A game that runs docked only, is a game they shouldn't make. The relationship between the portable specs and docked specs are important, and a 720p to 1080p scenario is the one that is easiest to maximize. Next gen consoles won't even be out for 2 years and won't have exclusive (non-current gen) games from yearly multiplats like Assassin's Creed for at least another 2, Nintendo putting out a 2nd iteration of the Switch, that is again more powerful with 7nm+ pushing 2TFLOPs+ when docked in 2022, would allow them to get cross gen games then.
You think that a 7nm chip would only allow Nvidia to produce a 921GFLOPs GPU + quad core ARM CPU?
"TSMC called their process at this "node" 16nm to reflect relaxed pitches. The initial process was 16FF followed quickly by 16FF+ with a 15% performance boost. 16FFC is now available and is reported to have 8 to 10 less masks driving lower cost while offering 0.55 volt operation for low power (50% lower power)." 12nm uses 60% less energy than 20nm planar in Tegra X1, while offering a 20% increase in performance. A57 is a power hog, at 1.8ghz it consumes 6.5 watts, while an A72 at 1.8GHz consumes less than 1.5 watts. 7nm is over 60% smaller than 12nm,
"The 7nm process offers 35% to 40% performance gains over 16nm or a >65% power reduction."
You might be investing in Switch currently using 16nm, it's not likely, but whatever, it's your thoughts to think about, what you should remember though is that Tegra X2 produces Switch's docked performance in 7.5 watts, with a 20% faster CPU, 10% faster GPU (437GFLOPs) and 100% more memory bandwidth, that is on 16FF, not even the 16FF+ or 12nm. 7nm would absolutely blow away those specs with TSMC's numbers. Remember the A72 is using 1/4th as much power than the A57 cores, well they could offer that CPU performance at 2.5GHz quad core A72, still have that extra 5 watts to use on the GPU side, offering a portable 1TFLOPs GPU would not be outside the realm of possibilities, remember Nvidia is the leading GPU manufacture on the planet when it comes to performance / power ratio. 7nm isn't what the next Switch will be though, it's 12nm. That is the most likely push because that is what they need the Switch to do, about 900GFLOPs docked.
And yes, the entire board for TX2 consumes 7.5 watts in Max Q and 15 watts in Max P, that includes RAM, Fan, and other aspects of the computer:
"Meanwhile the board's Max-P mode is its maximum performance mode. In this mode NVIDIA sets the board TDP to 15W, allowing the TX2 to hit higher performance at the cost of some energy efficiency." I also like Anandtech as a website, they rarely get anything wrong and they are very careful, which is why I am certain they didn't make a mistake with the use of Board TDP here, SoC in Max Q is allowing the board to run at 7.5 watts, not the SoC, much like Max P places that TDP to 15 watts.
Nintendo's iterations will likely go 12nm in 2019 and 7nm in 2022, there is a pattern these devices would create, allowing for an expected 3nm Switch in 2025, again increasing docked performance by 2.5x, so Switch in this scenario
looks like this:
Switch Spring 2017: 393GFLOPs docked, 196GFLOPs portable
Switch Fall 2019: 944GFLOPs docked, 393GFLOPs portable.
Switch Fall 2022: 2.35TFLOPs docked, 944GFLOPs portable. (this is a year or two before a probable PS5 Pro arrives)
Switch Fall 2025: 5.8TFLOPs docked, 2.35TFLOPs portable. (this is a year or two after a probable PS5 Pro arrives)
It's important to remember that docked performance is targeting 1080p on the newest Switch units and 720p on the previous ones/portable modes, while PS5/PS5 Pro would be targeting 4K. Nintendo would likely offer a 4K dock after 4K becomes relevant enough for Nintendo to care about, likely sometime in or after 2022.