Buying ARM to stop its cores getting to competitors doesn't seem a good idea. Other chip manufacturers would rush to fill that vacuum. Freescale and the other Power ISA guys would love that kind of boost. I don't see it slowing down Android significantly. GCC will recompile to whatever other platform is chosen; there will be some porting effort but it isn't huge.

If Apple is really doing this I see it a play to take on Intel. The ARM instruction set is clean and well-designed, unlike Intel's. This enables ARM chips to draw significantly less power for the same performance. Now, who cares about performance-per-Watt? Anyone who has a cluster does. Anyone building devices that run on a battery does. These two groups are becoming everyone -- it's laptops/tablets and cloud computing all the way, baby. If ARM gets a 64-bit implementation and a fast floating point unit it can compete against Intel in the server market. Apple can fund that with their giant pile of cash. ARM already has the low end market sewn up. Imagine a laptop with 10 hours battery life and Google buying a million 64-bit ARM cores. That's two nice revenue streams to have, and it gets Apple into a big market where they currently have very little presence.

I did a blog post about some of these issues here: http://www.untyped.com/untyping/2010/02/02/is-the-ipad-the-b...