4 ms·

I was initially very skeptical about the viability of space-based data centers but after a couple hours reading papers, studies and summary technical assessment

I was initially very skeptical about the viability of space-based data centers but after a couple hours reading papers, studies and summary technical assessments I realized there are a range of credible expert viewpoints from, "pretty unlikely" to "it could actually work". There at least appear to be plausible, though unproven, solutions to the most obvious drive-by objections I had off the top of my non-expert head. Of course, there are still a lot of unknowns, any of which could prove fatal to the concept but I'm no longer comfortable just dismissing it as "obviously ridiculous."

Teever 5mo ago
Putting a datacentre in space may be feasible but the scale that he's suggesting is really unbelievable. And if he's actually capable of producing solar panels in the quantity that he's talking about in the time frame that he's talking about -- why doesn't he just put them on earth to solve our growing climate change problems and fuel shortages?
- mrandish 5mo ago
  > but the scale that he's suggesting... Well, yeah but that's just Elon being Elon. At this point I think even the most pro-Elon folks freely admit "The first rule of Elon is: 'Ignore everything he says about timeframes and scale.'"
- johnsimer 5mo ago
  Permitting/regulation issues
foxylad 5mo ago
Did you find a credible solution for heat dissipation in the papers you read? I fear the laws of thermodynamics will kill this project.
- hvb2 5mo ago
  https://youtu.be/FlQYU3m1e80 https://youtu.be/FlQYU3m1e80 Is a good start
luke5441 5mo ago
I think what often gets confused is people saying it isn't viable with "it is not physically possible". It is physically possible, but it won't have positive ROI so it is not viable. If you have a paper/post doing the calculations for positive ROI, I'd be all ears. It can even have the optimistic Elon assumptions about price of mass to orbit.
- geertj 5mo ago
  I don’t think people are confused between the two. What’s happening is that drive by objections have no real way to assess viability. That calculation you ask for has line items in it that greatly depend on engineering optimizations that, unless you work in the field, are hard to estimate accurately.
  - luke5441 5mo ago
    On those line items make the most optimistic assumption. E.g. use Elons Starship cost to orbit. Assume useful GPU lifetime of 6+ years. Assume max physical possible radiators etc.
  - IshKebab 5mo ago
    I don't think you really need those calculations to know that it is always going to be less viable than a traditional data centre. It's like the idea about putting data centres under water. I'm sure you can do it and you can come up with estimates over how much it will cost... but it's always going to cost more than not putting your data centre underwater.
saberience 5mo ago
Yeah you would need 10-20 American football fields worth of radiators for a single data-center... so yes, it "can" work, but it's completely inefficient and unrealistic.
- geertj 5mo ago
  The plan is to launch a constellation of smaller AI sats not a monolithic large data center. The calculations I have seen actually have a smaller radiator area than solar panels. Scott Manley’s has a video on this where he goes into some numbers.
  - saberience 5mo ago
    Scott Manley's video uses 20kw as a reference number which isn't even half of the power usage of a modern GB200 rack. I.e. not even close to the power usage of an actual datacenter. In fact, not even 1% of the power usage of a datacenter... Also, how is a constellation of satellites any easier in this case? They all need extremely large radiators, they all need maintenance, they all need high bandwidth communication. If you calculate the actual cooling requirements for megawatts of server, you end up with needing many, many football fields of cooling. It's nonsensical. Sure you can make the numbers sort of work for a single server, but a single server on earth costs MUCH MUCH less to launch, maintain, etc. So why bother doing it in space? We just end up with loads of unusable space servers as they gradually breakdown and cannot be repaired.
    - amoss 5mo ago
      Later in the video he runs through the changes needed for 100kw per rack.
      - saberience 5mo ago
        Wow, one 100kw rack! Dude, you realize that right now there are 100+s of data centers in construction around the world often in the 500MW to 1PW range? I.e. there are many, many datacenters (100s) in construction, right now, with 100s of MW up to multiple PW? Scott's analysis is out by several orders of magnitude! Everyone knows its "theoretically" possible to have a single server or a single rack in space. The big and most obvious glaring miss is, how would it be economically viable and operationally viable to have datacenters in space which compete with datacenters on the ground. We are experts are creating economically profitable (very profitable) datacenters on the ground, which work really well for inference and training, which are cooled really well, can be maintained easily, etc. The idea that we are going to have 100s of MW clusters or 1PW clusters in space, and they are going to be competitive economically, and we are going to be able to maintain them, and they are going to actually WORK (i.e. how will the networking be competitive with datacenters on earth), is frankly laughable. Scott is totally talking rubbish on this.
xnx 5mo ago
There are a lot of things that are possible that make no financial sense.
saalweachter 5mo ago
If you spot Datacenters In Space as making sense, why does SpaceX acquiring XAi make sense to do it? Why not have SpaceX build datacenter satellites and lease them to XAi et al, or XAi design and manufacture satellite datacenters and pay SpaceX to launch them? Heck, why not start a third company which focuses exclusively on datacenters in space, buying services from one company and selling them to another, so they can concentrate on this rather specialized skillset without distraction? How does SpaceX acquiring XAi help anything but Elon Musk's personal portfolio?
- mrandish 5mo ago
  > why does SpaceX acquiring XAi make sense to do it? I never said it did. My post didn't even mention SpaceX, Elon or XAI. I just shared that my initial drive-by opinion on space-based compute, as a general concept (regardless which company does it, when or how), was reflexively dismissive but after a tiny bit of research, it's now risen to the lofty epistemic level of "not completely ridiculous." And I discovered it's a much more interesting and complex question than I assumed, involving highly uncertain estimates of rapidly evolving technology, physics and economics. For me the question starts more akin to "could a Dyson sphere ever be viable for human civilization in this solar system" than "should grandma invest her life savings in SpaceX this afternoon." But it seems some can't reason about the potential future viability of space-based compute independent of current politics, personalities, culture war or whether AI is a bubble inflated by circular deals. To be clear, I do think AI valuations are wildly over-inflated and propped up by financial engineering bordering on fraud likely to trigger a huge economic crash. But it can be simultaneously true that there will be hugely profitable AI businesses in 2035 and it's possible space-based compute might be viable for some company in 2040. As for Elon, ever since I met him a couple times closer to the Paypal days, I've felt he's an eccentric, impulsive nut but that he's also a savvy, technically-minded entrepreneur with a sharp eye for opportunity who's compelled by obsessive, visionary zeal. And everything he's done since has been consistent with that assessment, from the successes of Tesla and SpaceX to his bizarre, self-destructive detours into politics, social media and AI. I don't understand why so many can't deal with the reality that a person can be deeply flawed and wildly irrational in some ways while simultaneously being highly effective and immensely valuable in other ways. Steve Jobs was a brilliant visionary entrepreneur yet also an asshole in his personal life who believed some crazy shit so deeply it contributed to his early death.

