I think there's other NIPs in the process of sorting out storage payments and such, but as far as I know, they're not trustless. Carbonado is probabilistically trustless, and it can do this without a blockchain. It uses stream verification encoding to periodically ask for short pieces of a file, 1KB each, in multiple random places that can't be known in advance. I can then verify that along with a checksum against a 32 byte Blake3 merklehash. If we chunk data into 32MB segments, only 1MB of hashes are needed to secure 1TB of storage. Also, the Nostr devs seem pretty adamant about sticking to SHA-256, which is pretty slow these days without hardware acceleration. And finally, the storage providers are paid in sats for the value they provide, and there's even geocoding metadata added as well for georedundancy: https://github.com/diba-io/carbonado/issues/13

As for your original point, you might appreciate some of what I said in this response: #190465

Maybe it sounds farfetched, but it's good to at least build the technology. Who knows if it will catch on or be successful, but it might be useful to be prepared just in case... Bitcoiners are like the doomsday preppers and gun nuts of economics.

And regardless, in the very least, it's good to satisfy the demand for shenanigans without having to tell people to buy something other than Bitcoin. Ideally it wouldn't be through some wasteful, naive, and risky implementation like BRC-20, either.