I won't argue there aren't some use cases for it, but it was massively oversold for what it actually is. It's essentially just shared spreadsheets, but almost always described in a way that make it seem inscrutable to most people (further sinking any propect of mass-adoption) and cool to people who want to feel like they're intelligent and in-the-know about tech. And it was pushed primarily for things it was unsuited for, like replacing regulated banks with FDIC insurance.

I think it has potential niche uses, though. Not every technology has to be widespread or applied to a wide variety of different things.

Tbh I think the same overselling/over-applying is happening to large language models/LLM "AI" now, though at least LLMs legitimately do have a lot of potential use cases. Just not as many as the everything people are trying to apply it to, and not as overpoweringly as many assume.