Inflation is irrelevant here and the price of previous chips directly isn't relevant and this is where your comparison falls down. Chip prices are down to size, production cost then how much profit you slap on top which is what your analysis fails to take into account completely. Sandy 216mm^2, Ivy 160mm^2, Haswell 177mm^2, Broadwell, can't find it. If all the chips on a new process had a massive transistor count increase and had about the same size as their predecessor then you could make the valid comparison on cost alone.
Skylake is back down to 122mm^2. Look back and you have Lynnfield at 296mm^2 at 45nm. They dropped a node and dropped 80mm^2 off it, dropped another node and went down near enough another 80mm^2, another node and another 40mm^2 gone. They are offering MUCH less in transistor/value/production terms and each year the size of the GPU has increased which for the massive majority of users is near enough useless.
People do buy AMD APU's for gaming because even at lower graphics settings AMD has both owned Intel in performance till the last gen or two(only because AMD chose to go a period of 2 years without a real update before the new arch which will almost certainly put them ahead again in GPU performance), but even today an AMD APU offers a massively better gaming experience than an Intel APU. Graphics drivers, graphics quality, IQ is light years ahead for AMD. They have also done almost nothing industry wise to push gpu acceleration while AMD and loads of partners have worked extremely hard to push on die gpu acceleration.
So 5-6 years ago you buy a Lynnfield, it was not an APU, 100% CPU power, the next gen you were getting almost 30% less 'chip' and what 25-30% of what you did get was now GPU. Today you buy a Skylake that costs a bit more and you are getting less than half of the 'chip' you used to and around half of that is GPU which you may or may not use. Some shrinking is expected, this much isn't. You are categorically getting nothing like the value you used to get, not remotely comparable.
This is all so they can push mobile chips which they still aren't a big increase. Every shrink, every power decrease, every tiny transistor improvement and every small performance improvement is because the main and really sole development path is improving mobile. They are offering you less, their margins have increased HUGELY in the past 5 years and they are charging you more for less. They then take these chips they have been developing solely with mobile in mind, charging desktop the same and charging mobile even higher prices for the same chips.