Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Cloud Space

Could We Lower The Carbon Footprint of Data Centers By Launching Them Into Space? (cnbc.com) 114

The Wall Street Journal reports that a European initiative studying the feasibility data centers in space "has found that the project could be economically viable" — while reducing the data center's carbon footprint.

And they add that according to coordinator Thales Alenia Space, the project "could also generate a return on investment of several billion euros between now and 2050." The study — dubbed Ascend, short for Advanced Space Cloud for European Net zero emission and Data sovereignty — was funded by the European Union and sought to compare the environmental impacts of space-based and Earth-based data centers, the company said. Moving forward, the company plans to consolidate and optimize its results. Space data centers would be powered by solar energy outside the Earth's atmosphere, aiming to contribute to the European Union's goal of achieving carbon neutrality by 2050, the project coordinator said... Space data centers wouldn't require water to cool them, the company said.
The 16-month study came to a "very encouraging" conclusion, project manager Damien Dumestier told CNBC. With some caveats... The facilities that the study explored launching into space would orbit at an altitude of around 1,400 kilometers (869.9 miles) — about three times the altitude of the International Space Station. Dumestier explained that ASCEND would aim to deploy 13 space data center building blocks with a total capacity of 10 megawatts in 2036, in order to achieve the starting point for cloud service commercialization... The study found that, in order to significantly reduce CO2 emissions, a new type of launcher that is 10 times less emissive would need to be developed. ArianeGroup, one of the 12 companies participating in the study, is working to speed up the development of such reusable and eco-friendly launchers. The target is to have the first eco-launcher ready by 2035 and then to allow for 15 years of deployment in order to have the huge capacity required to make the project feasible, said Dumestier...

Michael Winterson, managing director of the European Data Centre Association, acknowledges that a space data center would benefit from increased efficiency from solar power without the interruption of weather patterns — but the center would require significant amounts of rocket fuel to keep it in orbit. Winterson estimates that even a small 1 megawatt center in low earth orbit would need around 280,000 kilograms of rocket fuel per year at a cost of around $140 million in 2030 — a calculation based on a significant decrease in launch costs, which has yet to take place. "There will be specialist services that will be suited to this idea, but it will in no way be a market replacement," said Winterson. "Applications that might be well served would be very specific, such as military/surveillance, broadcasting, telecommunications and financial trading services. All other services would not competitively run from space," he added in emailed comments.

[Merima Dzanic, head of strategy and operations at the Danish Data Center Industry Association] also signaled some skepticism around security risks, noting, "Space is being increasingly politicised and weaponized amongst the different countries. So obviously, there is a security implications on what type of data you send out there."

Its not the only study looking at the potential of orbital data centers, notes CNBC. "Microsoft, which has previously trialed the use of a subsea data center that was positioned 117 feet deep on the seafloor, is collaborating with companies such as Loft Orbital to explore the challenges in executing AI and computing in space."

The article also points out that the total global electricity consumption from data centers could exceed 1,000 terawatt-hours in 2026. "That's roughly equivalent to the electricity consumption of Japan, according to the International Energy Agency."
This discussion has been archived. No new comments can be posted.

Could We Lower The Carbon Footprint of Data Centers By Launching Them Into Space?

Comments Filter:
  • by WoodstockJeff ( 568111 ) on Saturday June 29, 2024 @05:41PM (#64588649) Homepage

    If the data center is solar powered, it shouldn't have a "carbon footprint", right? So why move it to a place that will have a very large carbon footprint to SERVICE? Are they intending to lift people by solar power?

    • by ls671 ( 1122017 )

      Just put them into the sea and have them powered by their own self contained tidal turbine. Only a data link might be required.

      • This is along the lines of what I was thinking, though I remember that Microsoft is closing its underwater data center. [extremetech.com]

        Broadly speaking, data centers only need 3 things: Power, Cooling, and Data Link. If you design them for sufficient redundancy, servicing isn't really a requirement. Just replace the entire module when the time comes.

