wbmw, then we will have to disagree on that point and revisit it during the holiday selling season. Here is my take on your specific concerns:
Some of the hardware you mention - NICs and IDE, particularly - are needed for servers and were optimized by the Linux folks already. I expect that Microsoft will certainly have competitive server product with Server 2003 64-bit Edition (and those drivers are common with XP-64), they cannot afford to look bad when benchmarked against Linux.
I do agree that there are many users who recycle old hardware into new PCs. I do not expect much Athlon64 sales into that group - they will purchase Celerons, P4s and 32-bit Athlons. Actually, not only is it fiscally questionable to pay the premium for a 64-bit PC and OS and then put in old peripherals, but 32-bit drivers just won't work in a 64-bit OS. Maybe some will make do with generic drivers, but not many. This just isn't the target customer.
Who is the target customer? The more well-heeled customer that Intel traditionally targets, the one who gets a complete new system every 2 or 3 years and then gives the old system to a charity (or young grade school relative).
I am making one leap of faith: That there will be a handful of very compelling 64-bit video games that will drive a bunch of people to get Athlon64 machines for the home. Those people will choose their peripherals from the supported set, and that supported set will be broad enough to meet almost all needs.
Right now my investing reflects my thinking in this area - I am holding shares long, and I have no short term trading positions in AMD. I'm expecting H1 to be unspectacular. If 90nm comes on line in time to be effective for H2 then this company can really take off in the closing months of the year.
