Your conclusion to the AMD page reads "...Another possibility is that low cost Opteron based systems could infiltrate the business world via departmental level purchases beyond the direct control of corporate IT policy makers. This bottom up approach was highly successful in the early spread of the Linux operating system."

The even better concluding statement would be "This bottom up approach was highly successful in the spread of Intel into the server business."

In the section on Sparc you write:
"The admission by computer giants like HP and Compaq that they couldn’t justify or sustain the increasingly expensive effort of designing house brand processors and building systems around them increases the pressure on Sun management to justify further investment in SPARC."

I have a problem with "couldn't justify or sustain"; Compaq and HP haven't demonstrated a viable strategy based on a commodity server chip.

In fact, I'd argue that Compaq management and BOD concluded that it couldn't survive no matter what it did. Compaq first recognized that it couldn't survive as a PC company and started buying non-PC companies of all sorts, but most notably Tandem and DEC. And then when Compaq management couldn't transform Compaq from a PC company to a computer company, the BOD pulled the plug on the company.

HP appears to be trying to transform itself from a computer company to an Intel reseller, but Dell is much better at that. I can only guess at how long it will be before the BOD decides to pull the plug and turns HP into a printer/scanner company, with a Unisys style consulting and service business on the side.

The failure of DEC and Compaq, and what I think will be the failure of HP (and in the spirit of disclosure, I have way too much HP stock to want my prediction to come true) "prune" the tree of possible business alternatives open to Sun. Dropping Sparc in favor of an Intel alternative will lead Sun to the same future as DEC and Compaq, and likely SGI and HP. DEC, SGI, and HP have suggested that they were willing to support other vendors developing competing chips and actually supporting them, but in all cases any alternative chips were undercut.

I guess what it comes down to is the almost insignificant role that a cpu architecture, chip implementation, and/or process and manufacturing prowness have in determining the success of a cpu architecture. (Process and manufacturing prowness definitely determine the profitability of a chip company.)