Do you really think that? Software emulation with graphical glitches is not what a professional enterprise like Nintendo can offer.
They would develop the emulators internally with intimate knowledge of their past hardware. Performance and fidelity could be much better than what you are seeing on enthusiast developed emulators for other platforms. Worst case scenario they'd have to optimize on a game to game basis and focus on the best sellers first.
And by the way, the Bobcat design (Jaguar wasn't available 1 year ago) is MUCH LESS efficient than the one on the WiiU.
Llano APUs feature the Husky core based on an optimized K10.5 microarchitecture with much higher per core performance than any next gen console (three-issue wide, big fast L2, high clocks).
A8-3820 crams a 4 core CPU clocked at 2,5 Ghz and 400 SP GPU at 600mhz into a 65W thermal envelope. A ~120W design I suggested before would facilitate a higher clocked GPU featuring 768 SPs and a wider 256 bit memory interface (or 128 bit GDDR5). If emulation required it, the CPU could turbo two cores to ~3 GHz or they could implement translation hardware on the graphics command processor to aid in Flipper/Broadway emulation.
This means more power draw for the same performance or even less, and Nintendo having to make a compiler optimized from scratch and re-do all their tools.
It wasn't an option, really.
The first part is irrelevant in pursuit of performance, of course the upfront cost would be higher across the board. They already had to re-work their tools for multi core support and that Frankenstein of a GPU.
Whether it was an option in the eyes of decision makers at Nintendo or not, frankly nobody here can claim that. It definitely was an option in a wider sense and an oportune one which would have allowed them to close the performance gap and vitalize 3rd party suport.
In my previous post I forgot to mention that they'd be bringing Xbone level performance to the market a year earlier than the competition. It would've been priced higher, I'm guessing between $399 to $449 to break even, if they insisted on including the tablet controller. By the time the PS4 launched the production cost would drop to profitability levels with adoption of recent technology like 20 nm DRAM and freeing up of 32 nm production slots at GloFo since AMD would be moving to 28 nm.
Meh... just a nice little fantasy that, had it realized, would drive me back to being their fan.