I say Nintendo, since they are pushing hard to get their next gen console ready to continue the Wii's domination.

The Cell isn't that hot - in fact, is one of the coolest processors on the market dispite its high performance. Even at 5 GHz it would idle at 63C and have a V-Core of 1.3. And these are IBM figures for the 90nm process. Look what they managed to do on the 65nm process.

Seriously one SPE at 2 GHz consumes ONE SINGLE freakin Watt and idles at 25C or the like. Surely it would need to be a PPE to be used on a console alone (since the SPE is what we call a 'dumb core') but still Ninty did very bad keeping the GC chip.

@Haxxiy: If i'm not mistaken, it runs cool for being so big.

I can't see that being a dumb move - 360 has the same chip in practice and PPE in PS3 shouldn't be that different.

Maybe Nintendo chose the technology that fit their needs the most? It seemed to work for them considering sales and profit figures. Technology does advance, but companies shouldn't be blinded by impressive figures.

There is nothing wrong with the PowerPC architecture, and the PowerPC 970MP (a dual core PowerPC processor) was used in the Mac Books and Mac Mini in 2004/2005 running at 1.5GHz to (IIRC) 2.5 GHz ...

With that in mind, I suspect that AMD sees a potential CPU/GPU hybrid as being very popular for low cost laptops and ultra-small and/or inexpensive PCs. At the same time, I'm sure all console manufacturers are looking at how physics APIs are starting to take advantage of the massively parallel architecture of the GPU and could see some benefit of that kind of parallelism being moved towards the CPU. While the console manufacturer and AMD may not want the exact same product at the end, there is a lot of similarity between their goals and AMD may have been fairly agressive in targeting a console manufacturer (possibly offering lower licencing fees or what not than IBM was willing to) in order to get the R&D money to build the groundwork for the product they hope to produce.

@HappySqurriel: Not to mention to lock customers for their new product. Of course, depending how early it's out. But bigger manufacture rates offer bigger competetive advantage in terms of cheaper price to customers.

Maybe. The standard CBE has 300 million transistors so that was as big as GPUs back then.

And perhaps it can look like a dumb move now. Back when Nintendo was risking to launch a new console with minimal improvements and revolutionary controls, besides then selling at profit they needed also fidelity from the GC's userbase to assure they would have no loss and a pre-set userbase. So indeed the Wii was planned pretty much like a 'canon fodder' - if it outsells every other console, nice and good. If does not, at least they were making profit.

But now that Wii stands on its two legs it does not look so smart after all. It could be getting 80% of userbase instead of 50% if at least was strong enough to handle PS360 multiplats with similar graphics/gameplay.

  A) I don't think equal to the PS3/360 graphics would help the Wii that much.  It would raise the price point and the gamecube generation showed that equal graphics don't sell Nintendo consoles.  They would quite possibly lose share if they went with a more expensive high performance box.  Remember when the gamecube had HD outputs before HD was ready for prime time and noone used them (well I did, but almost noone else did) so they got rid of them on the later versions?  I think Nintendo learned their lesson there.

  B) the PowerPC platform can easily handle anything the 360 or PS3 can do.  The cell, which the PS3 uses, was developed along with IBM and is relatively close to the power PC architecture.  IBM makes tons of high end processors for things like enterprise servers and super computers, they can easily handle a chip for a mid range console.

@Haxxiy: I do agree that Nintendo had prepared for potential failure with Wii. They could easilly had went though this gen with another 20M sales and make a nice profit. But you're not seeing the big picture here:

If Wii could handle the same games with PS360, devs would be porting their games back and forth. This would really cause Wiis controls just to be a gimmick.
Now that Wii can't handle PS360 games, devs can't just port their games to Wii or the opposite (though, multiplatform developement from the beginning makes this a lot more viable option) and the devs are forced to rebuild their games so that it's practically the same to do a completely new game.
Now Wii differentiates itself from the competition by its software and as time passes, Wii keeps selling and devs shift their focus to Wii, Wii starts to get similar titles with PS360 (and the majority of games), Wii is getting more and more desirable option even for the PS360 audience. In a way, it doesn't matter to Nintendo whether people only buy a Wii or Wii to go along with PS3 or 360, since Wii is always going to beat both with its games library. Of course, Nintendo knows it can't reach the graphically most demanding audience with Wii, but this group is the minority anyway.

So, at the surface it may look like that in retrospect Nintendo did a dumb decision, but digging it a little deeper reveals that cheap developement costs combined with expensive porting, Nintendo has locked a shitload of 3rd party exclusives to itself - which was the plan from the beginning and makes the strategy of lower powered hardware genious.

And, Nintendo wouldn't be at any higher marketshare due to them selling everything they manufacture at the moment.