I have been looking at laser cutting as well. My brother is a CNC machinist who has a custom CNC laser cutting company (
http://www.skog.ca), so I have ready access to a machine for trials. There are also a couple of local online companies who will mail results back to you when complete. In general the "costs" of laser cutting are in proportion to not just the cut length but also the cut depth. If the company is using a non-CO2 laser (i.e.: solid state or one of the pro-level diode lasers) then the burn time of the laser head is amortized into many many cuts, and therefore run time is the deciding factor, which is a function of length and depth of cut (deeper cuts are slower). If they are using a low-cost CO2 laser, the burn time of the laser head also has to be factored in (they have a much shorter working life and are not cheap).
For many materials like thin metal, thin plastic and thicker metals, plastics and ceramics, I usually send items for CNC waterjet cutting. In thin materials like what we as modellers would typically use for underframes, waterjet cutting of metal frames would be inexpensive and fast. I get aluminum, stainless, carbon steel and plastic cut by waterjet all the time for work. In materials under 1/4", the taper of the cut is negligible.
I have a couple of friends who run CNC and manual machine shops (and I am a manual machinist as well as an engineer), and it is the setup times that kill you. Great if you have a distribution chain set set up to sell the large box of parts you just got done in order to amortize the costs, not so great if you need a single item. I have looked at having a CNC shop run off machined wheel rims for use with 3D printed plastic centers and the realistic number for production would be 1000 units. Anything less than that isn't worth the setup charges. I just cannot find a use for 500 axles each of 10", 12", 14", 16", 18" and so on wheel sizes...or afford the costs involved in those numbers. Getting the axles machined is just as expensive. Few CNC shops will handle that small a material. For that matter few CNC shops have the required double-head lathe (both the shops I work with have them...but they are precision aircraft sub-contractors so they are unusually well equipped).
As for Shapeways metals, they use two methods. You are correct that the "Precious Metals" are 3D printed in wax and then investment cast...so if it will print in the castable wax, it will investment cast in precious metal...theoretically. The steels, however, are laser sintered as a single process, and the porous material structure is then infused with bronze...about 60% 420 Stainless Steel/40% Bronze. It is magnetic, but not as much as regular steel...I still have to test to see if it will have any impact, but I suspect it will be no worse than older plastic HO/OO freight cars which used a stamped steel plate as their body weight. I am thinking of making a steel underframe with a FUD or WSF cosmetic exterior (customer buys 2-3 pieces)...this gives a very strong underbody with significant mass (better running) while giving the external detail desired. The third piece would be optional - the decking, which could just as easily be hand-cut wooden planks.
If I was going to set up my own distribution channel, I'd have the support frames waterjet cut and purchase a FORM1+ printer to make my own body cosmetic exteriors and detail parts with laser cut wooden parts. Unfortunately I haven't won the lottery yet!

Until then, FUD has worked for me for some parts and it seems that if the part has sufficient structure and is not under stress it is adequate. WSF seems nice for cars (I am trying several of Tebee's designs to save the time redesigning mine to work with the WSF requirements), and I agree the textured finish seems to work for larger scales, but I am not entirely satisfied with it in 1:32 and not at all in 1:43.5. It seems OK for cars which are intended to be rusty and cruddy, but the texture is wrong for cars with planking or where it should be smooth paint. I have tried some of the other materials and will be trying polished WSF for another project shortly. I am also going to try to source some "Mr Surfacer" and try to smooth out the WSF surface with that.
Despite the occasional frustration or setback, it is quite interesting to explore the limits of this technology.
Thanks for all your comments and help.