        As such, while an underwater turbine might not be a practical option, we can do a lot on the cooling front simply by not building in areas with extremely high temperatures an

    • Carbon footprint includes manufacturing and transport, not just the final state
      • by dgatwood ( 11270 ) on Saturday June 29, 2024 @10:03PM (#64589119) Homepage Journal

        Carbon footprint includes manufacturing and transport, not just the final state

        Yeah, this. Let's say that you have a 3,000 pound server rack that draws 60 kW. In one year, that rack will draw 525,600 kWh. At the U.S. average of 0.86 pounds of CO2 per kWh, this is 452,016 pounds of CO2 per year. Or for pure coal, at 2.30 pounds per kWh, this is 1,208,880 pounds of CO2 per kWh.

        Assuming you are putting these into geosynchronous orbit (which would be horrible for latency, but would reduce the complexity greatly compared with trying to keep all of the servers sufficiently in sync when each user is forced to switch servers every four minutes, as would be the case if you put them into low-earth orbit), a Falcon 9 has a limit of 7,700 pounds, so you'll get only two racks per launch (non-disposable), not including power hardware, which might bring it down to only one.

        But assuming you can get two into orbit per launch, at probably O(700) tons of CO2 (1.4 million pounds), that's 700,000 pounds of CO2 per server rack. In other words, the launch itself will use almost as running the data center on coal for 7 months, or 19 months using U.S. fuel mix. Figure a three-to-five-year lifespan, and you've burned up half of the savings right off the bat.

        If you can only launch one, then double that number, and you're no longer saving any emissions at all. And that's before factoring in losses from being unable to do any maintenance.

        And for that, you will spend $67 million for the launch alone, not counting the cost of the hardware itself — enough to provide solar power for 410 such racks even at consumer prices. Or you could spend that $67 million to buy 63 megawatt-hours of storage from Tesla to provide power during the non-daylight hours for 55 of those racks (assuming 5 full-sun-equivalent hours per day).

        Basically, unless my math is way off (like three orders of magnitude off in cost and an order of magnitude off in CO2 emissions), this approach would still make no sense at all even if it didn't add huge latency penalties. Add that into the mix, and in my opinion, we can safely defer to Betteridge's law of headlines without the slightest hesitation and answer the question with a resounding "no".

        There is exactly one reason to build a data center in space, and that is to serve illegal content beyond the reach of any world government. For any legitimate business, there are exactly no plausible reasons to build a data center in space.

        • They have assumed a 1,400km LEO orbit, and a new launch vehicle with 10x lower emissions that doesn't exist yet.

          I also assume they haven't accounted for a global network of ground stations to communicate with the data centres.

          • by dgatwood ( 11270 )

            They have assumed a 1,400km LEO orbit, and a new launch vehicle with 10x lower emissions that doesn't exist yet.

            I also assume they haven't accounted for a global network of ground stations to communicate with the data centres.

            Yeah, that makes it even more bizarre, because you're not just limited by the link to the ground station, but also the speed between satellites, or else you're only going to be using a fraction of your capacity, because there's no way all your satellites in LEO are going to be within range of a ground station unless you have an absurd number of ground stations (like tens of thousands).

            Even at 1400 km, such a satellite would orbit the planet every 113 minutes. That moves you from the four-minute switching t

        • You forgot the weight of 60kW of solar panels, heat sink and gyroscopes. That's an extra 500 pounds for the solar panels and maybe 400 pounds for an active heat sink. Unfortunately your heat sink will also be your greatest source of failure.
          • by dgatwood ( 11270 )

            You forgot the weight of 60kW of solar panels, heat sink and gyroscopes. That's an extra 500 pounds for the solar panels and maybe 400 pounds for an active heat sink. Unfortunately your heat sink will also be your greatest source of failure.

