I think we're probably looking at 160 or 320 shaders which is the level you see on the Llano APUs.
I imagine a Fusion APU with 3 cores and an "old" 64->80 VLIW5 shader GPU could be had for as little as ~75€ by mid-2012.
A 320 to 400 RV7x0-level ALUs @ say 500MHz would put it head and shoulders above the X360. Which is much more than "a notch." Then again, we only have vague second hand opinions on the expected overall "power" of the machine. And since opinions are highly subjective, that's hardly enough for us to get by and set a proper frame of reference for the GPU.
If the source of that 01.net rumour has access to a devkit/doc, and thus this "a notch above X360" has any credence, then we should expect something like 240 ALU
based on the RV7x0 architecture having 80 ALUs per cluster.
If the source just bases its prediction on the overall picture Nintendo (or others) gave of the what to expect from the GPU, then a 4 cluster setup (thus 320 ALUs) is a possibility.
BTW, rumours are saying the 6" screen isn't multi-touch capable. Damn Nintendo, getting greedy in the wrong places again!
You'd need hard coating for a capacitive (multi-touch) screen. Which would make the gamepad heavier with a 6'' screen... and obviously much more expensive than just resorting to a simple resistive screen. And in this case, price would be a much wider issue than just Nintendo being cheapskates, since you'd have to factor the fact that the gamepads couldn't be too expensive, or else that will reduce the sales of supplementary gamepads for the Wii2/Café.
The bad news is that less than a year later there will be new consoles from Microsoft and Sony setting new benchmarks for performance so there goes the AAA 3rd-party games compatibility down the toilet, again.
If it isn't indeed just a ~3Ghz 3-Core POWER, with a 320 ALUs at best RV7x0 series GPU and ~1GB of RAM, then it would indeed make multi-platform support for 3rd parties an uneasy task once again.
With that said, one of the biggest issue Wii had compared to the other two, as far as multi-platform support went, was that it was very different technology wise. PS360 had multi Gigahertz multi-core CPUs and fully programmable GPUs with extensive Shader supports, whereas the Wii was stuck with a 700Mhz single core CPU and a GPU that supported fixed function TnL and some color combining effects.
The PS4/X1080 won't be that different technologically from the Wii2/Café, they'll just be a lot faster/more capable. In other words, it would be possible to imagine creating a game engine around Café's capabilities, and then port it up to the PS4/X1080 without it having to be fundamentally different (Like it had to be with Wii/PS360, if you wanted to make use of the PS360 capabilities). The Café version would obviously have to cut many more corners, have lower resolution textures, PCF samples, lower LOD, etc. But the overall game could be similar.
Of course, if the game relies on heavily on its aesthetics or in the sheer number of objects displayed on screen, and if the developers/publishers are not willing to sacrifice the image quality too much, it would sadly rule out a Café version.
Personally, I think what would be the biggest reason why publishers/developers refuse to port their big "core" games to the platform would be entirely related to the perceived success of "core" games on the platform early on. If, for some reason (founded in reason or not), the publishers decide that the Café is not a viable platform for some games, it will be a case of self-fulfilling prophecy. No or little core games > no audience cultivated > no success of the rare core games > no more core games.