Does anyone actually believe AMD will be able to compete at the top end against nvidia this year?
Why do you believe they won't? How much of a performance increase is Ampere over Turing? RDNA 1 is mathematically capable of beating the 2080 Ti, and even if AMD only manage half of what they intend with RDNA 2's performance improvements it'll be trading blows with Ampere. Unless Nvidia are taking competition seriously and actually give Ampere a proper performance uplift.
AMD being able to compete with Nvidia this gen is entirely down to Nvidia. They've already let their hubris cost them 7nm manufacturing, and if they arrogantly dismiss AMD again and don't make Ampere a properly chunky upgrade then AMD could very well get at them. I know TFlops means very little in isolation, but are you honestly suggesting what we've seen in the XSX so far is all that AMD have? How much could AMD wring out of that alone if it was unshackled from a console's power limit? Smash 300W through those 52 CUs and see what happens.
If you're going to take AMD rumours with a pinch of salt then you should also do so with Nvidia. Nvidia rumours suggest 4x the ray tracing performance for Ampere. Nvidia rumours offer wildly different performance numbers. Nvidia rumours suggest they're actually pretty worried about Big Navi.
I think AMD will do it this time, but my sky isn't going to fall if they don't. There's being optimistic and there's being a fanboy. There's also being a realistic and being a hater. Tread these lines carefully, one is a stance for discussion, the other just makes one look like a fool.