            I didn't forget the solar panels. That's what I was thinking about when I said "not including power hardware, which might bring it down to only one". Gyros, engines to boost its orbit or make it reenter... there are a lot of things missing that I didn't even bother to think about, because I concluded it was infeasible before I even dug in that far. :-)

            Unfortunately, your estimate of the heat sink's mass is actually low by an order of magnitude, which makes the odds of a "two racks" solution less than zero

            • I assumed the heat sink was about 300K, so it would radiate 1kw per m^2. So only 60m^2 of surface is needed. Liquid water at 1/4 atmosphere could be used to move heat through the heat sink. The heat sink would be in the shadow of the solar panels and likely need further shading but that is light. An active heat sink like that could be as little as 3kg per m^2. I envision the out going water at 35C and the returning water at 15C leaving a 15C margin of error. It wouldn't be ridged on earth. It would
    • Not really sure here, but I know the sci-fi plot version of this plan assumes that there's vastly more direct sunlight available just outside the atmosphere, and combined with the micro-gravity practically eliminating restrictions on structure sizes, maybe they think there will be enough extra free solar power up there that it will offset the launch costs.

      Mind you, my first concern isn't the power consumption, it's the latency. Only someone who's never played video games online would think this is a good id

    • So why move it to a place that will have a very large carbon footprint to SERVICE?

      Forget servicing, you need to launch the thing in the first place and your carbon footprint to launch the thing is going to make it much worse and massively more expensive than the "boring" solution of just running a normal data centre on Earth using solar power and batteries. In addition to servicing cooling is going to be a major issue too as vacuum is a great insulator and that cooling will add to the launch weight as well.

      It's great to have out-of-the-box ideas but this one really should not have ma

    • Collect energy from the sun and convert it into thermal energy via data centre CPUs and network infrastructure. Pump it out into earth's environment. What could possibly go wrong?
      • Most of the energy from that sun is converted directly into heat when it hits the ground. May as well put a solar panel in the way and do some work with that energy.

  • by locater16 ( 2326718 ) on Saturday June 29, 2024 @05:42PM (#64588651)
    Space, it's cold in space right, that means cooling is easy? I bet datacenters don't weigh much either. By god I've solved it better write an opinion article thinly veiled as a "study" and then see if I can get a government grant for doing so!
    • by msk ( 6205 )

      You're going to keep it permanently in Earth's shadow? How much ground-extracted energy will that take?

      • by ls671 ( 1122017 )

        It could take none actually to keep it always in Earth shadow but I guess latency would suck! /s
        see L2:
        https://en.wikipedia.org/wiki/... [wikipedia.org]

        • by msk ( 6205 )

          I completely blanked on Lagrange points. It could still take a little bit of thrust to keep it there and apart from our other probes at L2, but an ion unit might handle it, as long as it has a power source. If it's in Earth's shadow, solar might be pretty weak.

    • by Darinbob ( 1142669 ) on Saturday June 29, 2024 @06:12PM (#64588721)

      The snag is that they will still get very hot, but there's no atmosphere to let that heat radiate away. Cooling is a necessity but also difficult to do in space.

      • by GrpA ( 691294 )

        Yep, I was thinking the same thing... Unfortunately, Hollywood gives this crazy idea that space is very cold. The only way to cool stuff in space is to radiate vast amounts of heat in the infrared, which is going to require heat pumps into black body radiators so as to lose energy through infrared photonic emissions - another technology we haven't developed yet.

        • Exactly. Few megawatts in space... I wonder if they really did their homework, or that this is just something political.
          Maybe they just want a free ride to the iss so that they can observe why it is... difficult?
        • Absolute vacuum has no temperature at all, just as it has no taste or color.

          In nearly empty space, heat can be radiated and absorbed through radiation. Earth's is heated by the sun's radiation, to about 14 degrees Celsius.

          But an astronaut in the Earth's shadow can only absorb heat from distant stars and galaxies, which is much less than what he emits. Eventually, without solar radiation, his temperature will drop to the average space temperature in space of 2.7 Kelvin.
      • by Roger W Moore ( 538166 ) on Saturday June 29, 2024 @11:08PM (#64589185) Journal

        ...but there's no atmosphere to let that heat radiate away.

