[Starlink] Re: Moving servers or production from Earth to space is not rational in the first half of the twenty-first century

[Starlink] Re: Moving servers or production from Earth to space is not rational in the first half of the twenty-first century

[Colin_Higbie]

A bunch of posts recently seem to think there may be economic or related benefits to moving some of the production or technical systems from terrestrial-based to space-based. While it's technically feasible to do this, of course, there are very few scenarios where this would make economic sense. The only pieces that make sense to put in space, at least in today's (and the near-future's) economic environment, are those that MUST be in space to perform their functions. For example, Starlink must have satellites in space in order to provide line-of-sight access to remote locations on Earth that don't have access to terrestrial-based Internet.

I don't know anywhere near as much as the networking experts do here on those topics, but I do have a physics and economics background and can speak to some of these questions:

1. Cooling in space, even if in the shadow of Earth is MUCH WORSE than cooling on Earth. The only cooling that can occur passively in space is radiant cooling (heat leaving in the form or infrared or other EM waves, but almost entirely IR). This is dramatically less effective than convective (contact with a cooling fluid, whether liquid or gas) or conductive (contact with a solid, like a copper heatsink, which is typically just a way to move heat a short distances to a convective cooling option, like from the inner workings of a CPU in your PC to the outer surface of the chip, and then from the surface again to the fins or a liquid cooling system where a fan then blows room temperature air over a radiator. In effect, the partial vacuum of space in the inner solar system, lacking many such molecules to absorb and move this thermal energy, is an excellent insulator.

Now, you could use laser cooling to cool specific locations via a different form of quantum-induced radiant cooling more effectively than purely passive radiant cooling. This uses lasers to excite atoms in a particular way so they emit high energy photons, taking more thermal energy away with them than incident by the laser and more than could occur via standard IR radiant emissions, yielding a net cooling effect where the laser hits the material, but per the second law of thermodynamics, there is still net heat production this way, with a new added nearby heat source (the laser's power source, which also adds weight and cost and increases the power requirements of the satellite). Those also need to cool somehow via radiant cooling. Still, if the goal is just to create one part of a space-based system that is very cold for superconductivity, transferring heat to other parts where temperature doesn't matter as much, this approach could be a worthwhile tradeoff.

2. It costs dramatically more to put or build something in space than ship/truck it around or build it on Earth. Unless there are particular benefits to low/zero-g or the near-vacuum of LEO, it's MUCH more efficient to just do it on Earth. With possible exceptions for some exotic materials, all of what is built in space now is done for R&D purposes, not actual commercial production where costs matter. Note that we can create vacuum chambers on Earth with fewer particles per cubic meter than an Earth-bound satellite would experience and at a lower cost than putting something in orbit. To the extent that there are advantages to working in space in the inner solar system, they are based not on the vacuum but on the effective zero-g of being in freefall (very, very slight resistance from non-total vacuum also means not quite pure freefall and so not quite zero-G, but close enough for most purposes). 

Also note that while LEO is not a pure vacuum, it's at least still protected by Earth's magnetic field. The vacuum improves at higher altitudes (above LEO), but those also lose protection from the earth's magnetic field against charged particles (and all orbitals, including LEO lack the protection against radiation available only down here below the ionosphere). Both of those conditions of space can wreak havoc on material production.

3. If the goal is a lunar (or other extra-Earth) base for long-term production and self-sufficiency, then it may be more cost effective to build it there using local materials (depending on what's available), but that's only because of the large expense of launching from Earth's high gravity. And maybe eventually with sufficient mining activity and economies of scale, that may one day even be more efficient for materials used in Earth's orbit (i.e., cheaper to launch them from the moon's low gravity than from the Earth's higher gravity), but that is a distant future scenario. For now, before that mining is ready to occur, we'd still need to find the relevant mineral veins on those extra-terrestrial locations, establish mining bases, launch and send heavy and expensive mining equipment, etc. all before any such investments could begin to pay for themselves. In the near term, and at least for the next decade or two (probably longer), working from Earth for anything intended for communication use on Earth or in orbit around Earth will be faster, easier, and cheaper. 

I don't write this to discourage extra-terrestrial mining and development. I'm excited for that future. I do think it's inevitable, just not for at least a few more decades, maybe a century or more. Rather, I write this to remind that Starlink works today because Musk and team understood the real-world costs and limitations of available technology and worked with that to make something viable in the near-term. They appreciated what they could and couldn't do and put together a realistic plan that could be executed with existing technology at an affordable cost. That approach is essential to move things forward. If an approach can't provide a return on investment in a reasonable timeframe, it won't be attractive to investors and will die. 

Exceptions: if the goal is not efficiency, but something else, like we need a space-based relay to talk with humans staying on the dark side of the moon or for matters of national defense, then perhaps moving some of the computing power and data storage (like a CDN) closer to those locations to reduce latency for them makes sense. I'm only rejecting proposals that imply there would be advantages for users on Earth to moving data centers or anything else that can be done on Earth into space.

