Good question. I've been thinking about this myself.

I think the Makerbots, Solidoodles, Form 1s of the world will shortly begin to eat into the low end models from the big players, for example, in situations where a company wants to give each engineer a 3D printer.  But I don't see that happening in the commercial and industrial area, which will always find economic reasons to have the most precise, fastest and versatile printers backed by a support team that responds quickly to their needs, whether it's for prototyping, specialized printing like metals or medical, or actual manufacturing line printing.

It seems like 3D printers are somewhere similar to where computers where in the late 1980's. The personal computer market started advancing quickly, but so did mainframe computers and higher end microcomputers. Even through the 90's you still saw big iron computers still rapidly selling into commercial use, all the while as PC also improved dramatically. As powerful (in those days) as PCs were, the bigger commercial systems were similarly more powerful.  

I do remember in 1995, I chose to sell an IBM mainframe-related business I owned because I saw how PC's were beginning to edge in on the mainframe -- thought I was at the top of the bell curve. And it was true, as networked PCs began to supplant mainframes.  

I think we've got quite a ways to go before the low-cost desktop 3D printers take share from the big systems.