But yeah, it’s all kind of moot. I had AWS suspend my account once because someone complained they found a copyrighted image on one of my servers. If it’s in the cloud, you don’t really control it.
Either way, though, this is going to cost five figures, annually, to store in S3. For half of one year of S3 you can have a fairly robust RAID setup, maybe replace 1 drive every two years, and enjoy the data any time you want it without fees-as-a-service.
High five for avoiding AWS!
Considering the material here, that's more likely to happen than not. So, unlikely to be "much more reliable" for this particular scenario. ;)
You'd need to consider method costs, egress, etc on top of all that, to be fair
Downloading 20TB would take me 74.074 hours @ 600 mbps.
Downloading 75TB would take 277,7 hours.
Both sound reasonable. Am I missing something?