In close proximity, and connected to metal structures radiating & conducting heat at 120 degrees. You still have to expel that heat somehow, and route it around the spacecraft core.
I think stopping the heat conductivity is a solved problem - space telescopes manage to do this. But all in all, cooling in space remains a very big challenge.
No, it isn’t solved. The James Webb telescope was shipped with cryogenic coolant, and put in an orbit in the shadow of the moon far from the radiating earth. Once it runs out in a few years, it’s done.
James Webb, being an infrared telescope, has to be extremely cold: It's Mid-Infrared Instrument is actively cooled to 7°K (-266°C).
Most other instruments (NIRCam, NIRSpec, FGS/NIRISS) operate at about 40 Kelvin (-233°C) using only passive cooling (i.e. the sunshield).
Clearly this would be cold enough for a data center. Clearly the sunshield can be very well thermally insulated from the rest of the telescopy. The sun-facing side is up to 300°C hot. The cold side is -233°C.
I'm not saying radiating the heat from the GPUs is a solved problem. Just the thermal insulation between the solar panels and the rest of the spacecraft.
There's very little on JWST that makes it hot. Furthermore, the spacecraft bus where the non-scientific instruments exist) is on the other side of the sunshade so that the waste heat from the RAD750 (a single processor clocked at 118 MHz - PowerPC 750 architecture (Macintosh computers in '97) that uses 5 watts of power designed to operate -55°C to 125°C) doesn't interfere with the scientific instruments.
Putting an AI rack there is a completely different scale of power and cooling than what JWST uses.
Yes. Radiating heat away is a big problem - I totally agree! But stopping heat conduction is a solved problem! Consider the temperature of your pot handles. That‘s all I was saying.
The energy collected from the solar panels must be converted into heat in the AI chips. It's really like putting the AI chips directly into the sun, just with extra steps. Sunlight gets transformed into electricity which gets transformed into heat in the chips.
So the satelite receives say 13kW of heat energy as it has a 10 square metre surface facing the sun.
That is converted to 2.5kW electic and 10.5kW heat. The electric is then used to power the computer and ply doom or whatever. That 2.5kW of electric is then converted to 2.5kW of heat. That means the satelite has to dissipate 13kW of heat.
But even without the electric use it still receives 13kW of heat and dissipates 13kW of heat.
Any object in space will emit as much energy via radiation as it receives otherwise it will continue to increase in temperature. The question is thus what temperature does it sit at to make this in equilibrium.
If the satelite was perfectly flat it would run at 120c for a black body. If it was a lump of stew it would be about 150c.
To reduce that to 20C you need about 50sqm radistor, or 5 times the surface area in shadow. The shadow is about 350-400m long
So you build it as a cone with the flat circular area facing the sun and the pointy area in its shadow you’d only need a cone height of about 20m to emit enough heat to keep the satelite to room temperature.
They actually need the entirety of the backside of the panels to cool them - if not they would literally burn out from the accumulated heat from being exposed to the sun.
