I agree, the bureau route is probably best right now, and I reckon will be so for some time to come. Better put your effort into the CAD modelling, and let someone else pay the telephone numbers required to buy and maintain a really good machine, than spend your cash on a device which will intrigue your for a few weeks, then irritate they heck out of you, when you discover what it really can, and can't, do. Remember, if it sounds too good to be true, it probably is.
I just don't see the kind of 'every household will have one' kind of demand for this sort of capability. I've been in the Rapid Prototyping industry for over a dozen years, and nothing would please me more than to have a machine that produced strong, smooth, accurate parts in familiar materials (ABS is OK) that don't warp, break, need sanding smooth, take paint, glue, costs peanuts, and comes from a machine that doesn't waste most of the material you put into it, doesn't need a technician to repair/calibrate it every month, doesn't spit out cartridges for no good reason and then the supplier refuses to credit you for them, doesn't need vacuuming out twice a week... I could go on!!
It will come. But at a price. And whether it will then be commercially successful, ie if they volume of sales of machines and materials make sense for the manufacturers, I have my doubts.
Then again, maybe my slightly jaundiced views are because my experiences are from the pioneer days of 3D RP, my hat full of arrows.
Anyway, however you 'make' the final model, your CAD skills need to be up to scratch. Anyone want to compare notes on Rhino 3D, drop me a line! Heres one I did earlier...

Tim