        You do not need an atmosphere to radiate the heat away - in fact radiating heat away is the only to lose heat in a vacuum. What you mean is convect the heat away which is much better at removing heat than radiating it.

      • by kevmeister ( 979231 ) on Saturday June 29, 2024 @11:16PM (#64589195) Homepage

        The snag is that they will still get very hot, but there's no atmosphere to let that heat radiate away. Cooling is a necessity but also difficult to do in space.

        Well, kinda close. The only way to get rid of heat in space IS radiation. Infrared travels though a vacuum just fine. That is how most satellites are cooled. And, yes, cooling is required by most satellites. Some satellites carry coolants. Liquefied HE or something similar.

        Cooling in space is a bitch and cooling a data center of any size will be VERY tricky. Since I know hat they have thought of this, I'll be very curious to learn exactly what.

    • Cooling is about the single hardest thing to do in space. First, no, itâ(TM)s not cold, itâ(TM)s just youâ(TM)re getting 0 conductive or convective heating. The problem is, youâ(TM)re also getting zero conductive or convective cooling. As far as radiative heating goes, youâ(TM)re getting a metric fucking ton of that for half your orbit. As far as radiative cooling goes, youâ(TM)d better have some huge ass radiators to get rid of all the heat. Radiators are usually some of

    • "I bet datacenters don't weigh much" you would loose that bet, data centers are so heavy most buildings for them have to be designed specifically with the mass in mind.
    • Space, it's cold in space right, that means cooling is easy?

      Deep space is cold.

      Space at one AU is 283K or 50F.

      That doesn't sound so bad, but shedding heat into a vacuum isn't easy. The only way to do it is with a big blackbody radiator shaded from the sun.

      Going to space for cooling doesn't make much sense.

      • Oh but don't worry, as they say at the end of the article "Space data centers wouldnâ(TM)t require water to cool them, the company said." Good thing they won't use water!

        Clearly some people are thinking hard about this possibility. I'd really like to see a list of benefits of space-based data centers. Given the effort to get there, they better be huge benefits.

    • I could see adding some limited CDN features (caching) to Starlink satellites.
    • Temperatures on the moon range between 121C in daytime and -133C at night.
      If you want to have solar power, then you are going to need to be in daytime, so, it isn't cold in space.

    • The important things about studies is that experts who know what they are talking about are involved. You didn't solve anything. You set fire to your datacentre. Space is cold, but cooling something in space is very VERY difficult. Even your phone would overheat and burn up in the vacuum of space. Let's go through what you wrote:

      Space, it's cold in space right, that means cooling is easy?
      Wrong. Cooling is difficult
      I bet datacenters don't weigh much either.
      Wrong. Datacentres are incredibly heavy.
      By god I've

    • Space is actually a *really* good insulator. It wouldn't be easy to shed all the heat generated by the equipment.

  • Oh Pulleeze (Score:5, Funny)

    by spaceman375 ( 780812 ) on Saturday June 29, 2024 @05:47PM (#64588665)
    Just like the dot com bubble included lots of bad patents that were "This same thing everybody does" but "On a computer!" we now have "But in Space!" I truly doubt that they have calculated ROI even remotely correctly. This is just fishing for a fool with money. Mind you, I'm all for "In Space!" but this ain't it. BTW, a plug for a fave site: centauri-dreams.org [centauri-dreams.org]. Give it a read.
  • by Retired Chemist ( 5039029 ) on Saturday June 29, 2024 @05:48PM (#64588667)
    Are you just going to throw them away, when they stop working? How often would they have to be replaced? Solar power is available, but the size of the solar panels necessary to power a data center in space would be immense and do you really want a high-powered nuclear reactor orbiting overhead. Cooling would be an issue as well. Space is cold but has little ability to absorb heat. Cooling would be by radiation, which is very inefficient and of course the reactor, if you used one, would generate a great deal of excess heat. The whole idea seems very half-baked.
    • Throw way data centers... It it fails, de-orbit and watch them burn!
    • There's already a solar array and cooling system on the ISS, which is scheduled to be de-orbited fairly soon. The solar array is said to have a peak generation capacity of about 240kW (though the average is about 100kW, as the system apparently alternates between gaining power from the solar array white heating up, and rejecting heat to the cold of space while in the shade of the Earth), so all that would be required is to send up a buttload of NVidia modules to prototype a 10% scale version of the system e

