It is already more expensive to performance maintenance on SOCs than it is to replace them. Remember, these machines are not for serving a database, there are practically no storage needs (and storage is the component that fails most often.)
Given that, the main challenge is cooling, I assume that will be figured out before yeeting 100 billion $ of computers into space. Plenty of smart people work at these companies.
I meant it specifically for figuring out cooling computers in space.
I am pretty sure this is going to be a solvable problem if this is the bottleneck to achieve data centers in space, given that newer chips are much more tolerant to high temperatures.
https://www.marketplace.org/story/2026/01/07/new-ai-chips-wi...