The Tegra X2 or whatever its called, is supposedly almost twice the performance/watt of the Tegra X1.

So if you made a new "switch" on 16nm (with X2), instead of todays Tegra X1 (20nm), it would be around double the performance (at same power usage).

393 Gflops -> 786 Gflops.   *edit: (nvidia themselfs say Tegra X2 at 15watts is 750 Gflops Fp32)

Thats still less than half of the PS4.

Even 5 years from now,.... with a Nvidia Tegra X3? and maybe 7nm, I suspect it ll still be abit weaker than the current PS4.

I know nintendo make great games, dispite the graphics of their hardware being weaker.

What Im saying is "what if nintendo didnt have weak hardware?"

You are talking about a five year span here. Tegra will likely not even be labeled "X-whatever" then. I suspect an "X3" would be long old technology, and the x-series analogous to the non-K/X series today. 

Again Tegra 2 had a max of 7 GFLOPS in 2011 and the X1 500 GFLOPS in 2016. Even if Moore's Law slows its pace for GPU technology like it has for CPU's (which is a wild prediction), it is unreasonable to believe that it will slow that much.

Switch's successor is going to easily be more powerful than a base PS4, and likely will match or exceed a PS4 pro.

Apples & oranges.

Tegra 2 was ment for smartphones. (0.3-1,5watts)

Tegra X1 is ment for cars. (10-20watts)

You cant really compair them directly because their at two differnt power consumption levels.

Nvidia:

2013 : Tegra 4 = 76.8 Gflops @4watts (28nm hpl)

2014 : Tegra K1 = 365 Gflops @8-10watts.  (28nm hpm)

2015 : Tegra X1 = 512 Gflops @15-20watts (20nm)  (shield TV uses 20-22+ watts when gpu is at max load)

These are big improvements over just 2 years time... however I dont think it continues at this rate.

Their big games would take even longer to release, many franchises would get abandoned because their big games would need more resources, you wouldn't be able to play their games on the go.

Tegra 2 was used in tablets as well as phones which should have similar power requirements to the Switch when gaming. What it meant for does not matter, it is what it was used for and how much power was available which matters. A fairer comparison is that between the Switch in handheld mode and a 2012 tablet with tegra 2. The difference is still quite large. 

There is no reason to think the theoretical performance gains are going to change in rate. GPU's are paralellizable (you add more cores with die shrinks) and therefore more closely realize performance gains from transistor count than CPU's. That is why performance changes so rapidly all else held equal. Nvidia has already hinted at Volta Gpu's with 512 CUDA cores. And that is for this year. 

I was thinking more like.... instead of Switch games running 900p docked,... with a PS4 they could run like 4k instead.

You would end up with something akin to this:

https://www.youtube.com/watch?v=7kzZxF-upm4

It wouldnt take more work, it would just be higher resolution and look abit better.

wow, that was very unimpressive, all that video showed me is diminishing returns.

So basically we would be losing out on portability to make the games slightly better looking.

I think its rather impressive.... this is what games would have looked like (or better) if Nintendo had done a normal home console like a PS4.

I dont have a switch, but for 299$ Im not sure how id feel about takeing it out with me. I wouldnt have minded a traditional console.

I thought the same, that was impressively unimpressive.

It is MS who should go to no party. Nintendo needs to step up their f*ing game.