Data centers in space - future expansion opportunity for 1776 Solutions?

[David Brown]

I really don't see the economic appeal of building a data center in quite literally the most expensive place you could possibly put a data center except for MAYBE at the bottom of the Mariana trench.

 

[They were once men]

does this mean aliens can use the kiwi farms?

 

[Uncle Sid]

This seems like the worst possible place to store a backup. One solar flare and it's toast.

 

[General Emílio Médici]

You can immigrate too, and set up a business on Svalbard without a Visa. It also has geothermal potential.

Something to think about.

Maybe I should... Maybe I should...

 

["Controlled Demolition"]

Not gonna work due to waste heat.

Data centers are notoriously both power hogs and heat generators. The nature of current computing tech means that a lot of energy is lost to heat as the computing gets done. In space getting rid of waste heat is a fucking nightmare because you have no air or water or anything like that to be a thermal conductor and so you are forced to rely on infrared radiation. These orbital data-centers would need huge radiator panels to keep their insides cool, and if they are on Earth orbit the sun is gonna make them useless about 50% of the time.

People are better off building it in places like Iceland and Greenland with geothermal electricity and where you can just have outside air circulate to cool the thing down.

Not a problem. Do you really think the ISS just overheats when it's in direct sunlight? No. Their radiators are in line with radiation and shaded by the hull itself. Permanent waste heat removal solution. I'd need to see some actual numbers, but if the ISS can manage a crew of up to ~10 100W space heaters plus a bunch of electronics, lighting, conditioning, etc. with a couple of dinky little plate arrays, even a 1MW continuous heat load should be achievable. And, you're forgetting one extremely important part: you don't need to run computers at human-comfortable temperatures. The radiators on the International Space Station have to radiate at or around 20°C. No such case for a radiator meant to eliminate heat from a computing system, which can be rated to run at 80°C and higher. The efficiency of emitted heat rises exponentially with the emitted temperature compared to background radiation, so the higher you can take it, the better. It might need some more exotic radiator concepts to get rid of much more than 10MW, like wire-loop radiators, but it should be workable

What isn't achievable is radiation hardening and maintenance. All the electronics up in space use archaic radiation-hardened 100ns nodes for a reason. If you try to put up enterprise grade in orbit, they'll be fried within months due to cosmic ray bombardment and random glitches – unless you pay millions extra to send up weighty shielding. And then how will you replace them? Pay up several million dollars to send Bob up. That's fucking stupid. The math doesn't add up, and won't add up for a good while. Not until there's some big cozy radiation-resistant asteroids to put the server racks in, and a permanent population in space, which might be another century or so.

 

[General Emílio Médici]

Not a problem. Do you really think the ISS just overheats when it's in direct sunlight? No. Their radiators are in line with radiation and shaded by the hull itself. Permanent waste heat removal solution. I'd need to see some actual numbers, but if the ISS can manage a crew of up to ~10 100W space heaters plus a bunch of electronics, lighting, conditioning, etc. with a couple of dinky little plate arrays, even a 1MW continuous heat load should be achievable. And, you're forgetting one extremely important part: you don't need to run computers at human-comfortable temperatures. The radiators on the International Space Station have to radiate at or around 20°C. No such case for a radiator meant to eliminate heat from a computing system, which can be rated to run at 80°C and higher. The efficiency of emitted heat rises exponentially with the emitted temperature compared to background radiation, so the higher you can take it, the better. It might need some more exotic radiator concepts to get rid of much more than 10MW, like wire-loop radiators, but it should be workable

The ISS doesn't generate as much heat as a datacenter would, and your point about computers being able to handle much higher temperatures doesn't work out because while it is true it would mean that the efficiency of the computation would be in the shitter. There is a reason people use cooling to keep them cold and why datacenters have such strong air conditioning bills.

The radiation shielding is a good point though. Enjoy the random bits being flipped by cosmic rays.

 

[not william stenchever]

you can't "vent waste heat outside" in the vacuum of space where there is no atmosphere. the only way to get rid of it is to radiate it off, which is a very slow process.
you also can't use waste heat for power generation without some kind of heat sink, and again the only heat sink available in space would be shitty radiator panels.

Space just really doesn't want us to have fun.

 

["Controlled Demolition"]

The ISS doesn't generate as much heat as a datacenter would, and your point about computers being able to handle much higher temperatures doesn't work out because while it is true it would mean that the efficiency of the computation would be in the shitter. There is a reason people use cooling to keep them cold and why datacenters have such strong air conditioning bills.

