I'm still of the opinion that AMD doesn't have to take the speed crown in order to remain competitive.
Navi is a good product, just like Polaris was good. It's not a replacement for Vega though.
Vega was... lackluster. Not horrible, but not great. HBM is a great technology but just never hit high enough volume to make it commodity, and Vega could never hit high enough performance levels to allow for margins to make it worthwhile. It may yet make a comeback later on, but Fiji was a poor product for gaming and while Vega was much better, it was a case of too little, too late, for too expensive to compare to what nVidia had to offer.
nVidia does have the edge right now, in architecture. And yes, once they get on the same process node that gap will probably widen again. But that really only plays in the top tiers.
Now top tiers are the exciting part, and where a lot of the readers here are looking at. But they aren't the majority of sales or what drive companies quarterly statements.
AMD is taking the safe path - keep plugging at the high volume products. Don't focus on the low volume halo products. Lock down high volume custom order contracts. Keep the R&D budget moving but under control and focused on your target markets, not chasing halos. I think it's a good strategy. It's just boring to watch it play out.