Design issues, manufacturing issues, or both, caused a pre-launch 180-degree reversal on Polaris positioning. In any even, AMD found itself stuck with no performance competitive parts and loads of slow parts. AMD then did what it always does when faced with a product disaster - engaged in marketing spin. "Oh, we never intended our GPUs to compete in performance with nVidia's. We always intended that our parts be low margin, low price, bargain basement alternatives to the superior products of our competitors."
This is nothing new. For decades Jerry Sanders tried (and lied about ) the "give away margin/make it up on volume" strategy. It never works. AMD's cost structure is much, much higher than peers'. It's margins are the smallest to begin with, so price competition is a dull blade in AMD's hands; it loses money at levels where the competition still generates healthy margins.
This is just another verse in the sad, sad song that is AMD's long history of over-promising, under-delivering and then doing damage control when the failed execution becomes public.
Polaris will lose a significant percentage of AMD's very small remaining GPU markeshare to Pascal. This is no different than when Maxwell cuts AMD's GPU market share in half. Polaris is a Maxwell response, not a Pascal response. Again, AMD is a generation behind. This is a very attractive place to short AMD. It will be back in the 2s very, very shortly when the market rejects the new products.