10tacle :
Well that I can see. But my thinking was they need to have an if/or approach to forecasting market demand. One for stability/growth in cryptomining and one for returning to sanity of "normal" GPU demand.
The problem with predicting cryptocurrencies is that they are unpredictable by nature as nobody has control over them, nothing prevents anyone from forking existing currencies, nothing prevents anybody from starting their own, there are no standards for developing currencies which translates into no way to predict which resources the next currency is going to favor, there are no guarantees that miners will pick your products for mining, no guarantees that difficulty increases won't deter people from mining currently known currencies by the time your products launch, etc.
The lead time for wafer starts at foundries can exceed a year. Can you accurately predict with any degree of confidence what will happen to alt-coins within the next year? That's what AMD and Nvidia would need in order to justify a significant production increase aimed specifically at offsetting mining. If they commit to thousands of extra wafer starts and end up not needing them, cancellation fees will be in the millions of dollars should they fail to find buyers for those now unwanted wafer starts.
It all boils down to this question: is attempting to solve the mining issue to make gamers happier by increasing production really worth taking on millions of dollars in unnecessary additional liabilities? Probably not. People who still want a new gaming GPU still have the options of waiting however long it takes to get one near its MSRP or buy previous-gen. Previous-gen parts remaining desirable due to current-gen parts being perpetually out of stock is win-win for GPU manufacturers.