Why AMD is suddenly in the chips
By David Kirkpatrick
(FORTUNE.COM) -- Advanced Micro Devices, the computer industry's spunky, pugnacious comeback kid, has had an amazing few weeks. On January 20 it announced that last year's fourth quarter was its first profitable one in over two years. Subsequent news underscores why.
AMD's unique 64-bit Opteron chip, its first-ever chip for servers, is becoming so successful so fast that it is forcing the industry's biggest players to change strategies. It's also allowing AMD to keep its average selling prices high despite the usual pressure from Intel. IBM and Sun are shipping servers with Opteron. Customer interest is growing quickly. And late last week bombshells emerged from both Hewlett-Packard and Intel, the two companies who partnered a decade ago to take another path toward 64-bit computing called Itanium.
HP quietly confirmed that it is likely to soon start selling Opteron-based servers. Separately, Intel President Paul Otellini effectively announced that the chip giant will soon introduce chips mimicking AMD's approach. In both cases these are moves the two companies had long resisted and/or vowed not to make. It is by no means the kiss of death for Itanium, but rather a fabulous, though reluctant, endorsement of AMD's strategy on the part of both companies. Both clearly hate to take this path, but they hear what customers are saying and have no choice.
AMD's advantage is clear. Intel's mainstay processor chips for both desktop and server computers process data in chunks of 32 bits at a time. Itanium, intended for heavyweight corporate servers, digests 64 bits at once, but it doesn't run the same software as Intel's Xeon server chips, so users have to choose one platform or the other.
AMD, by contrast, designed a way to add 64-bit capability to a 32-bit chip, so customers essentially got two for the price of one. They had a choice -- they could run old applications the old way, but when they wanted to move to 64-bit software, they could simply add those applications on the same machines. In designing such chips, AMD also cleverly included a few features that make them extra-efficient, even for 32-bit applications. The strategy was AMD's first fundamental departure from its longtime path of simply offering a lower-priced version of Intel's X86 chips.
But as always, AMD priced its chips below Intel's, even though this time those Intel chips were only 32-bit. In short, AMD offers more value for less money.
Itanium is making steady progress, but at the high end of the market. AMD is bringing 64-bit processing to the masses, and they seem to want it. (The company's similar chips for desktop computers are known as Athlon-64, and are also doing well.)
Businesses and consumers stand to gain from the shift to 64-bit processing because 32-bit computers can't take advantage of more than 4 megabytes of memory at a time. But two important kinds of software can benefit from access to more memory -- database-intensive applications for business, and for consumers, state-of-the-art games. Demand for the chips should only continue to grow, as Microsoft is expected to deliver a much-delayed 64-bit version of Windows for PCs and servers sometime in the third quarter of 2004.
The latest good news for AMD came last week, when the European Union approved a massive $682 million subsidy from the German government for a $2.4 billion factory the chipmaker is building in Dresden. Now AMD can afford to make the chips its customers appear to want.
Recently, at the World Economic Forum in Davos, I spoke to AMD's upbeat CEO Hector Ruiz, who got the job only in April 2002 but has put an amazing stamp on the company in the short time since then. "We have never been in such a strong position," he told me in his low-key way.
Ruiz was not only celebrating the developing success of his 64-bit chips, but the company's recently-won position as No. 1 in a crucial segment of high-end Flash memory chips. Ruiz crafted a joint venture with Fujitsu on Flash in which AMD owns 60 percent, so it can include the sales in its own numbers.
AMD's success in these chips, which are critical for cell phones and other contemporary appliances, has come at Intel's expense as well. Intel was the leader in this kind of Flash a year ago. AMD's stock, by the way, has tripled since then.
For all of AMD's recent successes, its history suggests that future triumphs are not guaranteed. The relatively tiny $3.5-billion-in-revenues company has crept up on Intel before, only to be squashed by the $30 billion giant. Intel will likely do what it can, with pricing, promotions, and industrial muscle, to elbow AMD back into the world of losses, where it has been for most of its 30-plus-year history.
Ruiz concedes that all those past disappointments represent "definitely a burden we're going to have to overcome," but he professes confidence.
He says when he joined AMD he found a company hugely talented in technology but sorely lacking in business discipline, "not just in manufacturing, but in terms of business objectives," he explains. "I said 'We've got this wonderful airplane but every time we go through turbulence everybody wants to change everything.'" He has focused especially on creating more discipline around costs.
This is probably AMD's make-or-break year. I hope they make it. For all Intel's wondrous abilities, it's nice to see a two-horse race.
E-mail David Kirkpatrick at firstname.lastname@example.org.