    • How about the permanently shaded areas of the moon where there's a lot of ice that can be used to absorb and radiate out the heat from a large surface area. The data center can be solar powered, because the panels can be in the exposed areas. The only issue is the one second of latency, which may be a non-issue for certain applications like supercomputing or AI training.

    • >The whole idea seems very half-baked.

      bah.

      Once you start by assuming non-existent launch technology, yet constrain the completion to today, anything else is a mer detail.

  • Why not just put them on north-facing mountain slopes in Alaska? It's far far easier to build, transport, and manage them there than space.

    • Alaska already has a problem with their glaciers melting [alaska.edu] faster [scientificamerican.com] than anticipated [nationalgeographic.com]. Do you want to exacerbate that problem with all that additional heat?

  • by Kiliani ( 816330 ) on Saturday June 29, 2024 @05:59PM (#64588695)

    Pigs. In. Space!! (imagine it said with big reverb ...)

  • by ghinckley68 ( 590599 ) <sd@glenhinckley.com> on Saturday June 29, 2024 @06:08PM (#64588715) Homepage

    Space is not cold it is not hot, it is in fact nothing. There is just nothing to keep them cool. The nothing of space means there is nothing to carry way the heat. A human would take years to freeze in space.

    Like rods of the gods stop watching sci-fi and read a learn a little more sci.

    • Technically, and pedantically, the best combination of -alities, space is not a vacuum. It's just a lot close to being a vacuum than on earth. Especially in any orbit these satellites are likely to be in. Still though, effectively not enough particles bouncing around to provide any good radiation of heat.

      • Still though, effectively not enough particles bouncing around to provide any good radiation of heat.

        If they/we could tap into quantum foam for heat dissipation a large problem of running things in space would be solved.

    • I was thinking the same thing and now think they must be planning some kind of phase change setup for this.
    • There is just nothing to keep them cool. The nothing of space means there is nothing to carry way the heat.

      But but but...the fine company says these datacenter modules wouldn't need cooling. They wouldn't lie to us, would they?

      • It's not lying when you're stupid enough to think you're telling the truth. This is one of those cases where it actually takes some understanding of science to know why this is so wrong. Very many people will take this at face value - space is cold - they saw it in the movies.

        • It's not lying when you're stupid enough to think you're telling the truth.

          Ah, so Trump didn't lie. (ducks)

  • Of all the dumbest of the dumb ideas ...
    like solar freaking roadways and AA batteries with a boost convertor
    this one has to be at the top of the pile of dumb.

    • Of all the dumbest of the dumb ideas ... like solar freaking roadways and AA batteries with a boost convertor this one has to be at the top of the pile of dumb.

      I haven't seen anyone with a really good reason this is a dumb idea. How do I put it? I would hope that anyone I was at war with put all their datacenters in orbit. Makes my job a lot easier to put all their datacenter out of action. A few hundred kilos of small ball bearings launched to the same orbital shells but in retrograde, should do the job on the cheap.

  • by edibobb ( 113989 ) on Saturday June 29, 2024 @06:34PM (#64588753) Homepage
    Have you seen a major data center? More than a square kilometer of multistory buildings. Now think heat dissipation, using radiation only. (Is this an AI article?)
  • by gavron ( 1300111 ) on Saturday June 29, 2024 @06:34PM (#64588759)

    > ArianeGroup, one of the 12 companies participating in the study, is working to speed up the development of such reusable and eco-friendly launchers.

