I have a feeling that this part hype, part inept programmers who don't actually understand SQL, or database optimization.
This is part of the problem... similar to PHP, most people learn some examples that teach some bad habits right off the bat (sticking SQL in your view, etc) because it's so easy to get started, but you've gotta get a grasp on the tech before you can do anything big.
Also, I feel that one of the root causes of the hype is that SQL and RDBMSs in general don't solve all your problems and sometimes get in the way of your application design. Between rigid schema definitions and the SQL language that has a bit of a learning curve when you start dealing with nested queries and handling shards/partitions/etc, I think that's the reason we're starting to see more non-RDBMS databases.
At work, we had a project that we started building on MySQL, but was falling short because we were constantly making schema changes. We begun to build a system where we could have arbitrary attributes attached to arbitrary objects, but then our queries were getting REALLY nasty. We discovered MarkLogic which is an XML database server and uses XQuery to query the data. We were ingesting around 100MB of XML a day, and we needed to be able to handle just about any XML that went into the system. MarkLogic was a natural fit since we needed to put XML in and we wanted XML out most of the time.
We're still using MySQL for tracking the ingestions and managing the frontend to the system (which is built on Rails), but having XQuery at our fingertips has been a godsend.
There's a lot to be said about new technologies that solve needs and get around shortcomings of the more ubiquitous technologies, but, as with anything that people see as a solution, it's not a silver bullet. You've gotta be careful not to get trapped in "everything looks like a nail" syndrome.