AMD, Intel and Sun Microsystems

Yesterday I listened to three analyst conferences, and you can read my summaries of them if you like: Advanced Micro Devices (AMD), Yahoo (YHOO), and Sun Microsystems (SUNW). I currently own stock in AMD; once I lost some money holding Sun waiting for a turnaround back in 2003.

Often you learn about general trends in an industry from listening to management trying to explain their results at an analyst conference. For instance, a number of semiconductor chip makers had slow third and fourth quarters of 2006. Several claimed that end-user demand was fine, but that manufacturers who incorporate their chips into products like cell phones had tightened up on inventories. Thus slower sales. So everyone is predicting growth in 2007, unless the global economy sinks. It is a reasonable story, but still might not be true, especially for an individual company.

Yesterday I learned something important about AMD's long and short term outlooks during the Sun conference. Sun has worked with AMD closely for about 3 years now, producing high-quality, relatively low cost servers based on Opteron chips. Monday Sun and Intel announced a deal: in return for also using Intel's new line of Xeon chips, Intel would allow anyone, including Sun's competitors, to license Sun's Solaris operating system, which almost every tech pundit acknowledges is the best Unix around. Solaris is way better than Red Hat Linux or Novell's Linux, so they say (I run Windows, so this is hearsay). By getting some of HP and IBM's and even Dell's customers to switch to Solaris, Sun can create an important income stream for itself (I think Solaris is now free but you have to pay for technical support).

With respect to AMD, Sun's management made it clear, saying it several times during the conference, that they expect more sales of servers with AMD Opterons in them, not less. They are adding Intel Xeon based servers which they expect to cut into sales of competitor's Xeon servers. Solaris is already available for AMD Opterons.

The timing of this move is important as well. There will be no significant shipments of the Xeon servers until the second half of 2007. By then the new generation of Quad core Opterons will be competing with quad-core Xeons. Both companies will be at 65 nm process, leveling that playing field. Xeons will still have external memory management; Opterons their famous, but improved, Hypertransport on-chip memory controller. AMD says their chips will be the clear leader in performance and performance per watt and performance per dollar. We won't know until independent testing is done, but if it is true, then AMD will continue to gobble up market share and maybe even have gross margins that pump up the stock price. But this is not a stock for short-term investors. While Intel is still stuck in short-term mode, the management at AMD thinks long term. They were not afraid to say that to analysts yesterday. Their honesty is refreshing.