The radiation shielding is a good point though. Enjoy the random bits being flipped by cosmic rays.

The point is that it's a solvable problem. There's the possibility of double-coating solar panels to use the obverse side as radiators; of engineering chips specifically for high temperature operation; of material science making future radiators lighter, with greater emissivity, and better thermal conductivity; and so on.

The radiation and the lack of people in space isn't, though. You can only solve one through radiation shielding and the other through being there. Radiation shielding is weight, and weight is loads of money on top. And nobody wants to be in space.

So it's just going to be low-orbit internet relay networks ala Starlink for the foreseeable future. Starlink satellites are largely disposable for a very good reason. Space is corrosive. Space fucking hates you. Things that operate in space for thirty-odd years do so at a great deficit, and with overengineering on top. Techstartups fucking hate overengineering because it cuts into their bottom line and therefore their investors' pockets. Unless Google or equivalent is launching whole datacenters into orbit as a wild billionaire's gamble, nobody's going to do this.

 

[DumbDude43]

The point is that it's a solvable problem. There's the possibility of double-coating solar panels to use the obverse side as radiators; of engineering chips specifically for high temperature operation; of material science making future radiators lighter, with greater emissivity, and better thermal conductivity; and so on.

The radiation and the lack of people in space isn't, though. You can only solve one through radiation shielding and the other through being there. Radiation shielding is weight, and weight is loads of money on top. And nobody wants to be in space.

So it's just going to be low-orbit internet relay networks ala Starlink for the foreseeable future. Starlink satellites are largely disposable for a very good reason. Space is corrosive. Space fucking hates you. Things that operate in space for thirty-odd years do so at a great deficit, and with overengineering on top. Techstartups fucking hate overengineering because it cuts into their bottom line and therefore their investors' pockets. Unless Google or equivalent is launching whole datacenters into orbit as a wild billionaire's gamble, nobody's going to do this.

yeah it's solvable in theory, but is that worth the cost?
like, you'd go through all this trouble, for what? the cost of launching such a space datacenter would be so high, a regular one on earth will be able to do the same things for a fraction of the price.

 

["Controlled Demolition"]

yeah it's solvable in theory, but is that worth the cost?
like, you'd go through all this trouble, for what? the cost of launching such a space datacenter would be so high, a regular one on earth will be able to do the same things for a fraction of the price.

Yeah. And that's why there won't be any "datacenters" in orbit anytime soon. Why go through the effort? What advantages are there? There are none. Compared to just using a LEO orbit network to achieve the advantages of lower latency, or using already established infrastructure and accessibility to develop and maintain a server, there is currently no reason to put actual servers in orbit, and won't be for as long as a server on Earth can't do the same job better, cheaper, and more reliably.

 

[The Masked Sneed]

If heat is an issue, then it makes perfect sense to build a datacenter in space, or on the moon. Just have it orbiting on the "night side" of earth or build it on the dark side of the moon, and vent waste heat outside.

I, uhh, dunno how to tell you this but the dark side of the moon isn't actually dark. It's just the side that points away from the earth. You might try placing it in some polar craters instead since some of those really are in permanent darkness. Actually, that would be a pretty good idea, considering those are prime locations for permanent colonies, which would solve the maintenance problem as well as allow the waste heat to be used for more useful purposes.

 

[Yandex Captcha Solver]

>can't do any maintenance or repair
>can't upgrade or replace any components

Just send tranny system admins up there. A return trip won't be necessary 59% of the time.

 

["Controlled Demolition"]

I, uhh, dunno how to tell you this but the dark side of the moon isn't actually dark. It's just the side that points away from the earth. You might try placing it in some polar craters instead since some of those really are in permanent darkness. Actually, that would be a pretty good idea, considering those are prime locations for permanent colonies, which would solve the maintenance problem as well as allow the waste heat to be used for more useful purposes.

Yeah, they are prime targets for actual planned future colonies for a reason, and will probably see some sensitive hardware placed there in the near future, likely dug beneath regolith that acts as both heatsink and as radiation shielding. The NASA-ESA-JAXA-co. collaborative Lunar Gateway (planned for a 2027 launch, extremely dubious due to the reliance on the notoriously behind-schedule and explosive SpaceX Starship One) plans putting a refueling depot around the Moon in preparation for one of these bases, which would be on the Moon's poles, with Artemis specifically aiming for the Moon's south pole. That's going to be the first datacenter in space – dug underground into the Moon, not in Earth orbit, where it can enjoy reliability, near permanent solar electricity, actual accessibility from on-site astronauts, and protection from space radiation. It's still at an absolute minimum 10 years off, if not multiple decades.

