Space-Based AI? Not Easy!

I see a lot of hoopla and speculation about why Elon Musk has “officially” changed his goal from Mars to the moon.  The answer is simple in the shorter term:  money talks! 

Musk’s companies SpaceX and Tesla are both serious government contractors.  SpaceX provides launch services to NASA and to DOD,  plus it is contracted to attempt to land humans on the moon for NASA’s Artemis program.  I am unsure whether the Tesla connection has to do with electric vehicles or the “Powerwall”,  but that does not really matter.

The point is,  much of the income for both SpaceX and Tesla come from their government contracts.  It is one thing to honestly try to fulfill a contract and fail.  It is quite another thing to default by not making the attempt. 

Musk is being paid by no one to go to Mars.  Why is this hard for anyone to understand?  He must focus SpaceX’s contracted efforts on the moon,  or else lose funding,  and worse,  all credibility as a government contractor.

There’s also been a lot of hoopla over recent Musk statements about AI data centers,  in space,  or maybe on the moon.  I see all sorts of speculation about this,  none of it based on any sort of facts. 

AI data centers involve enormous amounts of power.  All of that power gets eventually converted to waste heat,  which must be gotten rid of,  somehow.  Yes,  space is cold,  but getting rid of waste heat in space is just NOT that easy!

In space,  there is no heat loss capability due to either convection or conduction.   There is only thermal radiation to the cold background of deep space.  On the moon,  there could be conduction into the lunar surface,  but no convection,  because there is no air.  Mars is similar,  with “air” that is close to vacuum.

As for “in space”,  the cheapest destination is Earth orbit.  And from there,  the only way to shed waste heat,  is to radiate it into deep space.

I ran some numbers.  They do NOT look very good,  if one is limited to a coolant temperature compatible with cooling silicon electronics,  which is near the boiling point of water at normal atmospheric pressure.  Thermal radiation is bound by physics to be inefficient,  until the radiating surface is well above 1000 F.  The relative effects are given in the figure.

-----  

Search code DDMMYYYY format:     21022026

Search keywords:                      aerothermo,  space program

-----