Tegra X1 is a fine chip but it was designed more for PS3/360/Wii U tier software and a little better than that.

When Nintendo releases a better Switch or a supplemental dock or something that can more adequately run PS4/XB1/PC engines and it's very easy to port, they will get a ton of 3rd party support.

When Nintendo release that we will already have PS5/XB2 on market and again there will be difference in power, and porting will probably again not be too much different compared to its now.

1. High lead-up times and a long development pipeline means AAA games in particular won't manifest very quickly. It took at least 2 years for most AAA game studios to change their baseline target to current generation hardware so who knows when Switch will catch on to the next release cycles ...

2. Tegra X1 is a low performance chip. By comparison home consoles have higher performance chips and it was specifically the reason why they were successful early in the generation with cross generation projects since doing a rough port meant an IMMEDIATE performance uplift. The same can not be said when porting current generation projects to the Switch along with the fact programmers have to deal with a different hardware architecture such as the Nvidia CPU/GPU ...

3. Switch has a different audience expectations and we can take a look at Japan for an example where many smaller projects which have development conditions sympathetic to the Switch fare far worse in terms of sales performance compared to PS4 ...

The problems with the Switch are currently 3 fold and it's probably not going to get any better for the Switch from this point onward before it get's worse when a new generation is less than 3 years away ... (even smaller projects some of which could target last generation hardware are starting to capitalize on the ubiquitous nature of current generation hardware and that can potentially put the Switch in further jeopardy)

There are a lot of smaller projects heading to Switch, whether it be indies or AA Japanese games. You have games like DBZ Xenoverse 2, Mega Man 11, and Disgaea 5 find success on Switch. Plus, Octopath Traveler shows that third party developers can find success on Switch with original titles, albeit not as expensive or graphically realistic compared to games on PS4 and Xbone. Thus, I don't know what you're trying to say here.

Install base and success of platform is also factor when comes to 3rd party support, and become obvious that Switch will be very sucesffuly platform and install base will keep growing so more and more 3rd party games will keep coming in any case. Also, Switch has good support of small even midsized games, but its lacking in big AAA 3rd party games.

Switch currently has listed 1231 games, that list is constantly increasing rapidly.

https://en.wikipedia.org/wiki/List_of_Nintendo_Switch_games

Depends. I think cross-gen is going to be a thing for several more years, so PS4/XB1 will be getting years after PS5/XB2 are out. The install base of those systems is just too big to ignore. So if a "Switch Pro" is a late 2019 or 2020 product, Nintendo could get a lot of games, and there's still tons of XB1/PS4 content that would be a lot of fun to play on the go as is. 

Switch Pro will be Tegra 2 most likely, so we dont talk about any huge difference in any case compared to current model, also you need to remember that same games would need to work on current Switch model also in any case, so from that point nothing will really change for 3rd parties.

We'll see. The leak in the Switch OS from hackers mentioned a Tegra 'Mariko' with custom model number that isn't the Tegra X2. 

X2 is actually a fairly old chip at Nvidia, they likely can give Nintendo something beyond that by 2019/2020. 

I'm not sure if Nintendo necessarily will mandate anything either, as far I know they didn't mandate for example that Minecraft had to work on all 3DS models (it only runs on New 3DS). It will be up to the developer I think. 

I mean maybe we will have bigger upgrade with Pro that X2, but offocurse that games would need to work with older Switch units in any case, at least huge majority liky in case of New 3DS or DSi. Buy time Pro model arrives (I dont think that first Switch revison will be Pro and that Pro will arive in 2019.) Switch will probably have install base of around 50m+, so making game that will not work on older units means huge majority of install base of platform cant play that game.

Well the alternative for some games is going to be 0% of Switch owners can play a game, if its too hard to port to the current Switch. Some % is always going to be better than 0%. 

There is also that supplemental compute device they patented, when you using chips this small, who knows maybe they could make something like a companion device for the older Switch models that has the new Tegra inside of it and can handle the graphics processing. 

I think Switch eventually is going to become more like Steam than a piece of hardware, an ever growing software library, as you scale up you get access to more and more games and your old games come up with you rather than something that is completed anchored and stuck to one hardware spec/hardware model for 5-6 years.