Lee DunaSpaceX plan for 1 million orbiting AI data centers could ruin astronomy, scientists say
dylanmorganAlso it’s stupid. Heat needs a medium to conduct into, and in space there is none. These data centers would remove waste heat through radiation, which is to say by incandescing.
I’m actually starting to come around on the feasibility of space data centers. I don’t think we’re there yet, but in 5-10 years I feel like the equation might work. “Free” cooling from new lighter radiators and cooling loops, “free” solar power, no land acquisition, no building permits, no building construction, etc., is all worth something.
The physics of radiative heat transfer haven’t changed and aren’t changing, the idea will be just as stupid in 5 years as it is right now.
But it doesn’t matter, because there isn’t a plan to actually build a bunch of orbiting data centers, it is just something Musk is saying to generate hype for investors pre-IPO.
I agree that this announcement is just another Musk hype cycle, and I can’t wait for a bunch of “AI” companies to crash and burn.
What I meant with the 5-10 year comment was that if people still insist on spending money on data centers, we might reach a point where falling launch costs, lighter radiator systems, and more available and cheap satellite components make it feasible for a space based server rack to have lower lifetime costs and be less of a headache than a terrestrial one.
A physics problem might be easier than a regulatory one. Big radiators are an acceptable solution if there are multiple reusable super heavy lift launch vehicles.