Almost becoming headache of all the stuff:
Those are no leaked results, there is no evidence that they barely speak the truth, its all a big assumption based on way to many minor factors which is actually based on another assumption. Someone can make speculations based on maybe a single accurate information and with some luck the construct is barely fitting. In term of 390X the "special cooler" may be true, so the construct is a "hot and beefy GPU" which is actually the probably only true stuff released so far and not more than that. But how good it will actually perform and whatever is not more than speculations and it should be handled according to this fact.
One thing is certain, i think AMD would not be stupid enough releasing a GPU that is in need of 2 times more Watt with only a minor performance gain. So far all they did is pushing a old GCN core by expanding the ressources to extreme levels, some may call it "tock" (new architecture) but there is neither a tick (shrink) nor a tock, its simply a expansion of a old architecture for even more performance and nothing more than that. The efficiency actually even was decreasing, for example 290X vs. 7870, but Nvidia wasnt better than that with the release of the more beefy "Kepler" versions. The Maxwell is a completely new architecture, i have to tell the obvious because people apparently arnt noticing it and they actually made a big step in efficiency, thats a fact.
Now what will happen to AMD and the Radeon? Surely they had over 3 years for the development of some better architecture, so i dont expect it to be unable to compete. In term they need almost twice the power for minor performance gain, the new architecture is unable to compete, thats a fact. Not necessarily when it comes to huge tower PCs, they can handle the heat, but SFF and Notebooks will be out of question... and honestly... AMD isnt going to be that stupid, they can lose a huge market. In term they can pull out 50% more performance at 300W it is surely competitive, at least on the desktop market but not SFF or Notebook market still. Stuff worse than that isnt gonna happen, and most likely not the case.
Anyway, lately the community seems overly focused on efficiency. In the past it was almost non existent, i remember when someone was boasting with a SLI 480 and it was a insane hair burner, although the user didnt care at all as long as performance was fine. Nowadays it seems even the enthusiast is becoming sensitive when it comes to efficiency... interesting changes and well... why not. But it would make huge tower PC pretty much obsolete because cool cards such as 970/980 can even be fitted inside SFF with zero issues (SLI too). Surely interesting changes even in the enthusiast markets and seems like Nvidia can afford to release cutted down GPUs because people now actually care the efficiency maybe even more than sheer performance.
I think, currently when it comes to "big PC" market, Nvidia was pretty much slaping those enthusiasts in the face because its below the thermal capabilitys of such a PC. The current spec is a great SFF GPU with the perfect TDP for SFF use. It can be overclocked yes, but only a small OC would make sense in term of efficiency. In term a high OC is involved (past the 1500 line) the current Maxwell chips will suffer crazy efficiency values up to the point where it would only be marginally better than the old GCN or Kepler. So its somewhat hilarious that the people value efficiency nowadays but may OC some Maxwell to insane levels because they can and the TDP is offering no challenges, a situation i dont enjoy at all. Once again, a Maxwell is only very efficient close to stock clock... not at 1500+ or whatever.
Regarding AMDs 300 series, i think AMD is doing a different approach, a GPU that is actually scaling better with OC compared to Maxwell in term of efficiency, so the basic value can look worse but the "enhanced value" might be smoother, so a good cooler can make sense and a big desktop enthusiast PC is becoming useful once again. There can be many different approaches and as long as i cant judge the 300 series i would not make to much assumptions... i could end up spreading way to many lies.
When did I ever say that it would double the performance? Please, show me where I did.
And how, exactly, would two 980's be more stressful than one 390X?
You said you would use double Watt in order to be on par (or maybe better) with a possible new Radeon, but it is not double efficiency nor double performance, so you are trading your so praised "efficiency" in order to beat a Radeon. Ok, why not, as long as performance is toward your expectations guess everything is OK. On top of that you may even have to pay almost double the price and in the end Nvidia will be releasing a 980TI (i guess this is only a question of time) and now that you was handing them over probably 1200 USD, you may not even become enthusiast grade for more than a few minor months. Such decisions is up to you but i would refrain from making it tasty to any users as the "ultimate solution", such a claim can be dangerous.
Yes it is more stressful, im not gonna argue because it would be same such as telling a child why a lollipop is tasting sweet.