It seems clear from the fact that Apple put the first of (what are likely limited supplies of) the merom CPUs into the iMac, rather than the much-anticipated MBPs, that the iMac enclosure simply can't dissipate heat quickly enough to run the higher TDP CPUs like Conroe without unacceptable compromises (i.e. loud fans).
So I predict these new, mid-to-high performance CPUs will either not be used by Apple (bad move, IMO) or they will appear in a new product. The long awaited xMac. The problem that Apple needs to chart a course around with this product is that it can't be so powerful that it eats up sales of the MacPro (that's easy, make it cheap), and it can't be a direct competitor with either the iMac or the mini (shouldn't be too hard, drop the price on the mini another $100, making it a cheap, 'Value-priced' system with very limited upgrade potential, and continue to produce the iMac as a beautifully designed AIO system for the office, lab, etc.). I do think the xMac will cannibalize iMac sales to some degree, but it should be manageable.
The problem with the xMac as a product for Apple is two fold. Firstly, it has to be agressively priced, because, of all the Macs, it's the one that will be facing the most head-to-head competition from other vendors, and it will have the fewest Apple-only features to justify significant price differences. Secondly, it will have to be easily expandable to be competitive, and consequently, it will suffer from 3rd-party hardware and software quality issues.
To deal with these issues, I think Apple needs to pull a rabbit out of it's hat WRT the industrial design of the xMac, making it an elegant, easy-to-work with, highly prized piece of technology that people won't mind paying a little extra for. And they need to be very explicit about using only 'Apple-certified' components or the warranty is void. This won't prevent people from using every standard PC widget under the sun, but it will give Apple an out when some of these systems fail.
Cheers