I loved when we would have an HPC server crash and we would analyze the issue and the vendor would respond, "probably a neutrino strike". I can't imagine this being more complex, given all the constraints of being in an orbit. The hardware depreciation, and the need to eliminate charge buildup and excess heat. Placing servers in the Arctic seems a better solution, at least the cooling issue is resolved.