When I paid a visit to my component supplier I spotted a few GAL22V10 available there. Asked the price - CNY24 (about US$4) a pop, NOS.
Then I asked about EPM240T100C5, an 100-pin 240-LU CPLD and the price? CNY6 (about US$1) each, brand new. Even the bigger variant EPM570T100C5 costs about just $3.
Why? Why an immensely more powerful chip is sold at 1/4 of the price of a significantly older, way less powerful one?
Yes, of course, supply and demand. But let me make a few comments on manufacturer costs. Newer IC's are typically built on newer fabs which can fit more transistors in a smaller area. Because of this, they can get many more IC's from a given amount of silicon. So the manufacturer cost is higher on older IC's. This is very significant. In competitive markets, high volume customer pricing asymptotically approaches cost of materials + some thin margin.
In addition, many manufacturers don't really like to keep old "legacy" fabs up and running. They only have a small number of customers buying the output of those fabs. So the price goes up. Those customers don't like paying the high price, so they eventually find a way to design in newer parts. This means that the number of customers shrinks even more, forcing the supplier to raise prices even more to make it viable. It is kind of a negative spiral. Eventually, it will simply not be worthwhile, and the manufacturer will announce an end of life date and a last time buy date.
The types of customers who buy "legacy" parts despite very high pricing are those who simply cannot modify the design. Nuclear power plants. Aviation. Some types of industrial equipment. If you have an electronics component that has been tested extensively over a very long time, running in a mission critical application, the cost of validating a new component may be many millions of dollars. Long before it finally retired, the space shuttle was shockingly obsolete.
Supply and demand. The more popular something is, the greater the market and economy of scale, therefore the more cheaply it can be made. This is especially true for integrated circuits since most of the cost is in development, equipment setup, packaging, and testing - not the chips themselves.
Older chips tend to be more expensive because they were produced when the technology was new and costs were higher. Without demand there is no incentive to use more modern (cheaper) processes, so the price remains high.
Those few customers who need them are willing to pay the higher price because newer chips won't work in their applications. Some older CPLDs are still selling for $30-70 each! (eg. $37 for a XILINX XC2C256)