Teever 5mo ago

Putting a datacentre in space may be feasible but the scale that he's suggesting is really unbelievable. And if he's actually capable of producing solar panels in the quantity that he's talking about in the time frame that he's talking about -- why doesn't he just put them on earth to solve our growing climate change problems and fuel shortages?

mrandish 5mo ago
> but the scale that he's suggesting... Well, yeah but that's just Elon being Elon. At this point I think even the most pro-Elon folks freely admit "The first rule of Elon is: 'Ignore everything he says about timeframes and scale.'"
johnsimer 5mo ago
Permitting/regulation issues

foxylad 5mo ago

Did you find a credible solution for heat dissipation in the papers you read? I fear the laws of thermodynamics will kill this project.

hvb2 5mo ago
https://youtu.be/FlQYU3m1e80 https://youtu.be/FlQYU3m1e80 Is a good start

luke5441 5mo ago

I think what often gets confused is people saying it isn't viable with "it is not physically possible". It is physically possible, but it won't have positive ROI so it is not viable. If you have a paper/post doing the calculations for positive ROI, I'd be all ears. It can even have the optimistic Elon assumptions about price of mass to orbit.

geertj 5mo ago
I don’t think people are confused between the two. What’s happening is that drive by objections have no real way to assess viability. That calculation you ask for has line items in it that greatly depend on engineering optimizations that, unless you work in the field, are hard to estimate accurately.
- luke5441 5mo ago
  On those line items make the most optimistic assumption. E.g. use Elons Starship cost to orbit. Assume useful GPU lifetime of 6+ years. Assume max physical possible radiators etc.
- IshKebab 5mo ago
  I don't think you really need those calculations to know that it is always going to be less viable than a traditional data centre. It's like the idea about putting data centres under water. I'm sure you can do it and you can come up with estimates over how much it will cost... but it's always going to cost more than not putting your data centre underwater.

