Akash_11 :
looks like coming year is not good for NVIDIA, still NVIDIA is on 16nm chip instead 14nm.
16nm vs 14nm is completely pointless. Physically, most details on "14nm" are multiples of 16nm anyway due to both processes using the same UV sources and being bound to the same wavelength fractions in their phase shift masks. 14nm is mainly a byproduct of screwy math, it has no real benefit over 16nm.
The problem with GPUs much below $200 is that an increasingly large chunk of that market is within IGPs' reach and this will only get worse when AMD launches APUs with 1-2GB of HBM on-package. At that point, it may become cheaper to upgrade the APU on an existing AMD motherboard than buying a $150 GPU with equivalent graphics performance since you only need to buy the chip instead of a whole board..