Not so fast...
Part of that statement is true.
Let's take it line-by-line:
Fair enough, not the end game... but it is a start nonetheless.
Yes, but I suppose nVidia, the PhysX team and every game developer feels pretty much the same way. Bit of a non-statement.
This is debatable.
On the one hand we have plenty of examples of games that were successful while riding mainly on their 'eye candy' hype.
On the other hand, we have plenty of examples of games that were successful simply because they had good gameplay... they were basically just good games.
If GPU physics is included in such a game, it will piggy-back on its success and gain traction that way.
Heck, the entire OpenGL API pretty much just piggy-backed on the success of GLQuake, and other major games which adopted the engine (eg Half-Life). It helped push Glide from the number 1 spot of 3D APIs.
Yes, but see above... that is not necessarily a bad thing. More eye-candy is pretty much what the entire GPU market revolves around anyway. It's a bit hypocritical. I mean, if you compare the early Voodoo accelerators to today's DX11 cards, they're a truckload of features on the new cards, but hardly any of them have much of an effect on gameplay and experience... They just give you more eyecandy (realtime shadows, HDR effects, sharper texturing, antialiasing etc).
Since pretty much all games make use of these features, and nothing looks like early GLQuake or such anymore, I would say that eyecandy is generally a good thing, and AMD made a lot of money selling it.
Okay, you said that more than 2 years ago... Where is it?
Seems like you haven't even made a start yet, let alone that you are anywhere near your proposed 'end game'.