If nothing else, the impending demise of K-Mart should broadcast a valuable message to retailers and manufacturers across the fruited plain: Complacency in a niche market, no matter how lucrative the position might seem at the moment, is not a strategy for long-term growth--or even survival.
Which brings us to AMD. Over the last 15 years, this plucky, not so little, CPU maker has evolved from being a second source for Intel processors to breezing along as a very competent processor manufacturer in its own right. In the past it's been short on innovation, acting more as a gadfly to Intel and the safety valve that's kept CPU pricing within a sane region for desktops. Its customer base is analogous to Linux's core group, composed of both those who appreciate the performance and low price of its products and, in this case, of those who simply hate Intel.
Conceivably, AMD could continue on in this mode for years to come, however it would eventually fall victim to the K-Mart Principle. Low margins and an encapsulated market are an inescapable downward path. Of course, growth is the solution, but how do you measure it? Selling more CPUs into its existing market segment is no indicator for AMD. We can turn to IBM Fellow Ravi Arimilli for a clue: "In my opinion, it's patents that indicate a company's upward growth. Innovation drives the industry." That sentiment is echoed by AMD's Bill Siegle, chief scientist at the company, and is borne out by AMD's patent track record over the last two years. It ranked fifth in patents among U.S. based companies and 14th and 12th worldwide in 2000 and 2001, respectively. It should come as no surprise that IBM was ranked first, by a wide margin, during the same period but, in a somewhat telling turn of events, Intel only ranked 19th and 18th for those years. (Statistics reported by IFI CLAIMS Patent Service.)
Although AMD has been in the server market for a scant six months, Kevin Knox, AMD's Director of Enterprise Segment Marketing and Business Development for North America, has been pushing that advantage--hard. Within that time, he has been leveraging AMD's existing desktop base to inch its technology farther up the corporate ladder, with some very positive results. Electronic Arts has been recommending AMD-based desktop and workstation systems to its development, sales, finance, and marketing departments. Mental Images GmbH & Co. KG, has certified the Athlon MP for workstation and server platforms using its rendering software. AMD's Multiprocessor Platform is both Linux- and Novell-certified. And a server cluster running 78 Athlon processors (PRESTO III) at Tokyo's Institute of Technology was listed among the Top 500 supercomputers at the June 2001 International Supercomputer Conference. One side of the IT acceptance coin is respectability, and that's snowballing.
Rounding up a larger base is important for AMD because its Hammer processor will debut somewhere around the end of the year. Hammer is the pivotal point for AMD's future success and, if you give credence to the Mercury News story, Intel has secret weapon against AMD chip--which discusses how Intel's new "Yamhill" technology, slated to appear in the next Pentium chip, will allow the CPU to run legacy apps as well as memory-intensive programs--some folk at Intel are not unaware of its potential. The Hammer is a dual 32/64-bit CPU that, according to AMD, doesn't yield devastatingly low 32-bit performance in exchange for its 64-bit prowess--something that Intel's IA-64 (Itanium) can't duplicate. Hammer offers investment protection in the low to mid-range 32-bit CISC server market and its arrival has generated quite a bit of anticipation--something absent for the IA-64. (Quoted in the Mercury article is a Gartner Dataquest report which estimates that "In the quarter ending September 30th, Intel shipped about 2,162 Itanium chips, with 2,000 of those going to IBM for two big servers that use 1,000 Itaniums apiece." There is no safety in those numbers.)
Don't expect to see AMD in the stratosphere of RISC servers, at least not soon. Do expect it to continue to emphasize the inherent value of its Hammer line, by handholds and inches if need be, within the low to mid-range server community. The other side of the IT acceptance coin is a verifiably good TCO rating, and AMD thinks that won't require a back flip thanks to the Hammer's x86 backward compatibility and a forward march to 64-bit processing.
Bill O'Brien is a freelance writer and frequent contributor to CNET and ZDNet. He writes Tech Update's weekly hardware column.
