couldn't even be bothered to write the text of the post himself.
modeless
I love the sliders, but note that the numbers on this site literally came from ChatGPT, so there is plenty of room for disagreement.
Seems like according to this analysis it all hinges on launch cost and satellite cost. This site's default for Starship launch cost is $500/kg, but SpaceX is targeting much lower than that, more like $100/kg and eventually optimistically $10/kg (the slider doesn't even go that low). At $100/kg (and assuming all the other assumptions made on the site hold) then you break even on cost vs. terrestrial if you can make the satellites for $7/watt (excluding GPUs, as the whole analysis does).
bediger4000
Aerospace industry has a long history of missing lower cost/kg to orbit. I'm extremely suspicious of $500/kg, which is about a third of today's cost.
modeless
OTOH SpaceX has a pretty good history of undercutting the industry on cost. If Starship full reusability works I would be very surprised if it only lowered launch costs by a factor of three. Of course it's not guaranteed to work, but clearly SpaceX's orbital datacenter plans are predicated on Starship working.
ted_dunning
SpaceX has never met any milestone that Elon has ever set.
enderfusion
OP here. Happy to answer any questions, appreciate the boost.
xnx
François Chollet https://x.com/fchollet/status/1999982683708150014 :
"Datacenters in space" make for a catchy narrative and an interesting demo, but the math simply doesn't work.
When considering factors like launch cost, maintenance complexity, and the cost of high-bandwidth communications (latency included), there is no realistic set of economic and engineering assumptions under which orbiting datacenters become cost-competitive with simply building conventional nuclear-powered (or renewable energy-powered) datacenters on the ground.
In fact we're off by 50-100x. Dramatic launch cost reductions still won't make it work. And of course if you invest a lot in specific lines of tech to make it work you then have to consider that the same can also be invested in better ground-based nuclear, bringing the cost of power down for everyone.
wmf
So the guy who shows his work says 3.4x and the guy who doesn't says 50-100x.
enderfusion
OP here.
man who quotes big number - often made a fool
octaane
The only benefit as I perceive it re: orbital data center hardware is regulatory avoidance. Think...DDOS machines that can't be shut off; or financial hosting services for unsavory individuals. However, it's very expensive by all metrics (including those talked about in the article), and frankly, these satellites are sitting ducks for the hunter killer satellites the various space powers have, if they actually wanted to do something about these hypothtical data centers and the problems they would cause.
null
sc68cal
When Starcloud put together that whitepaper the first thing I looked at was the launch costs[1]. It references a $5M cost to launch, which right away made absolutely no sense to me. Just a cursory search shows launch costs are around $50M per launch, if not more.
It's great that this site drills down even further to demonstrate that there is absolutely no point at which the launch costs ever make this economical or viable, so I really don't understand what people are doing.
Especially because this site was harping for years about the cost of launches and putting things in to orbit, the whole reason why SpaceX got started and has grown as it has. As soon as that became an inconvenient number, we now just make things up (Just pretend that launch costs are 10% of what they actually are to get people to invest?).
JumpCrisscross
> Just a cursory search shows launch costs are around $50M per launch
I think datacentres in space are predicated on Starship bringing launch costs down. Way down.
indolering
Spinlaunch is also promising drastically reduced cost per launch. The payload size for their first launcher is pretty small and they appear to be struggling to get the kinetic launcher online.
brotchie
Did a similar back-of-the-napkin and got 5x $ / MW of orbital vs. terrestrial. This article's analysis is ~3.4x.
I do wonder, at what factor of orbital to terrestrial cost factor it becomes worthwhile.
The greater the terrestrial lead time, red tape, permitting, regulations on Earth, the higher the orbital-to-terrestrial factor that's acceptable.
A lights-out automated production line pumping out GPU satellites into a daily Starship launch feels "cleaner" from an end-to-end automation perspective vs years long land acquisition, planning and environment approvals, construction.
More expensive, for sure, but feels way more copy-paste the factory, "linearly scalable" than physical construction.
notahacker
It becomes worthwhile if its actually cheaper (probably significantly cheaper given R&D and risk), or if you're processing data which originates in space and the data transfer or latency is an issue
You can set up plant manufacturing chips in shipping containers and sending them to wherever energy/land is cheapest and regulation most suitable, without having to seek the FCCs approval to get launch approved and your data back...
This is an interesting analysis, and I like the sliders that let you instantly show the impacts of system trades.
The one glaring hole that I see is the challenge of moving the data to/from the datacenter while it's on orbit. Bandwidth to/from space isn't free. FCC/ITU licenses are required, transmitters/receiviers/modems/DSP/antennas all add to SWAP (size, weight, and power). Ground-stations are needed to move the data up/down, but those have recently become a commodity too. Still, they're not free. (see: https://aws.amazon.com/ground-station)
There is also the added latency between earth-based users and space-based datacenters, which may be a deal breaker for some applications.
Another issue I don't see covered are the significant differences between space-based hardware and terrestrial hardware. The space stuff needs to be radiation tolerant, and that usually makes it a lot slower and a lot more expensive than the terrestrial stuff, all other things being equal.
In the end, space-based datacenters are highly impractical even if you assume that Starship can put anything into orbit very cheaply.