In addition to the NO-LOCAL-STATE rant:
Data archival doesn't (although it could and, to an extent, likely, will) need to be solved GLOBALLY.
Like most problems in life, data archival needs to be solved at the margins:
i.e. I don't care about the GLOBAL availability of all data; I care about MY segment of the data I'm interested in (or might be interested in, which is an interesting problem), and at the margin I am the one who will make sure that the data I stored in nostr is sufficiently persisted.
These are all relatively simple problems to solve that, to an extent are already solved, and, I expect, will be fully solved by multiple infrastructure offerings, given how cheap storage is.
TL;DR:
We don't need to solve data archival GLOBALLY by one or two providers; individual users need to be solve data archival MARGINALLY, for the data they care about. (this will, obviously, be fully managed for 99% users)