Thanks Jeff:

I realize that I've got a double thread going now, and I just had a light bulb go on when I looked at elapsed time for writing to disk, as opposed to elapsed time for displaying data.

Essentially, when displaying data I was seeing elapsed times for simple display being double that of concatenation and I was attributing the bottleneck to the SQL Server. That made me want to stick with concatenation because it was taking 11 seconds to run as opposed to 24. But when I wrote the results to disk (on my laptop), I saw elapsed times similar to yours. So I hit the reset button in my brain.

Bob