 

[Miller]

This gives me flashbacks of that time during GG when 8chan people were thinking about sending their server in orbit so no activist could take it down.

 

[Cedric_Eff]

>one gazillion ton datacenter space station flying around at 8 km/s View attachment 7380632

VS.

>one little speck of dust

You’d actually be surprised, if the station’s modules have an inflatable Kevlar layer, the dust and debris can bounce off of it. They figured that out with the BEAM module on the ISS.

Also despite how impractical it is, Space Data Centers are something to look into.

I’d kill for Space Kiwis

 

[The Mass Shooter Ron Soye]

That's going to be the first datacenter in space – dug underground into the Moon, not in Earth orbit, where it can enjoy reliability, near permanent solar electricity, actual accessibility from on-site astronauts, and protection from space radiation. It's still at an absolute minimum 10 years off, if not multiple decades.

I forgot about using the whole damn Moon to dump energy into.

Clearly, hot datacenters turning lunar ice into water will be the economic breakthrough venture capital is looking for.

 

["Controlled Demolition"]

I forgot about using the whole damn Moon to dump energy into.

Clearly, hot datacenters turning lunar ice into water will be the economic breakthrough venture capital is looking for.

The Moon is also a tad too far away to be attractive to any tech startup, fortunately. The light-speed delay between the Moon and Earth is 1.35 seconds at minimum. Still within voice call and streaming range, but no longer realtime, but the upper case I Internet could ostensibly spread to it with some minor allowances. Once people settle other planets, however, that's when the Digital (Internet) Age officially ends. All of humanity will no longer be connected by a singular near-realtime communication network, and it will only get worse from there on. Venus has a closest approach at ~41 gigameters, Mars at ~54 gigameters, making their best-case communication latency 137 seconds and 180 seconds respectively. You're no longer holding a video call then. No longer will every single public server be accessible at once, no longer will you be able to rely on largest piece of infrastructure in history to community, and no longer will you have a near-direct speaking path to close to seven billion people. The end of the Internet age may be closer than you think.

 

[The Mass Shooter Ron Soye]

No longer will every single public server be accessible at once, no longer will you be able to rely on largest piece of infrastructure in history to community, and no longer will you have a near-direct speaking path to close to seven billion people. The end of the Internet age may be closer than you think.

Null will make SneedForo the gold standard for DDoS-resistant interplanetary forum software by making optimizations such as removing the need to communicate to the fucking server to quote a post.

 

[74164978231]

I forgot about using the whole damn Moon to dump energy into.

Clearly, hot datacenters turning lunar ice into water will be the economic breakthrough venture capital is looking for.

Large scale computing on the moons of gas giants might end up being very attractive if the energy/heat computing dynamic remains the same into the far future. Could essentially cool your computers with liquid nitrogen for free on a Pluto or Triton. Or just submerge the entire thing in the liquid ethane seas on Titan.

 

[The Mass Shooter Ron Soye]

Large scale computing on the moons of gas giants might end up being very attractive if the energy/heat computing dynamic remains the same into the far future. Could essentially cool your computers with liquid nitrogen for free on a Pluto or Triton. Or just submerge the entire thing in the liquid ethane seas on Titan.

This is obvious but the computing moving far out has to be asynchronous, i.e. the kind of scientific supercomputing that would be done on El Capitan, Aurora, Frontier, etc. Not anything we want access to in seconds like Grok putting an Israeli flag in the background of an artist's work.

There is speculation that Titan could host alternative forms of life with its surface liquid hydrocarbons. So there will be planetary protection concerns, at least early on. Our first good Titan probe should arrive in 2034. We should know long before Titan could ever be used for commercial purposes. It's worth noting that Titan, Pluto, and many other icy objects could have subsurface liquid water oceans. Titan could have ethane/methane up top, but water deep inside.

It's possible that the energy/heat computing dynamic does not remain the same, i.e. 3D chips become the ultimate form of computing but have to scale back the energy consumption to not melt them. On the other hand, it might simply be another solvable engineering challenge, and we'll see heat carried away by microfluidic cooling channels. And then the chips will end up stacked on each other with high speed interconnects much like server racks. In which case there will be plenty of heat to go around.