Too many reasons to count.  Firstly the relay is in a state of willful neglect, so users see problems all the time.  Secondly relays by their very nature are not designed to *store* notes, just to *relay* them.  They are GOOD at relaying notes, but not good at storing notes.  Relaying is more resource intensive than storing, because you have to be always on.  Added to that you need a caching layer to scale, which nostr didnt start with, and still hasnt developed.  Also nostr never had an architecture, it was just a quick thing put together that just about worked, it was never intended to scale, it was intended as a proof of concept.  But the critical thing, is that for scaling you need a uniform interface ie scaling architecture, and it has to be pretty good.  Getting people to agree on such a thing is very hard, especially with the current state of NIP centralization.  Best that we can hope for is that someone will invest in hardware, but we've still not seen that yet, despite large donations.  tl;dr nostr still has a chance of scaling, but it would take some work and some luck.  That work is not being done, so, we would need a game changing event.