I can say with a fair degree of confidence what the solution to Nintendo's OS load time issues with the Wii U is, and it's not faster NAND, snappier processors or anything like that, it's
LPDDR4.
It's occurred to me that the Wii U's "slow OS load times" compared to the PS4 and XBO have pretty much nothing at all to do with the hardware performance or software implementation, it's all to do with how Nintendo implements the Wii U's sleep state compared to the other two. Effectively, when you put the Wii U to sleep the system is almost completely powered off. Espresso probably keeps going to monitor the network, handle downloads, etc., but it presumably does so using purely the 32MB of eDRAM, as it would appear that the DDR3 is completely powered down.
By comparison, the sleep states on the PS4 and XBO keep a lot more functionality powered up, most notably the PS4's GDDR5 and the XBO's DDR3, which is what allows them to keep features like "suspend application" working, and what allows them to have near instant-on response when you press the on button; the OS was already loaded up and ready to go before you pressed the button.
You can see the effect of these two approaches by
the standby power consumption of the three consoles:
PS4: 10W
XBO: 12.9W
Wii U: 0.4W
Obviously those PS4 and XBO power draws aren't purely RAM, but it does seem that Nintendo's drive to keep standby consumption low has prevented them from implementing an "instant-on" wake from standby; they're not willing to increase the power draw by keeping the DDR3 wired.
I actually did a bit of an experiment with this on my PS4 and Wii U (no XBO to test on, unfortunately). I fully shut down my PS4 (which requires navigating through a few menus) and then turned it on again and timed it from pressing the on button to getting to the main game select screen. I then did the same thing with my Wii U. The results are actually quite informative:
PS4: 31 seconds
Wii U: 18 seconds
The Wii U actually takes significantly
less time to cold boot than the PS4, despite whatever technological advantages the PS4 has. The perceived notion of the PS4 (and, I assume, XBO)'s advantage over the Wii U in OS loading speed is down entirely to the different standby implementations.
While I wouldn't be surprised if Nintendo pushed the envelope in power consumption for NX over Wii U, I would be very surprised if the console consumed anywhere near 10W in standby. However, if they want snappy OS loading, they're going to need to find a solution where they can keep the main system RAM wired during standby, and GDDR5 (or even DDR3/4) would simply consume far more energy in standby than I can imagine them being happy with.
Which brings us to LPDDR4. As we've discussed earlier in the thread, it does have a couple of advantages in that it provides very high bandwidth per module, and would allow them to use identical RAM modules between the home console and handheld. The big argument I'm thinking of, though, is that LPDDR is specifically designed for extremely low standby power, as the phones, tablets and laptops it's used in need to be able to be left on standby for days or weeks on end without the battery draining. Although I can't find any precise data on LPDDR4,
Samsung claims that their LPDDR3 uses 87% less power in standby than DDR3, and I can imagine that a similar gap would exist between DDR4 and LPDDR4.
LPDDR4 (or LPDDR3) seems to be pretty much the only way for Nintendo to ensure NX has snappy loading times while sticking to their low standby power philosophy, and given how much Nintendo has apologised for the Wii U's loading times, I'd be very surprised if it wasn't something that was taken as a priority when designing the new console.