Sharp, incisive thinking. Also don't forget that AMD is cutting R&D like crazy and lost many key technical people (i.e. most of the K8 team, the low power core team, etc.). AMD cannot work miracles.
Of course they can, I've been part of teams (in software) that have done it. We were very professional and highly motivated to succeed. We worked or collective butts off. But, you can't do that on every project - eventually you would burn out, unless you are an extreme Type A. However, we were not nearly as limited as AMD financially - we had pretty much everything we needed. Does AMD have everything it needs - the most obvious answer would be no, but we really do not know how AMD is allocating it's resources or how well automated design is improving their efficiency. We don't know how inefficient their prior designs were, and, if through better knowledge of their target market, they have been able to make great strides in improving the efficiency of their newer designs. While I wouldn't take the numbers that AMD is publishing at face value (let's face it, this is PR, not science); I wouldn't totally discount them either - not without having some inside knowledge of just how AMD has transformed themselves.
And speaking of that transformation; Dirk and Ruiz make a mess of AMD and spent a decade doing so, I think, especially given the product development lead times in this industry, that expecting AMD's new management to completely revamp AMD in just two years, is expecting an even bigger miracle (a point mrmt brought up). IIRC, Keller himself doesn't expect his most meaningful changes to come till 2015 (again because time to design and bring up a new microprocessor). You have to remember that even though the A6/X came out after he left Apple, he was in charge of that product during it's design phase, he probably made key decisions on the architectural direction of the A7. We are talking about ~3 years to design a CPU/SoC and bring it up in manufacturing (+ time to ramp to good yields depending on process maturity) - or longer for a completely new architecture. I don't know, but I am inclined to think, that at this point in time, AMD only has enough engineers to have one team working on each product class (big core, low power, ARM, etc.) as opposed to two or three - that will clearly lengthen their development cycle since there is no overlap in design cycles. IMHO, YMMV, yada, yada...
1) The perf/watt numbers are PR, and as is often the case with PR, they are the absolute best numbers marketing could find from tests that were not likely done under real life conditions. That doesn't mean AMD didn't make very substantial improvements on Jaguar - we'll have to wait and see.
2) The new management has only been in place for around two years, I think they deserve another 12-18 months to see if the changes they are making will succeed because of the nature of their product pipeline and the total wreck they inherited.
3) AMD's ability to stop it's marketshare decline is hampered by many things, most are internal and still have roots in AMD's recent past. Some are external including the still relatively weak world wide economy and the pressure being applied by Intel in it's own attempt to transform it's product mix to maintain it's high margins**.
4) Full disclosure: as an investor, I wouldn't touch AMD stock with the proverbial ten foot pole. That doesn't mean that as a technophile, I can't be hopeful that AMD will have a softer landing than VIA, even if that's their current trajectory.
5) Intel is going to maintain it's lead over AMD for the foreseeable future. Intel's financial strength and technical depth are simply unparalleled in the industry; they are the 800 lbs. gorilla. I don't have a problem with this. I can hope for the best for AMD and respect Intel and their products at the same time.
** thank you for pointing out that latter point in another post.