I've helped with a couple risk registers at tech companies. Two things I've never seen appear in a risk register: The company runs out of money. Human society is wiped out. I've been laughed at once for bringing up variations on these. They're out of scope; risks stop being a threat when there's no one left to care about them.
I think the goal of keeping an internet-scale level of data accessible and searchable, for longer than one lifetime, is an impossible task. Maybe Archive.org/Archive.is can pull it off; I doubt it. Its an insane amount of data. Most of it is totally pointless, but its really difficult to pick-apart what's useful and what's useless, so you have to keep as much as possible without bias. All of that is on hard disks which violently spin around at 8 meters per second, accessed by software which we all know breaks every day but are too afraid to admit it, over a network of other computers with all the same flaws, distributed globally, yet can be significantly disrupted by one roadside construction worker and a jackhammer.
The internet didn't increase the lifetime of data; it decreased it. Sure, we have far more of it at our fingertips than any other point in history, but that's not lifetime; that's just volume. And that volume has desensitized us; its fundamentally impacting our innate biological memory capacity, and the social structures we form around memory. We know the Library of Alexandria existed because people wrote about it; the pages laid for thousands of years; its memory passed verbally from person to person.
If all computers stopped functioning tomorrow, not even disappear, they're still there, they just don't work: Would the memory of Stranger Things still be known in two thousand years? I doubt it, but: if the only thing which offers us a satisfying "Yes" is "we keep the computers running, accessible, indexable, searchable"; that seems, at the very least, given the extreme challenges we as a species will be facing over the next century, beyond the scope of human possibility