    Ariane:
    ArsTechnica has a great article (that's been referenced in two slashdot articles) where Arianegroup poo-poos reusability and is working hard on the long-delayed Ariane 6, none of which is reusable. They're quoted as far back as ten years ago saying reusability will never be worth it (pretty much calling SpaceX stupid). There is no chance in hell they'll develp any "reusable" anything in the next ten years.

    Eco-friendly:
    Eco-friendly launchers don't exist, although if SpinLaunch were a real thing instead of an investor ripoff, that's half eco friendly. Nobody talks about how they get the energy to "spin up" for that launch. You can't wind a spring with wishful thinking.

    RTT/Latency:
    At 3.34s/1000m that's 4.7sec each way, or an RTT of 9.4sec. That won't work for live two-way anything including VoIP, televideo, video games, and that magic telemedicine 5G carriers promised. In other words, it's good for data storage, retrieval, and one-way (e.g. HTTP/S) transfers.

    Constellation:
    A constellation at 1400km orbit of only 12 satellites guaranteed period signal so poort only fixed-base dishes on Earth would be usable. For practical purposes it wouldn't be "a datacenter in space" but more of a space "ground station".

    Power:
    The solar grids on the ISS have shown that with multiple launches and in-space assembly they can generate 240KW. A datacenter that only uses that little power would be somewhere between worthless and "give me government money boondoggle."

    Financially unfeasable:
      It's about as realistic as SpaceX building 30,000 Starships to colonize mars. The sheer amount of materials, the cost to put them together, and the lack of any ROI makes that idea DOA as well. If a hypothetical man worth $220B (today) wanted to fund those at today's dollar and stock value he'd have to spend his entire fortune AND each Starship would have to cost less than $7.6M including testing, fuel, LoX, launchpad, regulatory fees, etc. The current cost per launch of Starship is estimated at 100B, or an order of magnitude higher. It kind of makes me wonder where they'd expect to find 100Bx30K=3 quadrillion (3,000 trillion) dollars.

    It's good to study things, propose novel solutions, test the math, etc. but from a purely networking perspective, a physics perspective, and an inability of the named player(s) to do anything of the sort, this is DOA.

    • by GFS666 ( 6452674 ) on Saturday June 29, 2024 @09:29PM (#64589077)

      The current cost per launch of Starship is estimated at 100B, or an order of magnitude higher. It kind of makes me wonder where they'd expect to find 100Bx30K=3 quadrillion (3,000 trillion) dollars.

      I don't disagree with your fundamental conclusion that having data centers in space is really a bad, bad idea. However, I would like to point out that current cost of Starship per launch is estimated to be approximately $90 Million. The entire cost to develop Starship will be around 5 Billion (estimated) with those sunk costs spread out over the life of the program. Link to estimated costs can be found here: https://forum.nasaspaceflight.... [nasaspaceflight.com] . Please note that I am NOT a Musk Fanboy and I have serious concerns about SpaceX's desire to capture an entire freaking booster with the capture arms. I don't like the vertical landing aspect of Starship either and feel that they should have done a landing via a runway like the Space Shuttle but then again you can't do that on Mars. Time will tell.

    • RTT/Latency: At 3.34s/1000m that's 4.7sec each way, or an RTT of 9.4sec.

      Looks like you dropped a prefix somewhere. It takes 3.34 micro seconds for light to travel 1000m, giving a RTT of 9.4 milliseconds. Of course, that's only if the satellite is directly above the receiver.

      For comparison, I have Starlink. Starlink satellites orbit at about 550km. There's no noticeable lag on video calls. RTT ping time from here to slashdot.org is about 35ms, and that's 4 one-way trips between ground and the satellite.

    • by Hank21 ( 6290732 )
      Your numbers are way off - latency for 826 miles is around 30-60ms.
  • by The Cat ( 19816 ) on Saturday June 29, 2024 @06:35PM (#64588767)

    Jake Busey screams into an iPhone as ten billion dollars worth of computers slams into the Atlantic Ocean at Mach 23.

