Navi looks to be an amazing architecture.
Its nice because since Vega was released on 7nm we can compare Vega to Navi on the same process. Vega 20 is 331 mm², and Navi 10 is 251 mm². That makes Vega 32% larger, yet from what we know performance will be very close. Geforce 2070 (TU106 ) is 445 mm². TU106 is 77% larger and performance, we are led to believe, is ever so slightly behind Navi 10. Let's call them equal in performance. So yes, NVIDIA has yet to use its node-jump, they are also on their brand new architecture Turing, which I doubt will get changed much due to the sudden important of RTX.
Far and away the most important caveat here is that we are comparing TSMC's 12nm to their 7nm process. NVIDIA will be hopping over to Samsung's 7nm not because of performance or density scaling, but cost - as reported by multiple sources.
AMD's midrange 241 mm² Navi 10 die equaling NVIDIA's 445 mm² die means that NVIDIA will at best match them if they are able to take full advantage of 7nm. Navi looks to be an excellent architecture.