Doesn't Intel do ARM foundry runs for folks?
Intel does foundry work for nobody (but themselves).
Intel sold off their ARM business to Marvell several years ago. They decided that x86 was going to attack the ARM business. At high power embedded end they have been somewhat successful (many netbooks), but still getting their butt kicked when performance/milliwatt counts most.
IBM, Samsung , and TSMC are much more likely candidates. (maybe AMD's spin off)
[ARM's list of foundary partners
http://www.arm.com/community/company_list/rw/ProductTypes/3/ ]
Apple has gotten stuff they where they weren't primarily design lead from IBM and Samsung before. IBM makes a bit more sense since they aren't a direct phone/mobile device vendor and typically have their stuff together rolling out new processes. TSMC would be a shorter supply chain (and align with Apple's penchant for putting as much manufacturing as possible out of the USA. )
Wouldn't Apple benefit from a three-way (PA semi, Intel, Imagination) conglomeration of key technologies?
No. Because it advances ARM ( not x86) so Intel is not interested.
Intel's embedded graphics gets whipped on a regular basis (by Tegra2 all the more so ). That's why they have invested in ImagTech. If can't do it themselves just buy it off the shelf. LOL.
It may be weird but given they control the software environment too, it could work.
It could work, but I could be about to go on lunch date with Halle Berry too.
Not particularly likely though.
Apple is not shooting for interoperability with other vendors here. They are shooting for "insanely great".
What? To a large extent yes they are. That's why they are buying shared IP (from ARM and from ImagTech). This way a significant fraction of the R&D costs are distributed over a larger number of devices than just what Apple makes. The differentiation will be which common components Apple combines onto the SoC package and which leaves off on other places on the circuit board(s). Apple's bundling may be a bit different and they may be able to drop certain bundles onto the market before others get there (perhaps a more expensive chip that will work better in a more expensive device with larger margins) but what is being merged is largely going to be off-the-shelf.
I'd be shocked if Apple pushes the envelope in terms of some radically "insanely great" better silicon tech. Much more likely that they are just going to take more of the profits by going more vertical on the design ( don't have to pay Samsung for design work) and have more flexibility to work on their own schedule for more wholistic system design ( drop new silicon about when new device is going to ship).
