It's not generational. I don't think developers these days are any lazier or less gifted than developers at any other times. We're at a point in history where it's much, much cheaper at a certain point to double the number of computers you have than it is to make all of your computers twice as fast. SQL databases, especially the ones we have right now, were built with the idea that you'd have one large database server -- replication and such exists but it's an extension of the single-server model. Things like NoSQL databases and Hadoop are meant to provide a model that supports horizontal scaling better than existing relational databases do. As they mature, they are starting to take on more and more features from relational databases. And SQL databases are taking on more and more features that let them scale horizontally. I suspect at some point in the future we'll see a lot more convergence.
I didn't mean to imply that the new generation of devs are lazy and stupid. They (and you apparently) are influenced by the perception that RDBMSs are relics not designed with modern hardware in mind. This perception is wrong. Vertica and VoltDB were both designed from the beginning to be distributed. SQL Server and Oracle were not designed that way in the beginning but incorporated those capabilities over time. All of these are incredibly sophisticated, modern, high performance RDBMSs. None of them can be used effectively without some effort to acquire relational concepts, modeling, SQL, etc. This is where the "database community" has lost ground to the prevailing NoSQL solutions IMO, by not making it easier for new devs to plug and play.
One example of the sort of thing that could make RDBMSs more plug-and-play is Vertica's flex tables - a schema-free, super flexible blob of a table that lets you move forward without having to completely define your relational model.
