I agree that APUs will take over the entire low-end market. Probably the middle of the market as well, at least after enough years go by to make it possible. Where I'm not so sure about is the high-end. It's entirely possible the high-end market could die out, especially if Windows gaming is relegated entirely to console ports. But I think it's large enough to survive.
Are CPU manufacturers really going to want to stick 50 billion extra transistors on their CPUs that are much more complicated (and likely to fail, therefore wasting the entire chip)? Or will they stick to a simpler, cheaper chip good enough for 95% of people that will give them higher yields and tell consumers to buy $600 graphics cards for those who really need the extra power? I think this is an open question, and probably something that not even AMD or Intel has figured out yet, or will even attempt to figure out for many years to come.
Also, your assumption that you can just plug in extra APUs for more power doesn't seem like a good solution to me - look at crossfire and SLI - even with 2 GPUs it doesn't always scale very well. Stick in 4 GPUs and watch scaling go way down. Just being APUs won't fix the scaling problem, at least not all the way.