I am increasingly finding that the complexity of the environment requires the need for a DBA with experience and knowledge.

A simple onesie twosie instance could be potentially handled by a dev guy, or someone without much in the way of experience, but once you get beyond that, have to deal with replication, log shipping, mirroring, monitoring, change management, troubleshooting, performance tuning etc... then that lack of knowledge and time with the product will leave people dead in the water.

What is potentially a scarier prospect are the folks out there who firmly believe that they are DBAs. I interviewed a gent a couple of years back, really nice guy, had been working as a contractor for the last 10 years, and was working on a lot of projects, but wanted a full-time gig. He interviewed well, called him back for a second interview, put him down in front of a computer and asked him to do some simple tasks like update a table based on a result set from a join of two tables. After many minutes of playing around with enterprise manager he couldn't do it.

He was the best candidate.

So long as there is the product there will be the need for DBAs to manage it.

Right now it really appears as though focusing on one facet of SQL Server might be the best route, just because there is so much more to it now, and with the fabric up and coming it adds another dimension (just as policy based management does right now). After all, how many people are there in your organization managing Active Directory, and what expertise do they have?