Hmmm, even then I think we were still very 68060 oriented... No one really took the PPC seriously as an Amiga CPU solution until Phase5 (who built Mac CPU cards) suggested it... And then did it 
Allow me to elaborate.
Back in the day, we were all hailing the 68060 as the holy grail, but let's not forget that it actually arrived late and with lower performance compared to the Pentium (particularly in floating point terms). It was clear to everybody back then that the 68K was falling behind. When apple introduced the first PowerMacs, people started talking about the possibility of making the same jump (the PA RISC idea pretty much died with Commodore). Amongst those people were Phase5.
The PowerPC was seen as an obvious choice back then as RISC was still seen as the future (which indeed it was, but just not in the way it was envisaged), it shared the same native endian representation as 68K, supported the same basic datatypes (except for long double). The x86 was still looked down upon (IMNSHO deservedly so as wasn't until the Pentium Pro that the architecture became any good) and was seen as too slow by most people in the day to provide any realistic emulation potential. Remember, JIT emulation of 68K didn't have an obvious precedent (I think even Apple's first 68K emulation for PPC was interpretive) and so things like register count mattered. You could statically map the entire 68K usermode registers into PPC and a well written assembler interpreter could fit the most common handlers into cache.
Not long later, Phase5 announced their intentions to produce a PowerPC accelerator for Amiga. Back then, the talk was of 68030 + 603e. The rest is history.
