Aside from what other have pointed out, one of the main reasons id entry barriers. Its hugely expensive to develop a competitive x86 cpu; hundreds of millions, if not billions are involved. it basically requires having your own state of the art fab(s) as well, which costs a few billion by itselve.
In the past, many companies have tried; centaur, cyrix, UMC, AMD, even IBM. competition and huge r&d costs made that none except AMD survived, and what remained of centaur and cyrix got purchased by VIA. Recently, only Transmeta tried with a radical different approach, but unfortunately, I doubt they will be with us at the end of the decade.
Other attempts have been made to replace the desktop hegemony of x86. There have been (windows supported) PowerPC systems, Alpha's, even MIPS "desktops", but none ever made a splash. Backwards compatibility and software availablitity where just too important. Only Apple sort of kept lingering around in its own (shrinking) niche.
As it is, I don't see anything replacing x86 for the next 10-15 years. After that, maybe power or cell, maybe not. I don't see any company with the resources and the will to bring out an x86 chip besides Intel, AMD or VIA either. Maybe IBM could, but they got burnt trying before, so I don't think they will again. They gave up on the desktop market long time ago. Well thinking about, there might be one other candidate: microsoft.
= The views stated herein are my personal views, and not necessarily the views of my wife. =