(this originally started as a response to a comment on the "Wii successor" article, but I decided it was too wordy...)
It's interesting to look back at Nintendo's hardware strategy; much of their success has derived from making skilfull use of three things.
The first and most obvious is their IP; aside from some early arcade licences (and the 5 titles which appeared on the CDi as a sweetener for Philips after Nintendo decided to abort the planned CD-drive for the SNES), the only place to get Nintendo games has been on a Nintendo machine, and for the most part, Nintendo has carefully husbanded their IP to keep player interest high from generation to generation.
However, the second is the application of Gunpei Yokoi's "Lateral Thinking with Withered Technology"[*] philosophy. In simple terms, this philosophy revolves around the use of commoditised technology, often in a previously unheard-of context. The benefits of this approach are twofold: it makes the hardware cheaper to produce and the novelty of the "lateral thinking" also provides a hook to intrigue potential buyers.
The use of this philosophy has been most apparent in the handheld arena: the Gameboy was woefully underpowered when compared to Game Gear/Atari Lynx, but was physically smaller, somewhat cheaper and had a much longer battery life. And it also benefitted from one of the biggest IP coups yet seen in the games industry: the exclusive availability of Tetris. Similarly, the DS was handily outperformed by the PSP, but the smaller form-factor, novelty of the touchscreen and the presence of several exclusive IPs (e.g. Pokemon, Mario Bros) led to it selling over twice as many consoles - 145 million vs 68 million, if Wikipedia is to be believed.
Admittedly, the use of this philosophy isn't quite as clear in the home-console market: on paper, Nintendo's consoles have generally had similar specifications and feature-sets as per the equivalent consoles from their rivals. However, things become clearer when you look at the release timelines:
Megadrive: October 1988
SNES: November 1990
Saturn: November 1994
Playstation: December 1994
Nintendo 64: June 1996
Dreamcast: November 1998
PS2: March 2000
Gamecube: September 2001
Xbox: November 2001
With an average delay of around 18 months, Nintendo were able to utilise the "withered technology" approach and take advantage of the cost-reduction effect implicit in Moore's Law: the hardware in their consoles is equivalent to their competitors, but the initial production/licencing costs are significantly lower. And while this approach gives their rivals a head-start in the market, Nintendo's IP has generally been strong enough to overcome this disadvantage.
Unfortunately, this approach became increasingly less successful as games shifted from 2D to 3D - though to be fair, this was perhaps more due to Nintendo's insistence on maintaining control over the physical media: the Nintendo 64's cartridges were a major hinderance when it came to trying to create realistic 3D worlds such as those seen in Metal Gear Solid and Final Fantasy, while similar applies to the Gamecube, with it's 1.4gb mini-DVDs vs the standard 4.5gb available to the PS2 and Xbox.
(it's worth noting that both consoles were profitable for Nintendo, but they failed to dominate the market in the way that the NES and SNES did. Indeed, if it wasn't for the multi billion dollar success of the Pokemon franchise on the GBA and DS, Nintendo could well have crashed out in the late nineties in much the same way as Sega did...)
In any case, Nintendo decided to apply the full "lateral thinking and withered technology" philosophy to the Wii, by combining a cheap motion-control system with low-end technology. The combination of novelty controls and low cost led to the Wii attracting interest from the relatively untapped casual market, helping it to become a runaway success.
So: that's a potted-highlights history of Nintendo. What does this imply for the Wii 2? Well, herewith are some predictions for the future...
Prediction: yes, but only with the Wii, not the Gamecube
I'd expect the Wii2 to be backwards compatible with the Wii - aside from Nintendo having a strong tradition of maintaining BC (e.g. GBA/GB, DS/GBA, 3DS/DS, Wii/Gamecube), I suspect it's going to be a vital element of their campaign to try and pull people over to the Wii 2. A bigger question is whether the BC will be based on software or hardware; the hardware approach seems most likely, given that the Hollywood GPU was a custom design built specifically for Nintendo.
Prediction: multi-core, PowerPC.
Arguably, CPUs haven't advanced that much from 2005, as the irresistable force of Moore's law has smashed against the law of diminishing returns: not only were the cores getting hotter and the yields lower, but it's been difficult to find mainstream uses for the additional power. Instead, CPU manufacturers have gone in another direction (partly driven by the expansion of the mobile industry), adding more cores and reducing power consumption.
Still, even a low-end modern multi-core PowerPC CPU should be far more powerful than either the Cell or the tri-core CPU of the Xbox 360. And sticking with the PowerPC architecture would also make it easier to maintain backwards compatibility with the Wii.
Prediction: AMD, 9th or 10th generation architecture, HMDI and component output
Where the Wii used a somewhat custom GPU (designed by Artx and Nintendo, then manufactured by ATi following their purchase of Artx), the Xbox 360 and PS3 essentially used standard GPUs - Sony opted for an Nvidia part, while Microsoft opted for ATi (now AMD). However, both chips hailed from the 7th generation of GPUs; we're now moving into the 12th generation and getting an order of magnitude more performance than anything which was available back in 2005.
Admittedly, the odds of Nintendo opting to use a 12th-gen part are low; it'd simply be too costly for the "withered technology" philosophy. But by the same token, I can see them opting for a 10th-gen part; not only would it be low-cost, but it'd also be significantly faster than the Xenos/RTS chips used in the PS3 and X360.
Of course, the second question is: who would they buy this part from? The main candidates are obviously Nvidia and AMD; PowerVR could also be a candidate and there's also a dark horse, in the shape of DMP, who produced the GPU used in the 3DS.
Of these, AMD seems most likely; Nintendo already has a relationship with them and it's debatable as to whether or not PowerVR and DMP chips have the horsepower to pump out high-end HD graphics.
