The fastest i7 and i9 should be used for an entry level model for those who want a tower design with the option for out of pocket expandability.
The i9 and Xeon W prices aren't all that far apart ( used to be almost none for previous "Extreme" versions and Xeon E5 1600 class ). It isn't going to dramatically cut costs.
i9 7900X $999
https://ark.intel.com/products/1236...-series-Processor-13_75M-Cache-up-to-4_30-GHz
W 2155 $1440
https://ark.intel.com/products/125042/Intel-Xeon--W-2155-Processor-13_75M-Cache-3_30-GHz
With the i9 you loose x4 PCI-e v3 lanes and top end memory capacity. Loosing uniformity of the CPU PCI-e lane count probably isn't cost performance for the line up at all (i.e., different motherboards for different versions gets you lower economies of scale. ). Since memory cspacity is a major part of the cost differential going to 8 banks of DIMMs would be a better value add for Apple.
If looking to make a more reasonably price entry model than the iMac Pro entry characteristics present Apple could easily go with either
Intel W 2125 $444 ( about what previous Mac Pro entry level processors started at )
https://ark.intel.com/products/126708/Intel-Xeon--W-2125-Processor-8_25M-Cache-4_00-GHz
or raise the floor slightly and start with a W 2135 $617
https://ark.intel.com/products/125040/Intel-Xeon--W-2133-Processor-8_25M-Cache-3_60-GHz
The 2125 will for folks who are primarily "drag racing" older single threaded apps and aren't particularly workload growth bound (i.e., their old Mac Pro was largley good enough but has dropped on obsolete list so time to upgrade. ). The 2135 has the 6 cores ( but not the base clocks) is core count spin has deeper traction with entry level market.
The mainstream iMac is likely to jump to 6 cores sometime in 2018 so it makes some sense to start the iMac Pro off at 8 cores. However, there is no good rationalization to start the Mac Pro out at 8 at all. It is not an iMac so it will have other significant characteristics to do differentiation with. ( it doesn't simply need to be be processor is better because the screen and about every other external characteristic is the same. )
Same with baseline of 32GB of RAM ( could be 4 x 4GB --> 16GB ), A 512GB ( or smaller ) SSD (especially if there is an option for a 3.5/2.5 HDD for raw capacity usage and/or a M.2 standard slot. ). Apple could have an entry level GPU that isn't HBM2 based (e.g., RX 570 or 580 ). All total that is probably around another $1,000 discount of the iMac Pro base price.
Not everyone wants or needs a Xeon and the Core i series would open up Apple to mainstream crowd such as gamers; users who want desktops that are not an all in one; or just a really nice tower. Pricing should be around 1,700 to 2,300, minimum 16 GBs DDR4, max out 128 GBs RAM. The Xeon's could go up to 512 GBs and 1 TB of RAM for those who can afford it. But it should come with 64 GB standard.
Apple isn't chasing the hard core gamer crowd with Macs. Apple makes money off of iOS games, they don't need Mac games to drive revenue. Mac gaming isn't going to move the needle against the vastly larger iOS ecosystem.
An i9 isn't going to get you into the sub $2,300 zone with an Apple system. Just not; you can just look at pricing of other $800+ CPUs that Apple has previous done in the Mac Pro class.
Apple is highly unlikely to build an xMac relatively high fratricide rate for the current iMac price zone. There are relatively high enough priced, sunk cost PCI-e cards that can drive significant differentiation between the iMac Pro and a revised Mac Pro (presuming they put a PCI-e socket in the updated Mac Pro). They can probably share an overlapping price zone with relatively low fratricide.
Apple's fear from this is how making a system thats expandable gonna help the bottom line; especially in this age where hardware has significantly outpaced software. Many users are likely to keep a tower for 5 to 8 years. Apple doesn't want that.
funny how Apple's vintage and obsolete policy has generally run 5-7 years. (and of late around 8 years for the Mac Pro due to the super long upgrade cycles. ).
They would rather have you upgrade every 2 years, 3 years max.
Folks on a faster than normal depreciation cycle??? It is more the extremely dubious Apple was, is, or will be counting on that.
A tower with expandability will definitely affect that.
This has been the road to the xMac. There are significant fraction of folks who just buy the trailing edge. Series of used Mac Pro , upgraded with "cheap" CPU , RAM , etc also from the used market. For the most part these folks don't buy from Apple so Apple puts relatively low priority on making something for them.
Apple could tap into the expansion market by providing authorized approved components: CPU, GPU, RAM and use the Apple stores as incentive to have them do it for you or even an authorized Apple reseller and make a profit.
But if this was so amazingly lucrative why where there no Mac Pro upgrades for a long time after 2010?
If ignore inventory costs and other issues this sound great in the abstract. It doesn't 'save" Dell, HP, etc from having substantially lower margins than Apple does.