Cheers,
Colin

[Starlink] Re: Moving servers or production from Earth to space is not rational in the first half of the twenty-first century

[David Lang]

I would agree in the next few years, but I wouldn't be confident projecting more 
than a few years out, a lot can change in 25 years.

David Lang

On Sat, 22 Nov 2025, Colin_Higbie via Starlink wrote:

> Date: Sat, 22 Nov 2025 20:19:06 +0000
> From: Colin_Higbie via Starlink <starlink@lists.bufferbloat.net>
> Reply-To: Colin_Higbie <CHigbie1@Higbie.name>
> To: "starlink@lists.bufferbloat.net" <starlink@lists.bufferbloat.net>
> Subject: [Starlink] Re: Moving servers or production from Earth to space is
>     not rational in the first half of the twenty-first century
> 
> A bunch of posts recently seem to think there may be economic or related benefits to moving some of the production or technical systems from terrestrial-based to space-based. While it's technically feasible to do this, of course, there are very few scenarios where this would make economic sense. The only pieces that make sense to put in space, at least in today's (and the near-future's) economic environment, are those that MUST be in space to perform their functions. For example, Starlink must have satellites in space in order to provide line-of-sight access to remote locations on Earth that don't have access to terrestrial-based Internet.
>
> I don't know anywhere near as much as the networking experts do here on those topics, but I do have a physics and economics background and can speak to some of these questions:
>
> 1. Cooling in space, even if in the shadow of Earth is MUCH WORSE than cooling on Earth. The only cooling that can occur passively in space is radiant cooling (heat leaving in the form or infrared or other EM waves, but almost entirely IR). This is dramatically less effective than convective (contact with a cooling fluid, whether liquid or gas) or conductive (contact with a solid, like a copper heatsink, which is typically just a way to move heat a short distances to a convective cooling option, like from the inner workings of a CPU in your PC to the outer surface of the chip, and then from the surface again to the fins or a liquid cooling system where a fan then blows room temperature air over a radiator. In effect, the partial vacuum of space in the inner solar system, lacking many such molecules to absorb and move this thermal energy, is an excellent insulator.
>
> Now, you could use laser cooling to cool specific locations via a different form of quantum-induced radiant cooling more effectively than purely passive radiant cooling. This uses lasers to excite atoms in a particular way so they emit high energy photons, taking more thermal energy away with them than incident by the laser and more than could occur via standard IR radiant emissions, yielding a net cooling effect where the laser hits the material, but per the second law of thermodynamics, there is still net heat production this way, with a new added nearby heat source (the laser's power source, which also adds weight and cost and increases the power requirements of the satellite). Those also need to cool somehow via radiant cooling. Still, if the goal is just to create one part of a space-based system that is very cold for superconductivity, transferring heat to other parts where temperature doesn't matter as much, this approach could be a worthwhile tradeoff.
>
> 2. It costs dramatically more to put or build something in space than ship/truck it around or build it on Earth. Unless there are particular benefits to low/zero-g or the near-vacuum of LEO, it's MUCH more efficient to just do it on Earth. With possible exceptions for some exotic materials, all of what is built in space now is done for R&D purposes, not actual commercial production where costs matter. Note that we can create vacuum chambers on Earth with fewer particles per cubic meter than an Earth-bound satellite would experience and at a lower cost than putting something in orbit. To the extent that there are advantages to working in space in the inner solar system, they are based not on the vacuum but on the effective zero-g of being in freefall (very, very slight resistance from non-total vacuum also means not quite pure freefall and so not quite zero-G, but close enough for most purposes).
>
> Also note that while LEO is not a pure vacuum, it's at least still protected by Earth's magnetic field. The vacuum improves at higher altitudes (above LEO), but those also lose protection from the earth's magnetic field against charged particles (and all orbitals, including LEO lack the protection against radiation available only down here below the ionosphere). Both of those conditions of space can wreak havoc on material production.
>
> 3. If the goal is a lunar (or other extra-Earth) base for long-term production and self-sufficiency, then it may be more cost effective to build it there using local materials (depending on what's available), but that's only because of the large expense of launching from Earth's high gravity. And maybe eventually with sufficient mining activity and economies of scale, that may one day even be more efficient for materials used in Earth's orbit (i.e., cheaper to launch them from the moon's low gravity than from the Earth's higher gravity), but that is a distant future scenario. For now, before that mining is ready to occur, we'd still need to find the relevant mineral veins on those extra-terrestrial locations, establish mining bases, launch and send heavy and expensive mining equipment, etc. all before any such investments could begin to pay for themselves. In the near term, and at least for the next decade or two (probably longer), working from Earth for anything intended for commu
 ni
> cation use on Earth or in orbit around Earth will be faster, easier, and cheaper.
>
> I don't write this to discourage extra-terrestrial mining and development. I'm excited for that future. I do think it's inevitable, just not for at least a few more decades, maybe a century or more. Rather, I write this to remind that Starlink works today because Musk and team understood the real-world costs and limitations of available technology and worked with that to make something viable in the near-term. They appreciated what they could and couldn't do and put together a realistic plan that could be executed with existing technology at an affordable cost. That approach is essential to move things forward. If an approach can't provide a return on investment in a reasonable timeframe, it won't be attractive to investors and will die.
>
> Exceptions: if the goal is not efficiency, but something else, like we need a space-based relay to talk with humans staying on the dark side of the moon or for matters of national defense, then perhaps moving some of the computing power and data storage (like a CDN) closer to those locations to reduce latency for them makes sense. I'm only rejecting proposals that imply there would be advantages for users on Earth to moving data centers or anything else that can be done on Earth into space.
>
> Cheers,
> Colin
> _______________________________________________
> Starlink mailing list -- starlink@lists.bufferbloat.net
> To unsubscribe send an email to starlink-leave@lists.bufferbloat.net
>

