Ask HN: Does Starship mean that data centers will soon be in orbit? | Hacker News Reader

Ask HN: Does Starship mean that data centers will soon be in orbit?

3 pointschrischapman8 comments

So, I've just been reading this HN link (https://news.ycombinator.com/item?id=38227463) and it got me wondering about near-term profitable use cases for Starship.

Would it be possible to move data centers into space? Would it be cost effective? It would be better for the planet from a climate change point of view but how practical is it? How many servers could you fit on board Startship? Maybe it would be better to send the raw material (silicon/graphene whatever) into orbit and do the manufacture there. How cool would it be to buy Raspberry Pi's made in orbit. Most manufacturing is already done by robots. Why not in space? I guess if Musk is fully focused on going to Mars, shipping servers into orbit will be the last thing on his mind but what if Amazon or the NYSE moved their servers into orbit? Would everyone follow?

If getting into orbit is no longer the key technical challenge, whats the next biggest challenge to putting data centers in space? All communication between the data centers and earth (as well as each other) would be laser/light speed. That's got to be a solved problem, right? But what about heating, cooling, solar power? Any guesses as to how long it will take for this to happen?

Just thought I'd ask.

Cooling is a bigger issue in space than it is on earth. The launches themselves have an environmental impact, so doing things in space doesn’t suddenly make it “eco free.” There are other things that should go in space first, like permanent stations and docks, before data centers.

>Cooling is a bigger issue in space

Is this due to the raw sunlight? If so, wouldn't using a sunshield immediately reduce the heat to far below zero? Or is circulation the problem?

You only have radiative (heat moving via electromagnetic emission aka photons) cooling in space. On Earth you also have convective (heat moving via moving fluid) and conductive (heat moving via permeating a material) cooling to move heat away from a source. Radiative cooling is by far the worst way to move heat.

The space station has a massive cooling system, https://www.space.com/21059-space-station-cooling-system-exp.... If you were to be ejected into a vacuum, you would first heat up from stored chemical energy in your body being released, and then cool over a very long period to a very cold temperature, not instantly freeze solid.

Something has to be very hot for you to feel it via electromagnetic emission (emitting in IR), and it will always transfer more energy by touch. If something is white hot, then it is emitting in the visible spectrum. But in an atmosphere, only a small amount of total energy emission is being released as photons, much more is released as kinetic energy to the surrounding air. https://en.wikipedia.org/wiki/Black-body_radiation

No, unfortunately there are so many reasons this wouldn't work. Just a few

- $3 million per 150 tons is still massively expensive.

- Mass rocket launches are horrendous for the climate because of the resources and energy they require (not to speak of the huge mass of hydrogen/methane for every ton).

- Servers certainly can't be manufactured in space.

- Housing humans to maintain them is too expensive.

- Power in space is expensive.

- Radiating heat away in space is very expensive, and a luxury we take for granted on Earth.

- Shipping parts to space is slow and expensive.

- Modern computers don't like all the radiation in space.

I could go on and on, but basically, doing anything in space is hilariously expensive, and theres no reason to do it unless you absolutely have to be in space for it, like space telescopes, probes, comms or zero gravity experiments. Go play Kerbal Space Program for a bit, and it will expunge whatever dreams of mass space industry you have from your brain :P

EDIT: Now, there is a case for sending small amounts of equipment for scientists to use, or maybe to process data before transmission. HP servers have already been sent to space, see https://www.datacenterknowledge.com/hardware/space-final-fro....

>Radiating heat away

Because no air molecules to carry the heat?

There is not a lot of benefits to it besides it being awesome. Maybe some HFT firms could benefit from the low latencies in orbit.

Why do you think latencies would be low?

https://en.wikipedia.org/wiki/High-frequency_trading#History:

“At the turn of the 21st century, HFT trades had an execution time of several seconds, whereas by 2010 this had decreased to milli- and even microseconds”

At the speed of light, a millisecond gives you 300 km, so even assuming zero switching and processing time, you’d have to put your data center in an orbit lower than 150km. That’s less than halve the height at which Starlink operates.

I don’t know how long a satellite lasts without maneuvering in space at that height, but suspect it will require lots of fuel to increase the orbit, or will be problematic to fill and backup such a data center before it burns up on reentry.

Also, of course, a data center in such an orbit almost never will be directly overhead. At times it will be at the other side of the planet, adding over 100ms.

No.

Reply on news.ycombinator.com