Personally, I question whether Phenom II is a good enough effort from AMD. It is just barely good enough to hold its own with Intel chips that have already earned back their R&D and it will not hold its own against Intel's next generation. So, where does AMD have a chance to make money in all of this? The reason they need to come up with something clearly better than Intel is not to beat them in a technical sense, but to have something they can sell for a premium to recoup some of their loses from the past several years.
As I've already said, AMD will have more trouble when Ci5 shows up. The thing with the latest CPU's is that AMD have made very small changes compared to the first Phenom: a die shrink and added cache, that's about it. If they had more time and money they would of course have done more. I'm really curious what AMD's plans are for the next two years, until the next CPU shows up.
AFAIK, the core logic have changed very little since the K7. They pretty much added some very good features like IMC, HTT, 64 bit, and L3 cache. I think they could have done more with the K10.
From a business perspective, I think AMD should be using 45nm to chase Atom rather than i5/i7. The whole point of Atom was to make something that was incredibly inexpensive to manufacture that could play in emergiing markets where power efficiency is more important than raw performance. Playing in those markets was the whole reason AMD acquired the Geode stuff, but they seem to have largely dropped the ball. For whatever reason the folks running ATI seem to have a better grasp on how to do things than those at the mothership. If AMD would concentrate more on being profitable rather than have equivalent solutions to Intel, they'd have a better chance to survive long term and be around and keep Intel on their toes. Maybe if they can actually get this Foundry spinoff done, it will help them focus.
They are doing that, although in their own way, and it's called Athlon Neo.