One of the situations that I used to encounter all the time (still would if I weren't "retired") is picking a microcontroller part number for a piece of (let's say) avionics equipment that's got a minimum service life of 20-25 years, then finding out that the part that was considered completely "marketworthy" when it was designed at the 45 nm node is considered "obsolete and unsaleable" when the 28 nm node becomes available. This is bad enough in the normal scheme of things, but (and the following really happened) when your vendor was Motorola then the only available source becomes NXP, who then gets bought by Qualcomm who then gets bought by Broadcom who then gets bought by...yeah good luck with that "25 year" thing!! (And this is from the same people who are heard all the time blasting Wall Street for being "too short-term oriented"!?)
Moore's Law has been used as a marketing propogenda for the time being and, as you mentioned, people have been blindly following Moore's Law. So, there has been slim chance for innovative technologies to change the rule of the game. Now, customers should find out game changers. Otherwise, customers should pay more.
Problem is that many people are still blindly following Moore's Law even though Moore's Law ends economically for logic products and physically for memory products. People won't change becasue big semiconductor companies are making money. Simply end custoers should pay more as we see the memory market sector.