In reality the cost of the wafer the CPU is produced from only represents a small portion of the total cost. Most of the cost of Intel cpu's is to recover the large sunken costs or research, development and the fabs needed to make the chips.
To illustrate my point... This article put the cost of a high end CPU wafer (300mm round wafer) at $5,000 in 2014
http://www.forbes.com/sites/jimhandy/2014/04/30/why-are-chips-so-expensive/#28f03e4a7789
A skylake cpu is 122.4mm2
A 300mm wide wafer has a surface area of 70685mm2 (area = 'pie - 3.14159' * radius [150]squared)
So 577 odd skylake 4c dies could fit on that wafer.
Now as the wafer is round so some of the dies wont fit in their entirety and some will be lost due to defects. Lets assume an unrealistically high loss of 50% from these two factors that leaves 288 odd chips.
So in my example each chip costs $17.36 each from the cost of the space on the wafer alone. A 6700k had a launch 1ku price of $350 so about 5% of the bulk price for the CPU!
To expect Intel to be selling 6700k's cheap (at or less than about £200 by the time they reach our shores) because the die is a lot smaller than previous gens is therefore somewhat ridiculous (a 14nm cpu occupies about a quarter the space of a 32nm cpu of the same sort of design)
Now in reality the cpu's from such a wafer wont all be 6700k's, subject to a 'binning' process some will be 6700's some 6600's some 6400's etc which of course sell for less money. But then the overall loss of defective chips wont be anything like 50% so its not unreasonable to assume a far lower cost than $15-20 cost per chip from a 300mm wafer representing around 5% of the details value of those chips
