There are people working on this, but "IA preserved as it was when it was killed in 2023" is nowhere near as valuable as IA as it would be in 2033 if it survives.
IA is constantly adding new content. ("New" meaning "not-yet-archived stuff from the past century", in addition to up-to-date web snapshots.)
What people need to be working on is to create a peer/successor organization that can take a copy of its archive and carry on its core functions, not just a static archive on a server somewhere.
I'd rather have "IA preserved as it was in 2023" than no IA at all. If we have a gap in the collection between tomorrow and when someone sets up a new IA in a few years, that's very different than losing most things collected about the inception of the Internet.
The main issues are storage, centrialisation, moderation and copyright; that if you wished, rebel and ignore these with a distrbuted model like BitTorrent, or IPFS.
If an library burns to the ground you loose everything, so you will have to decentrialise. That then causes moderation issues as if you were to truely go decentrialised; what's stopping a bad actor uploading icky stuff? It would resolve the take-down from copyright, as they couldn't kill all nodes. But all sounds like a lot of work for a very little return.