There's also the "3D or not 3D" question, but I tend to think Nintendo won't actively implement this on the Wii 2: the market penetration isn't there, much as happened with the Wii and HDTVs. And the jury is still out as to whether the 3DS is going to sell well or prove to be a flash in the pan.
Note that this doesn't mean that Nintendo won't support 3D displays, but I'd expect it to be an optional feature rather than the standard.
Prediction: 1GB RAM
A quick dig around on Google indicates that the retail price of ram has dropped from around $200/GB in 2005 to about $15/GBin 2011. So, for all that console manufacturers have traditionally skimped on RAM, I'd expect Nintendo to put at least 1gb into the Wii 2. But it's unlikely that they'll put more in; aside from the cost, the "single-tasking" nature of game consoles means that they don't see as much benefit from additional ram as a multi-tasking PC does.
Prediction: DVD drive, internal flash memory, SDHC card reader
Nintendo are unlikely to move away from the DVD format; for all that Blu-ray has several advantages (capacity, movie playback), it's heavily associated with Sony, and I can't really see Nintendo betting the farm on a DLC-only system.
Internal storage is a bigger question; having a hard drive increases the flexibility of the system - being able to cache data on the drive can significantly improve performance and reduce loading times and hard drives are stilll significantly cheaper than flash storage. However, HDDs are less reliable than flash memory and I've no doubt that Nintendo are keenly aware of the way USB hard-drives have been used to make piracy easier for Wii owners. I'd therefore expect Nintendo to stick with a relatively small amount of internal flash memory (4gb?) and an SDHC card reader.
Prediction: as per the Wii
I can't really see them changing things from the Wii here; for all that an ethernet port would be nice, it'd increase the cost of design and production - it may only be pennies per unit, but when you're planning to produce millions, every cent counts.
Prediction: backwards compatible with the Wiimote, but not the Gamecube controllers
This is where things get a bit trickier. On the assumption that Nintendo is interested in maintaining backwards compatibility, support for the Wiimote is a no-brainer - I'd actually expect the Wii 2 to ship with the Wii Remote Plus.
However, this is also the area where Nintendo is mostly likely to apply their "lateral thinking" approach - after all, it's what they did with the DS, Wii and 3DS. The only question is: what else can they do?
There's two candidates which spring to mind, though both come with significant disadvantages:
1) A Kinect-style video camera
On paper, this seems perfect: camera technology is cheap and as the Kinect has shown, motion-detection integrates well with the local-multiplayer games which have been Nintendo's bread and butter on the Wii.
However, Nintendo haven't dabbled too much with video technology (the Gameboy Camera doesn't count!), and while camera hardware is very much "withered technology" (the Kinect runs at just 640*480/30hz), a considerable amount of R&D would be needed for the software side of the equation - not to mention the cost of the lawyers needed to ensure Nintendo's solution doesn't infringe any Microsoft IP. There's also the political aspect: Nintendo would look as if they were playing catch-up with a rival, rather than being innovative.
2) Touchscreen controls
This seems more likely - Nintendo's second-favorite design approach is to reuse old designs; the DS was inspired by the Game and Watch, while the Wiimote was based on the NES controller, and the lessons they learned from the Virtual Boy were ploughed back into the 3DS. And they've previously dabbled with "informational" controllers on the Gamecube, which could transfer games or information to a linked-up GBA; Final Fantasy: the Crystal Chronicles is probably the most famous example, but it's far from the only one.
All told, it's relatively easy to envisage Nintendo shipping something similar to a DS with the top half removed, leaving a controller with a LCD multi-touch screen and wireless connectivity to the Wii 2, which could then stream a display back to the controller in realtime. On paper, this ticks all of the LT/WT boxes: it's a novel approach (at least for a home console), and it's cheap, since it reuses existing Nintendo IP and technology. It offers some interesting possibilites for local-multiplayer, by providing players with unique views of the action and the ability to secretly communicate with individual players - perfect for boardgames, cardgames, strategic wargames and RPGs. And this approach might even steal a march on Apple, who may well do similar with the iPhone/Apple TV in the future.
However. Touchscreen controls aren't as intuitive as the Wiimote and it's easy to imagine casual gamers becoming frustrated with the need to keep switching between two displays; if there's one major flaw with touchscreen controls, it lies in the fact that there's little or no tactile feedback. And a touchscreen large enough to be practical (e.g. the 3" display of the DS or the 3.5" of the iPhone) will be far too large to incorporate into a Wiimote. Similarly, it'd be difficult to use a touchscreen in combination with a Wiimote, limiting how much motion control can be incorporated into a game. And touchscreens are difficult to use one-handed, as you generally need to be able to stabilise the screen while using it.
Then too, "informational" controllers haven't exactly proven popular to date, though it can be argued that the cost, limited functional and complexity of the two main examples to date (i.e. Sega's VMU and the GC/GBA hookup) were a major factor.
Beyond that, I'm racking my brains a little. You could stick a microphone on the default controller, but voice-command systems are distinctly incompatible with local-multiplayer. You could maybe stick a video camera on the touchscreen controller and use it to drive some form of AR system (by throwing up AR icons on the TV), but it's difficult to see any non-gimmick uses for this.
Still, Nintendo has several teams of people dedicated to R&D, so it's entirely possible that they've dreamed up something new and interesting.
The big question is whether the Wii 2 will be able to appeal to the casual market in the same way as the Wii did: with profits dropping like a rock and a relatively alienated "hardcore" audience, this is all looking similar to the situation Nintendo found themselves in with the Gamecube.
Can they pull the rabbit out of the hat twice? Time will tell...