(page 2 of 2)
Is there a light at the end of this tunnel? Doug Freedman of American Technology Research thinks there is: He rates the stock a "buy." "I don't think what we're seeing with AMD has much do with its fundamentals as it does with nervousness about the credit markets," he says. "But that's not to say AMD doesn't have its fundamental challenges."
The indicators of turning the ship around are encouraging in a few spots. PC sales in the second quarter, typically a seasonally weak one, bounced back stronger than most people realized, Freedman says. "If PC sales come in anything close to seasonal patterns, we'll have a strong year," Freedman says. And AMD has recaptured some of the market share it lost in the first quarter. Encouraging numbers from other PC makers, such as Hewlett-Packard (HPQ), which reports earnings Aug. 16, could be revealing.
AMD also has a new graphics-chip family and a server-chip family, code-named Barcelona, coming soon. They are aimed precisely where they need to aim—at higher-priced products. Barcelona is a server chip with four cores—a core being the central brain on a chip—that will be sold for a higher price than a typical chip that goes into a PC or notebook, and as such will generate higher margins per unit than the chips sold for lower-end systems. A successful Barcelona launch would go a long way toward generating some much-needed cash.
But AMD's main challenge with Barcelona will be fighting the perception that it's not as fast a chip as previously hoped. Barcelona is expected to ship with clock speeds in the range of 2 to 2.5 gigahertz, when analysts and customers had been hoping to see speeds closer to 3 GHz. AMD says it expects to release speedier versions of the chip by the fourth quarter. After Barcelona, AMD has another chip, code-named Shanghai, in the pipeline for 2008. This one will be built on more advanced 45-nanometer manufacturing technology, allowing smaller transistors on a chip and thus more computing power than current 65-nanometer manufacturing technologies. After that is a 45-nanometer chip called Sandtiger, set for 2009, which will have eight cores.
Intel isn't sitting still, though. Among other things, it plans to introduce chips built on 45-nanometer manufacturing technology before the end of this year. Code-named Penryn, the chip will use a new technology that Intel says will reduce its power needs while boosting its speed. "AMD's biggest problem is that it doesn't have much on its road map that's really exciting until 2009," says Jim McGregor, an analyst at InStat/MDR in Scottsdale, Ariz.
How much lost market share has AMD managed to retake? Mercury Research in Cave Creek, Ariz., pegged AMD's total share of the market for PC, notebook, and server chips at 22.9% in the second quarter to Intel's 76.3%, which for AMD is an improvement over its 18.7% share in the first quarter, and even slightly ahead of its share in the second quarter of 2006.
Dean McCarron, head of Mercury Research, says much of AMD's sales strength, however, came from the lower end of the market, where profit margins tend to be leaner. "AMD demonstrated that the first quarter was an artifact of arithmetic and not as dramatic as it appeared to be," he says. "Its share started to come back, but the average selling prices for both [AMD and Intel] were down pretty significantly."
AMD does have one advantage that Intel doesn't have—at least not to the degree AMD does. AMD has long made its server chips pin-compatible with prior models, so servers using AMD chips are easy to upgrade—just swap in a new one to replace the old. It's often not as easy to upgrade servers using Intel chips, McCarron says. Intel's chips are less often pin-compatible between generations, so servers are more difficult to upgrade. "I think the Barcelona release that's coming before the summer is over will kick off a wave of server upgrades," he says. That would boost the volume of chips with higher margins and give AMD some much-needed cash.
Hesseldahl is a reporter for BusinessWeek.com.