  • You probably can't lower the carbon footprint of anything by launching it into space, unless it's getting launched with renewable power by a railgun or a space elevator. The carbon footprint of even a small rocket launch is just too huge to offset the activity of anything it can lift. And then you have to ask about the additional carbon footprint of building the thing to work in space vs. what would be required to make it work on land.

    And what's harder than lowering the carbon footprint of something by laun

  • by hotdogee ( 2696891 ) on Saturday June 29, 2024 @06:47PM (#64588795)
    Written on there website https://ascend-horizon.eu/ [ascend-horizon.eu]:

    position data centers in space, in order to have low-carbon solar electricity and to avoid the energy necessary for cooling (assured by the vacuum of space).

    • Right! We use vacuum-insulated mugs (and other stuff) because a vacuum is an excellent insulator. It wouldn't be easy to shed all that CPU heat and radiate it into space.

    • Ah, just crack open a window and turn on the fan. It'll be fine.
    • There is no need to understand cooling if you understand how EU science funding works.

      What can also be found on the site is "This project has received funding from the European Union’s Horizon Europe research and innovation Actions under grand agreement n101082517". That should explain it nicely.

      So what the people behind this proposal do is to leech taxpayer money which is spent by stupid bureaucrats in Brussels. It ticks all the boxes:

      • - Zero Emissions
      • - Sovereignty
      • - In space!

      Just what th

  • Cooling (Score:4, Informative)

    by bill_mcgonigle ( 4333 ) * on Saturday June 29, 2024 @07:00PM (#64588819) Homepage Journal

    What thermodynamics breakthrough do they have?

    I know a guy who designs satellite cooling systems for .tla and I've seeen how it's incredibly difficult even for small computers because everything needs to be radiatively cooled and not while the sun is heating it! Like even the microscopic texture of a copper heatsink is critical.

    An orbital data center might need a radiator hundreds of times the size of the satellite.

    Are they hiding some new tech?

    Surely they have not gotten this far on an elementary mistake?

    • by JBMcB ( 73720 )

      I remember reading an IEEE article a while back about how vacuum tubes are still used in some applications for satellites because they are more resistant to damage by solar flares and radiation than semiconductors, and their performance doesn't degrade when they get hot.

      • I remember reading an IEEE article a while back about how vacuum tubes are still used in some applications for satellites because they are more resistant to damage by solar flares and radiation than semiconductors, and their performance doesn't degrade when they get hot.

        Yup, there are some things that hollow state technology does pretty well. They are also resistant to emp. They certainly don't look like Grandpa's radio tubes, for certain.

    • What thermodynamics breakthrough do they have?

      I suspect the "breakthrough" is a white paper with some bad math.

      This is just space based solar power by another name and the thermodynamics on space based solar is horrible, we have far better options on terra firma.

      Elon Musk has been asked about space based solar power many times because of his association with solar power and rockets. He's stated in interviews on how he's quite annoyed about these questions because it means having to explain again the horrible thermodynamics of space based solar power.

  • new type of launcher that is 10 times less emissive would need to be developed.

    and order of magnitude improvement in non-digital technology I have to dream up, build, test, and mass-produce. But that's no big deal. Can I have my money now?

  • by Baron_Yam ( 643147 ) on Saturday June 29, 2024 @07:30PM (#64588867)

    We're going to need a series of space elevators around the Earth, then down within the upper atmosphere you connect them into a ring. You can add the occasional small-scale extension and orbital counterweight to keep the ring lifted.

    Then you get your data centre and hitch it to that ring, and it travels around the Earth fast enough that it's always midnight wherever it is. You use the counterweights on the space elevators to mount large solar arrays, and whichever ones are exposed to sunlight can energize the ring to power the mobile data centre.

    This gives you excellent atmospheric cooling and uses the Earth as a heat shield against the Sun.

    Some people will tell you it's impossible. To them I say, it's merely really, really impractical... and probably impossible.

