The company's CPU market share dropped from 23% to 17.7% in the past year. AMD's weakness in desktop CPU performance, coupled with its limited mobile PC processor line, all contributed to the company's decline in share.

AMD also continues to post quarterly losses, though the $100M loss reported recently was a significant improvement.

Meanwhile, we're seeing the first reviews of Intel's new Core i7 processors, with the official product launch of Intel's latest architecture slates for later this month. Previously, AMD had argued that Intel's approach to quad core CPUs was flawed, since a Core 2 Quad CPU really consisted of a two distinct Core 2 Duo dies in one package. Some of this was reflected in certain server applications, where Intel's shared front-side and off-chip memory controller allowed AMD CPUs to better run certain applications sensitive to memory latencies.

But only a few apps benefited from that, while most server and workstation applications running on Intel outpaced those on AMD platforms.

That didn't prove the case in most desktop applications, as lower clocked Intel CPUs easily outpaced higher clocked AMD parts, which also drew more power. On the desktop, AMD was relegated to the budget and low power play, trying to make the best of its strengths.

Core i7 pretty much eliminates whatever slender advantage Opteron may have had. Core i7's integrated memory controller and QuickPath interconnect (QPI) replicate in functionality, if not in specific technologies, AMDs strengths. Core i7 is "true" quad core, and QPI's high bandwidth and point-to-point nature removes all the bottlenecks inherent in the older front side bus approach.

There have been a few tidbits of good news from AMD recently. AMD's move to shed its fabs should improve its balance sheet. Meanwhile, on the GPU front, AMD's ATI subsidiary has been doing well, punishing Nvidia's balance sheet. The company continues to roll out cost effective GPUs, including the recent release of the Radeon HD 4830.

Still, the company's bread and butter is CPUs. AMD's current product mix is still heavily mired in its aging 65nm manufacturing process, but the 45nm ramp seems to be moving right along. The latest 45nm Opterons have been sampling already, and 45nm desktop CPUs based on the Shanghai core look to arrive in early 2009. Early indications, however, are that Shanghai won't really recapture the performance crown, or even achieve performance parity. But it will at least enable AMD to be competitive across a wider range of products.

Competition is a good thing. It was, after all, competition from AMD's x86-64 product line that force Intel to move away from the point heat source that was Prescott to the much more efficient Core architecture. Intel's never been particularly good at competing with itself, with its tendency to believe the direction its currently heading is the right one. Like any large company, Intel occasionally needs to be prodded, so that it will sit up and notice different ways of accomplishing the same goals. And AMD has certainly been good at that prodding.

Still, something seems to be missing from the new AMD. Gone are the wild men, the Jerry Sanders and Fred Webbers, who loved tweaking Intel, but were willing to back up the tweaking with interesting new approaches that were nimble and made efficient use of AMD's lean resources. AMD's last CTO, Phil Hester, was a very middle-of-the-road kinda guy--someone that would be good to have a beer with, but not someone who had new insights into how CPUs could work, or inspire other architects to that end.

So what's next for AMD? Will the newly streamlined AMD pull of a cool new Fusion product, melding the best of the CPU and GPU in a whole that's greater than the sum of the parts? Or will it simply limp along, envying Intel's seemingly limitless resources and forgetting that being nimble and having new insights doesn't require a huge budget?

I can't say for sure, but I can say that an Intel without an AMD will be less driven and less innovative. And that would be unfortunate.