Noobs lol. I was selling cpu's in the late 90's and we sold 10 Intel cpu's to every 1 AMD cpu. Its now more like 3 Intel cpu's to 1 AMD since core 2. Before core 2 it was 2 AMD's to every 1 Intel. AMD isnt going anywhere, they survived this long at number 2.
You need to keep in mind that AMD has now taken on LOTS of expenses that it can't easily shed... your logic makes about as much sense as saying "Well, Henry Ford made it just selling 10 cars a year at first... don't worry about Ford... they'll be fine today selling 10 cars" I'm not a doomsday prophet or anything, but AMD is in dangerous financial territory. If it doesn't turn the ship around in 2 or 3 quarters, it'll go belly-up and will be acquired by someone like IBM (not necessarily a bad thing in my book)
Ah, the typical American Business accountant outlook: "We lost 2% on our stock for this quarter! We need to cut expenses, kill our customer support funding, and start hounding our managers to do their jobs and make money for us!"
It takes money to make money, and all things come in cycles. Japanese businesses have known this for years. Your typical Japanese business is not nearly so concerned with this quarter, next quarter... or even next year. They are focused on the long-term goal and what needs to happen before they can get there.
That aside - All I've really seen behind Phenom is speculation, for the most part. According to what I've read - the core doesn't even really begin to function until about 2.6 GHz. More importantly, though, is how instructions are executed within the core and special instructions the processor can handle. I can compile processor-specific code that can take advantage of the various instructions available to each architecture. For desktops - this is of minimal concern. Servers, though.... being able to take three standardized instructions that take four clocks a piece to carry out, to a single seven-clock instruction can make worlds of difference.
And with the way CPUs are beginning to take shape, architecture is only part of the battle - precisely how data is distributed amongst the cores (if at all) will determine who wins out on performance. Bang for the buck as well as supporting components (how good are the chipsets?) determine far more than the CPU alone.
It also wouldn't surprise me if we begin moving away from the X86 architecture. But that's another discussion.