I'm not sure I agree about his leadership skills. Vega was a terrible gaming card, HBM was a terrible bet in general for gaming cards and 4GB HBM1 was a death blow for Fury (not sure how much this was his fault, as this was ~2 years after he returned, but he still kept talking about HBM all the time) - but he's a well respected engineer, and maybe with unlimited funds at Intel he'll do better. However, I mostly dislike him because of marketing and his huge mouth, and he couldn't deliver. Not saying it's his fault, it's difficult combating both Intel and Nvidia with lower R&D funding, but marketing under his leadership was terrible.
It's so weird to see people criticize GCN iterations as if it was Raja's fault. Those were already on the roadmap, well under way without his input.
His only fault is as you say, his tendency to over promise to the media and not realizing the financial problems AMD was under so he should not have promised an entire Polaris stack when AMD could not afford to design and bring up so many. That was peak Zen time.
Funny thing is, Polaris 10 is one of the best GCN design from AMD, it competed fairly well with the 1060, similar die size, not massively more power, and only slightly more transistors. You can see AMD was broke, they recycled it as 580, 590.
The biggest fail was Vega, end of the GCN line and expensive HBM2. Maybe he should have kept his mouth quiet and don't talk to the tech media. But the way I see it, that is GCN's life cycle, it ran into the dead end just like Carrizo for Bulldozer.
GCN was always behind in Tflops to gaming performance vs NV, some by a huge margin, it needed many more transistors & power to compete, making high-end off-limit as they were already busting 300W to compete vs mid-range NV stuff. But RDNA1 brought them back from the wilderness IMO, just look at its perf/transistors and Tflops, very comparable to NV's best.
David Wang + Suzanne Plummer has a good base to build upon, and for all the ppl thinking Raja and his big mouth deserves disrespect, at the very least, he did deliver to AMD what he was hired to do, a new architecture beyond GCN.
Perhaps Raja should have been like Keller and stay quiet, focus on the work, and not be the face of the current roadmap GCN launches, as he becomes the punching bag for what is really his predecessors design and decisions. IF RDNA2 turns out amazing, a lot of ppl will praise David Wang, however, he has always been one of the chief graphics architects, and one of the key player who designed GCN back in the 2009 to 2011 era. Nobody blames him for it, rightly so, because AMD wanted a compute GPU for datacenters to chase after lucrative markets that CUDA opened.
It really comes down to AMD itself, it was broke, approaching bankruptcy, it has no right to chase after so many markets. Should have focused, pure gaming, both CPU & GPU. Instead they got FX, and GCN and stuck with it for a decade until something new was developed.