By Olga Kharif For as long as anyone can remember, Advanced Micro Devices (AMD) has been the company that would never say die. It may have boasted only a 20% share of the microprocessor market to Intel's (INTC) 80%. It may have turned a profit only every five years or so. But it sometimes ran close or even with Intel in chip innovation, and it usually matched or even bested the microprocessor king on price (though Intel may have pulled some punches to avoid the attention of antitrust regulators). And AMD regularly signed up big-name customers, such as Hewlett-Packard (HPQ).
The prolonged tech downtown has taken a toll on AMD, however. It hasn't earned a profit since 2000, when, partly thanks to a new chip called the Athlon, its revenues rose 60%, to $4.6 billion. Its revenues this year will be almost 30% below that -- and only one-ninth those of Intel. AMD's cash on hand has doubled since last December, to $1.1 billion -- but mainly because of more borrowing. Its long-term debt has ballooned to $2 billion from $1.37 billion a year ago.
THE 64-BIT QUESTION. As analysts look forward to a revival of the microprocessor market in 2004, many of them think AMD has lost its resilience in this business -- the only major product on which it now makes a profit. Chris Danely, an analyst with J.P. Morgan, believes that AMD won't even achieve profitability for all of 2004, the anticipated peak of the current semiconductor cycle. Its best hope, he and others say, is to diversify -- a tough trick considering AMD's already stretched resources.
Its dilemma is dramatized by the reception for its next-generation desktop Athlon chip, which debuted in September. The first 64-bit microprocessor from either AMD or Intel (Apple's latest PowerPC chip is a 64-bit processor from IBM [IBM
]), the Athlon 64 theoretically can crunch data much faster than competing chips, making it ideal for buyers who use their PCs to view and edit video. That kind of high performance is why AMD's similar server processors, released earlier this year, have been successful in that market.
However, the desktop chips are a different story, especially in the short term. Athlon 64 won't ramp up to high-volume production until mid-2004, says Dean McCarron, co-founder of chip consultancy Mercury Research in Cave Creek, Ariz. What's more, at the moment some PC makers complain that the new chip doesn't outperform AMD's older chips, much less Intel's best, says Tai Nguyen, an analyst with investment bank Susquehanna Financial Group in San Francisco.
That's partly because most PC programs haven't yet been modified to take advantage of the chip, Nguyen adds. John Crank, the Athlon's product manager, nonetheless argues that PC makers "are extremely pleased with" the performance it delivers for the price.
NO SMALL PROBLEM. The Athlon 64's slow takeoff means AMD will likely continue to struggle in the microprocessor market. Its stock rose to a 52-week high of $14.56 the day after its Oct. 16 announcement of a much lower-than-expected $31 million loss on sales of $954 million in the third quarter, vs. a $254.2 million loss on sales of $508.2 million a year earlier. Yet analysts say AMD's chip production costs are higher than those of Intel, thanks to its less favorable economies of scale. Adding to that problem, AMD is way behind its rival in adopting the latest manufacturing processes.
The penalty it pays for using older chipmaking techniques is striking. AMD plans to move from 200 millimeter-diameter wafers -- silicon plates from which chips are stamped -- to 300-mm plates in mid- to late 2005, a transition that Intel is making right now. Such a shift allows for more chips to be made from every wafer, at cost savings of up to 30%. Similarly, AMD trails Intel in moving to making smaller chips, and the smaller the chip, the more that can be stamped out of every wafer.
Intel is also about to introduce new processors that will prove tough competitors to the Athlon. In the fourth quarter, it will release the Pentium 4 Extreme Edition, a chip for PC gamers that will go head-to-head with AMD's new Athlon 64FX. Also in the fourth quarter, Intel will start shipping its latest Pentium for normal PC use, called Prescott. Next year, analysts expect Intel to crank up Prescott to a clock speed of 4 gigahertz (vs. 3.2 Ghz for the fastest chip today).
JUST A MEMORY? So daunting is the competition from Intel that some investors have given up on the idea that AMD can match up in microprocessors. Peter Hofstra, senior investment analyst at the $115 million AIC Diversified Science & Technology fund, whose largest holding is Intel, says "we wouldn't own AMD, period: They've lost the [processors] war." For proof, he cites Intel's gross margins, which peaked at 64% in 2000, but which Hofstra thinks could reach 65% within two or three quarters -- because of decreased pressure from AMD.
Seemingly, that leaves AMD CEO Hector de J. Ruiz with the option of building a "new AMD," as he has called it -- one that should focus less on competing with Intel in processors, says Nguyen, and more on markets that might help AMD turn a profit on a sustainable basis.
Already, many analysts view AMD as a memory maker more than a processor maker. Sales of flash, a type of memory used in cell phones, now account for half of AMD's sales, up from 37% in the third quarter of 2002. Flash doesn't make money for AMD yet, in part because its price fluctuates wildly, but it could be profitable by next year's first quarter, analysts estimate. Because AMD is the largest maker of one of two primary types of flash memory, it will benefit from a rise in the overall flash market, which analysts think will grow 47%, to $11.3 billion, this year. That's why J.P. Morgan and UBS Warburg have both upgraded AMD's stock from underperform to neutral.
FLASH SPIN-OFF? Focusing on both flash and processors could make sense for AMD, particularly as cell phones get smarter and require more processing power. Still, it might not have the financial power to swing both: With Intel's continued marketing push, AMD needs more resources to promote its processors. So it might have to spin off its flash unit to get the money to compete, argues Jim Handy, an analyst at Semico, a chip-industry market researcher and consultant in Phoenix. (AMD declined to comment on such speculation.) Alternatively, AMD might issue more debt or shares to keep its balance sheet strong or diversify further.
Whatever else it does, most analysts expect AMD to persevere in the processor market, where Intel will continue to give it some slack to avoid antitrust problems. But that could leave it where it has been for years: Playing second fiddle to a rival it can never hope to catch. By Kharif writes about technology in Portland, Ore.