Just recently, Steve Jobs leaked the information that the PA Semi people would be designing an integrated system on a chip for the next generation iPhone.
The two best known system on a chip (SoC) implementations are IBM's cell - which puts a grid computer on a chip - and Sun's Coolthreads initiative putting an SMP machine on a chip.
Freescale's efforts to adapt the SoC ideas to PPC networking and computational devices are less widely known but equally successful and motivated by the same idea: that taking distance out of electronics reduces cost, power use, and failure rates while improving performance.
Thus Apple's rejection of Intel's CPUs for the iPhones just reflects the obvious: SPARC and PPC are years ahead of x86 on this issue - and, more importantly, Unix software is far ahead of Microsoft on making the new hardware work.
As I've said before Apple's PA Semi acquisition comes with ten year commitments to DOD PPC needs and thus the built in opportunity to reinvent the PPC based Mac -either Apple branded or via OS licensing to an existing DOD contractor like Curtis Wright.
Now I'd like to add a further prediction. It's been obvious from the first announcement that the iPhone design can be modified to accept plugin microcards - things the size of memory sticks but with processors and code of their own. Health monitoring and secured data storage and transmission are, for example, two natural applications for this kind of plug-in. So far the big constraints have been power and space: power to run plugins, and physical space for the connectors and the card housing. Moving much of the iPhone's circuitry to a SoC format will improve performance, offer power and space for the plugins, and improve the isolation of Apple's own circuitry from that provided by the plugin maker.
So plugins are inevitable, but here's a bet: the first major - meaning paradigm destroying market success - such plugin will offer full scale projected keyboards and HD capable screen projection using TI's DLP technology. By next year.