  • salt water is more corrosive than fresh water, why not a cold fresh water lake? build a submersible water cooled datacenter that can be easily hoisted up to the deck of a small ship or large boat in case maintenance & repairs are necessary
    • salt water is more corrosive than fresh water, why not a cold fresh water lake? build a submersible water cooled datacenter that can be easily hoisted up to the deck of a small ship or large boat in case maintenance & repairs are necessary

      Considering the amount of heat thrown off by a datacenter, they'll end up with a pretty warm freshwater lake. It would be an environmental disaster.

  • It's the data center that will be the vacuum, and the problem is that the combination won't be changed from the factory default. OMG it's the same as my briefcase! 1-2-3-4....

  • - latency much?
    - space walk to replace a failed drive?
    - repeated damage from orbital debris?
    - and we'd eventual have to rescue Matt Damon

  • Will look back at this decade and laugh their fucking arses off at all the crazy hare-brained ideas that people have been proposing in the name of climate change. They will be saying, "were we really that stupid?"
  • This feels like a bait article honestly. And here I am falling for the bait.

    Launch costs oh yeah lets just handwave away the climate impact of that. Never mind how expensive it is to launch things either, the cost of hardware and installation doesn't matter if the CO2 goes down (except that it does because that money could be used to save CO2 emissions somewhere else, obviously).

    The obvious latency issues. The wildly unreasonable cost of what it takes to fix anything that fails in orbit. I could go on but m

  • Given the eco impact of an infrastructure required to launch data centers into space. ... Who's asking these questions? A weed smoking 9-year old or something?

  • AI datacenters should be launched straight into the Sun for maximum efficiency.
  • I people weren't so averse to modern nuclear reactor designs (lot safer than the old pressurized vessel water cooled reactors), the solution is a small modular reactor (SMR) that can generate up to 100-150 megawatts per SMR to power a data center. That way, the data server center don't need to hog a lot of land for a nearby solar and wind generating farm, and any power generated by the SMR could be put into the local power grid, too.

  • https://ascend-horizon.eu/acti... [ascend-horizon.eu]

    This is the plan, not the conclusion, but it does detail the solar panels and radiators required, for example.

    Hopefully more details of the study would get published.

  • Nobody is going to spend the billions it will take to do this, even if there were a chance of it working.

  • It would take a LOT of rocket launches to put a data center in orbit. And rocket launches spew out a LOT of carbon. It would be really hard to overcome that, in terms of achieving a carbon-neutral data center.

    https://www.bbc.com/future/art... [bbc.com]

  • An orbital DC is supposed to have a lower carbon footprint, what about all the CO2 involved in getting the tones of equipment up there in the first place, not to mention ferring all the people there to get it installed. + the rtt would suck, hmm I might be an idiot but this does seam to have quite a few issues
    • They're already assumed a reusable rocket with 10x fewer emissions will exist.
      They didn't count any of the emissions in designing, testing and manufacturing the imaginary rocket though.

  • Space is actually a bad place to put things that need to stay cool. Spacecraft and the ISS need radiators to help get rid of heat or they would cook the crew alive. With no convection to get rid of waste heat (no air to transfer the heat to) they quickly heat up and cook the crew. The ROI of launching the computers and cost of the radiator systems that will need repairs when they spring a leak is horrible. It's cheaper just to put them seaside and pump in cold seawater into a heat exchanger, and a lot c
  • Better launch them into the Sun
  • One thing about being trained as an engineer: you can just look at some ideas, and immediately see how utterly stupid they are. This is one.

    Literally the only advantage of this idea is the increased efficiency of the solar cells (24 hour sunlight, no atmosphere). Launch costs? Maintenance?

    Thales Alenia, a joint venture between aerospace-and-defense companies Thales of France and Leonardo of Italy, collaborated with partners such as aerospace company Airbus, server and cloud-software company Hewlett Packard Enterprise and telecommunications group Orange in the study.

    They got themselves some government money. By saying "may be feasible" they can apply for more government money. In my younger days, I had the misfortune to be on an EU research project like that - the lead investigators

/earth: file system full.

Working...