saberience 5mo ago

Yeah you would need 10-20 American football fields worth of radiators for a single data-center... so yes, it "can" work, but it's completely inefficient and unrealistic.

geertj 5mo ago
The plan is to launch a constellation of smaller AI sats not a monolithic large data center. The calculations I have seen actually have a smaller radiator area than solar panels. Scott Manley’s has a video on this where he goes into some numbers.
- saberience 5mo ago
  Scott Manley's video uses 20kw as a reference number which isn't even half of the power usage of a modern GB200 rack. I.e. not even close to the power usage of an actual datacenter. In fact, not even 1% of the power usage of a datacenter... Also, how is a constellation of satellites any easier in this case? They all need extremely large radiators, they all need maintenance, they all need high bandwidth communication. If you calculate the actual cooling requirements for megawatts of server, you end up with needing many, many football fields of cooling. It's nonsensical. Sure you can make the numbers sort of work for a single server, but a single server on earth costs MUCH MUCH less to launch, maintain, etc. So why bother doing it in space? We just end up with loads of unusable space servers as they gradually breakdown and cannot be repaired.
  - amoss 5mo ago
    Later in the video he runs through the changes needed for 100kw per rack.
    - saberience 5mo ago
      Wow, one 100kw rack! Dude, you realize that right now there are 100+s of data centers in construction around the world often in the 500MW to 1PW range? I.e. there are many, many datacenters (100s) in construction, right now, with 100s of MW up to multiple PW? Scott's analysis is out by several orders of magnitude! Everyone knows its "theoretically" possible to have a single server or a single rack in space. The big and most obvious glaring miss is, how would it be economically viable and operationally viable to have datacenters in space which compete with datacenters on the ground. We are experts are creating economically profitable (very profitable) datacenters on the ground, which work really well for inference and training, which are cooled really well, can be maintained easily, etc. The idea that we are going to have 100s of MW clusters or 1PW clusters in space, and they are going to be competitive economically, and we are going to be able to maintain them, and they are going to actually WORK (i.e. how will the networking be competitive with datacenters on earth), is frankly laughable. Scott is totally talking rubbish on this.

xnx 5mo ago

There are a lot of things that are possible that make no financial sense.

saalweachter 5mo ago

If you spot Datacenters In Space as making sense, why does SpaceX acquiring XAi make sense to do it? Why not have SpaceX build datacenter satellites and lease them to XAi et al, or XAi design and manufacture satellite datacenters and pay SpaceX to launch them? Heck, why not start a third company which focuses exclusively on datacenters in space, buying services from one company and selling them to another, so they can concentrate on this rather specialized skillset without distraction? How does SpaceX acquiring XAi help anything but Elon Musk's personal portfolio?

mrandish 5mo ago
> why does SpaceX acquiring XAi make sense to do it? I never said it did. My post didn't even mention SpaceX, Elon or XAI. I just shared that my initial drive-by opinion on space-based compute, as a general concept (regardless which company does it, when or how), was reflexively dismissive but after a tiny bit of research, it's now risen to the lofty epistemic level of "not completely ridiculous." And I discovered it's a much more interesting and complex question than I assumed, involving highly uncertain estimates of rapidly evolving technology, physics and economics. For me the question starts more akin to "could a Dyson sphere ever be viable for human civilization in this solar system" than "should grandma invest her life savings in SpaceX this afternoon." But it seems some can't reason about the potential future viability of space-based compute independent of current politics, personalities, culture war or whether AI is a bubble inflated by circular deals. To be clear, I do think AI valuations are wildly over-inflated and propped up by financial engineering bordering on fraud likely to trigger a huge economic crash. But it can be simultaneously true that there will be hugely profitable AI businesses in 2035 and it's possible space-based compute might be viable for some company in 2040. As for Elon, ever since I met him a couple times closer to the Paypal days, I've felt he's an eccentric, impulsive nut but that he's also a savvy, technically-minded entrepreneur with a sharp eye for opportunity who's compelled by obsessive, visionary zeal. And everything he's done since has been consistent with that assessment, from the successes of Tesla and SpaceX to his bizarre, self-destructive detours into politics, social media and AI. I don't understand why so many can't deal with the reality that a person can be deeply flawed and wildly irrational in some ways while simultaneously being highly effective and immensely valuable in other ways. Steve Jobs was a brilliant visionary entrepreneur yet also an asshole in his personal life who believed some crazy shit so deeply it contributed to his early death.