So you focus all of your effort on the short term, and then what happens long term? You get left behind.
NVIDIA is winning because they invested in the long term, they built the CUDA ecosystem that is now so difficult to dislodge even Intel's massive resources will require massive amounts of focused execution to make a dent. NVIDIA built the GPGPU focused Fermi before AMD brought GCN, and they kept investing heavily in CUDA even when initially it was giving them massive trouble to have multipurpose GPU's (HD 5870 vs GTX 480... yeah). AMD is now investing into machine learning and building their own open ecosystem to make inroads into the markets NVIDIA is dominating, and Vega 20 will likely be the first AMD GPU to never see the light of day for gamers (or only in expensive "Titan" form like Titan V).
I am always impressed by forum strategists and their mental gymnastics.
As for Koduri, AMD, by his own admission, lost focus on dGPU's after 2012. Management expected iGPU's to be the future. He was severely underfunded for the amount of markets he needed to target with his architectures.
I also wouldn't underestimate the major contributions he made with RTG's driver focus. As an R9 390 owner I felt the difference in quality and consistency that he brought about.