[Starlink] Lower latency in the Netherlands?

[Inemesit Affia]

https://x.com/WietseWind/status/1992549150479516078

Does anyone have probes to check?

Datacenter in RIPE?

I know SpaceX is present in Amsterdam but the traffic has historically gone elsewhere

[Starlink] Re: Lower latency in the Netherlands?

[J Pan]

no obvious rtt https://atlas.ripe.net/probes/1004353/results/wellknown?zoom=custom&start=1763826161&stop=1763913124
nor routing

"RIPE 209.198.140.138 14593 Twekkelerveld Overijssel NL
customer.frntdeu1.pop.starlinkisp.net. still associated with the
frankfurt pop, 209.198.140.0/24,frntdeu1,fra
Traceroute from 209.198.140.138 to 149.19.108.213
Using RIPE, Probe 1004353

STARTED QUERY AT 2025/11/23 15:56:40 UTC

Fetching Measurement: 140344362
Traceroute from 10.13.37.1 to 149.19.108.213 (149.19.108.213):
1 10.13.37.254 4.494ms 0.265ms 0.239ms
2 100.64.0.1 30.856ms 21.841ms 16.651ms
3 172.16.252.36 22.768ms 32.524ms 27.214ms
4 * * *
5 undefined.hostname.localhost (206.224.65.164) 19.894ms 37.878ms
33.879ms in frankfurt
6 2(SERVFAIL (149.19.108.186) 97.792ms 101.878ms 97.871ms to new york city
7 2(SERVFAIL (149.19.108.40) 109.852ms 105.844ms 101.892ms in new york city
8 2(SERVFAIL (149.19.108.67) 117.803ms 125.879ms 121.844ms to chicago
9 2(SERVFAIL (149.19.108.20) 121.835ms 117.946ms 119.654ms in chicago
10 2(SERVFAIL (149.19.108.213) 141.826ms 139.762ms 141.827ms to denver

Completed in 24.39s"

changes internally or externally

"Traceroute from 209.198.140.138 to 1.1.1.1
Using RIPE, Probe 1004353

STARTED QUERY AT 2025/11/23 16:07:13 UTC

Fetching Measurement: 140345358
Traceroute from 10.13.37.1 to 1.1.1.1 (1.1.1.1):
1 10.13.37.254 5.065ms 0.243ms 0.242ms
2 100.64.0.1 17.386ms 16.309ms 17.526ms
3 172.16.252.36 19.757ms 18.453ms 21.799ms
4 undefined.hostname.localhost (206.224.68.196) 25.787ms 21.891ms 16.519ms
5 undefined.hostname.localhost (206.224.70.83) 26.913ms 21.86ms 25.881ms
6 customer.lax3.mc.starlinkisp.net (206.214.226.9) 21.754ms 25.882ms 16.637ms
7 162.158.84.109 17.515ms 26.909ms 21.845ms
8 one.one.one.one (1.1.1.1) 25.867ms 16.51ms 21.646ms

Completed in 8.68s"

according to https://atlas.ripe.net/probes/1004353/overview
(Starlink@UTwente). ping without traceroute is not conclusive (ping,
traceroute and tcpdump are the trio for all problems ;-)
--
J Pan, UVic CSc, ECS566, 250-472-5796 (NO VM), Pan@UVic.CA, Web.UVic.CA/~pan

On Sun, Nov 23, 2025 at 3:34 AM Inemesit Affia via Starlink
<starlink@lists.bufferbloat.net> wrote:
>
> https://x.com/WietseWind/status/1992549150479516078
>
> Does anyone have probes to check?
>
> Datacenter in RIPE?
>
> I know SpaceX is present in Amsterdam but the traffic has historically gone elsewhere
>
> _______________________________________________
> Starlink mailing list -- starlink@lists.bufferbloat.net
> To unsubscribe send an email to starlink-leave@lists.bufferbloat.net