Jeez I hate sounding like such a fanboy but here goes. They switched because IBM's only option they provided Apple for the future of their laptop line was just straight up embedding G5s into PowerBooks. This wasn't really feasible at all, they were already running dangerously hot and had abysmal battery life. PPC chips were just nowhere near energy-efficient enough. It really had nothing to do with benchmarks or power or whatever (in fact the initial Core Duo chips were much slower than their PPC counterparts).

I mean think about it, Apple makes next to no money on desktops, why would they prioritize power over energy consumption? Every once in a while they'll trump up some weird graph/metric like "24X FASTER" when talking about new chips in their devices, yet just about every keynote they talk up (in concrete numbers) how much longer batteries last. Hell, the latest release of OS X's biggest feature is enhanced energy management.

This is one of the biggest things that makes comparing Macs to PCs kind of silly to me—if you want raw power, you should absolutely go with a Windows box. Apple loses on that every single time, and they are fine with that.

The PowerPC as a processor really was a powerful chip. And it was arguably more powerful than contemporary Intel chips. It's hard to directly compare the two because to unlock the performance for either architecture requires an optimized toolchain. I'm not sure that the toolchain Apple was using was ever quite up to par with Intel. But since they are difference architecture styles (RISC vs CISC), it was really hard to get a good comparison. On HPC workloads, the Power chips were really good. I'm not sure that ever translated to workstations well.

The differences between the PS3 and Xbox 360 processors and toolchains is probably a good analogy. While in theory the Cell of the PS3 is more powerful, it was easier to get the Xbox to work optimally. (Although, ironically, they were both PowerPC based CPUs).

But the reason why Apple changed over to Intel chips had nothing to do with raw performance. It had everything to do with power - as in watts. Apple was beholden to the PowerPC manufacturers (Motorola/IBM) for their chips.

The major markets for PPC chips was either at the very high-end (HPC) or the very low-end (embedded). While Apple was the main customer for the mid-range chips, they didn't have much relative volume, so they didn't have much sway in the chip roadmap. Because of this, it was difficult to get high-performance/low-power chips. Apple went for years with a G4-based Powerbook because they couldn't get a G5 version that would work in a laptop - it required too much power.

The reason why they switched to Intel was that Intel was already working on the high-performance/low-power chips that Motorola just couldn't offer.

The PowerPC architecture itself was very high-performance, but recent years have shown that raw performance isn't as important as performance per watt for consumer use. If you're building a super computer, that's another story. From a pure CPU architecture standpoint, I'm still a fan of RISC.