I'm interested in the development cost of different kinds of chips in the 70s and 80s, both for its own sake and because it aids in the understanding of historical events; in technology and business, cost is often an important reason why things happened the way they did. For example, why did some computer companies develop custom chips, and others make do with what they could get off the shelf, or try to squeeze their designs into Ferranti ULAs that still didn't actually work reliably at 16 MHz.
A valuable data point: How much did it cost to develop the Z80 CPU?
So we did the CPU, the development system, the software and everything else with $400,000. Which, in this day, would be considered totally impossible. But even for those days was probably one-tenth, or at least between one-fifth and one-tenth of what it would take anybody to do.
The Z80 is reckoned around 8500 transistors.
Other things equal, we might start with the assumption that the development cost for a given chip, at a given tech level, is directly proportional to its transistor count. But of course, other things are not necessarily equal. Maybe microprocessors are qualitatively more complex than their transistor count would indicate.
Setting aside memory chips, which are obviously a different kind of thing.
Another important and well-studied category of integrated circuits from the 8-bit era, is graphics chips.
Atari 2600's TIA: Transistor count of early video chips 6193.
VIC-II: VIC-II transistor count maybe around 13000.
NES PPU: What are the circuits captured on this photo of the NES PPU die? no numeric estimates of transistor count seem to be available, but generally speaking it looks in the same ballpark as the VIC-II.
Atari 800: ANTIC on-chip memory again I don't have any numeric estimate, but roughly speaking, I get the impression that ANTIC + G/CTIA together are in the ballpark of VIC-II. (Why two chips instead of one? Three years time difference; maybe the Atari chips needed to be made in a larger process.)
Amiga: three custom chips, around 20k transistors each. These were 16-bit chips, but they were designed in the early 80s, each chip was not dramatically more complex than the VIC-II, they were made in a process that was not much smaller, and I gather the design tools available to the (quite small) company, were similar to those used for the 8-bit chips.
Do any historical sources record the development cost of any of these chips?