I'm highly doubtful that NX will be a PS Now style streaming service, with Nintendo using those kinds of hosting servers and Wii Us or handheld devices piggy backing off of that, in that way.
PS Now only has to run PS3 level games, it's a supplementary service, not the basis for the whole Playstation gaming ecosystem.
Latency is too high and bandwidth barely sufficient for this to be used by a small portion of overall users, even if Nintendo only expected Wii U numbers to do this it's probably too much, not even taking into consideration the extra demands of more demanding 8th gen console games.
What I think NX is, is a new OS, that can run all of Nintendo's future games, flexible to the point where Nintendo can release new hardware devices for a variety of needs of gamers (Nintendo have basically said this, with their comparisons to Apple's approach to Apps).
It may release with a new console or a handheld or both in March.
Now it's possible that the whole "supplementary cloud processing tech" patent found be people online recently could elude to people being able to buy more than one console to boost their own processing needs for games or when you have multiple handhelds in the vicinity of each other, playing online together they don't have to each handle the processing needs of that whole game alone, but rather they can all share an even portion of the game's demand, maybe even getting a boost from nearby NX consoles that are also playing the same game or playing with those handheld gamers.
With that kind of a network of users the processing power grows as people buy into this ecosystem, so things can evolve as the install base grows.
Games can scale in performance very easily (PC graphics/performance settings are a clear example of this), if developers plan for that in their product designs.
Nintendo have said NX is a long term strategy, which a local, expanding cloud network would be and using a constantly iterated on OS, with improved devices released every few years, it just makes sense in the modern video gaming world.
This kind of a product would allow Nintendo to adjust to a constantly changing market and deliver what gamers need.
Overall the OS centered focus on core software and flexible, evolving hardware design, with Nintendo branding (though Nintendo will need to work on their brand to make their products attractive once again) would allow Nintendo to actually put the games more at the forefront of their gaming ecosystem.
The reason for the March 2017 launch could just be because the processing tech Nintendo are using in their new devices won't be ready to start shipping to Foxconn until late 2016 and that has a knock on effect with Foxconn being able to ship devices to Nintendo for packaging and then the retailers won't get sufficient launch stock until some time in March.
Zen APUs aren't going to be ready to ship until 2017, so there is at least one example of tech that fits the bill for this.
I doubt Nintendo would bank an entire new gaming platform on streaming, maybe streaming will be a part of the function of NX as a platform, but to only be a streaming service is a bad idea at the current time or any time.
When processors are getting to the point when you can put something as powerful as a PS4 in a notebook by the end of the year and take that on the go with you it makes more sense to use it.
Hell it's not even that costly going by what AMD wants to launch Polaris 10 40 CU GPUs capable of powering VR games for.
Zen will likely be affordable, given how many chips you could probably get off of a 14nm wafer.
Nintendo could launch with a 28nm device, but I wouldn't expect delays to have happened if that's what Nintendo were using, unless Kimishima's comments on games requiring more time to develop is true.
Nintendo are meant to have undergone a big restructuring in how they handle development of handheld and console software, so perhaps that did effect moving games to a new development environment.