... for me Canon has a HUGE profit margin on their full frame 1D's ...
In short, those big 24x36mm sensors are expensive to make. They are about three or more time larger than even the biggest, fanciest, most expensive CPU chips and such, and that unusually large size makes their fabrication rather expensive.
Kodak seemed to cut prices desperately on its Film Format (24x36mm) DSLRs before giving up, getting the price down to US$3,500 after rebate. And those cameras had no low-pass (AA) filter, which apparently saved Kodak some hundreds of dollars on the manufacturing cost. And surprisinlgly, those DSLR aparently sold more units than the 1Ds series (but at lower revenues and profit margins, I am sure), so Kodak had better economies of scale than the 1Ds, but less good that Canon is planning for the 5D.
Since I severely doubt that Kodak was making a high profit margin on those cameras at $3,500, I similarly doubt that Canon is planning to make big margins on the 5D at $3,300.
From what I have read from sources such as Thom Hogan, it is quite likely that the planned 5D production levels will be about enough to occupy a chip fabrication line full time at full capacity, maxing out one of the main cost advantages of higher volumes: optimal utilization of those very expensive fab. lines.
On the other hand, Canon's stated 1Ds production levels of 2,000/month are maybe only about one quarter of full fab. line capacity for a chip of that size, so there is potentially a significant cost of underutilization of fab. lines, raising unit costs.
I imagine that Canon hopes eventualy to use its several 24x36mm products to maximize fab. utilization across the board. It should help if they move to using a single sensor for the "EOS-1 D" product line, instead of the current two separate ones for the 1D and 1Ds.