If the promised benefits of RISC architecture would have actually been realized, I don't think the processors would be fading away. Early and substantial inroads into the processor market showed that RISC is viable and worthwhile. Unfortunately, in order for RISC to maintain momentum, chip manufacturers have to stay behind it and continue development.
And this is indeed a catch. For example, IBM did not put enough emphasis on progress to improve the performance-particularly the power consumption issue-of the PowerPC processor, while Intel, having always taken abuse for how much power their chips consume, has been working 24/7 to address that very issue. At the same time, they've reduced the "process" dimension, reduced the overall die size, and INCREASED the number of transistors on their mainline chips, Their internal architecture has been tweaked to improve performance as well; their current chips outperform their earlier chips even when run at the same, lower speeds of the older chips.
One thing to think about in the discussion of Intel processors; they are not entirely "CISC." On a very low level, they are closer to RISC processors, but internally they run "microcode" which tells the core processor how to interpret the CISC instructions, making them very flexible. Don't bash this architecture too much; it's outlasted several other concepts and doesn't look like it's going to go away any time soon.