In part one of this two-part series, we look at the evolution of AMD from a second-source supplier for companies using Intel processors towards CEO Hector Ruiz's ideal of a "premium" chipmaker that could sell to the likes of Dell and Intel.
On June 10, 2000, Advanced Micro Devices (AMD) wanted to party—and party big. The company’s CEO, Jerry Sanders, arranged to rent out the entire San Jose Arena (now called the HP Pavilion) and then paid big bucks to bring in Faith Hill and Tim McGraw, the husband-and-wife country music superstars.
Employees “could bring anybody, your wife, your kids, your friends—it was big doings. There were celebrations, gifts and awards,” recalled Fran Barton, who served as AMD’s chief financial officer from 1998 to 2001. The boss even got in on the fun. “[Sanders] was on a high wire, he did a unicycle ride. It was totally Hollywood. He could really put on a show when he wanted to put on a show.”
And why not celebrate in style? AMD’s successful Athlon chips—Ars named the Athlon its "CPU of the Year" in 1999—had finally put the screws to archrival Intel, and in 2000 the company earned nearly $1 billion in profits.
By 2005, years of solid chip design and technological execution had the company walking with a swagger, as seen in marketing stunts which challenged Intel's then-current server processors to a "dual-core duel.” Nowhere was this attitude more apparent than AMD's 2005 lawsuit against Intel for anti-competitive business practices.
Doubters didn't think the good times could last. "I rode through a few such cycles and can recall the zenith of decadent exuberance and breathtaking spectacle in 2000 when AMD's then showman CEO, Jerry Sanders, clad in tight leather pants, shirt cracked open to the waist, descended to the stage of the HP Pavilion in a cherry-picker and announced that AMD's stock would soar to over $100 per share and that another [chip fabrication plant] would be built in Austin,” recalled Bill Bushnell, a rank-and-file veteran software engineer at AMD, in an e-mail to Ars. "Neither occurred."
AMD has been on a notable drop for nearly a decade now. To put it mildly, 2012 was a rough year: AMD lost over $1 billion, effectively wiping out its $471 million profit in 2010 and its $491 million profit in 2011—its two most profitable years in the last decade. Over the last 15 years, AMD has sustained a net loss of nearly $7 billion, and the company has been downgraded by credit rating agencies, burned by lower demand for PCs (and hence, for its products), and even called "un-investable" by one Wall Street analyst.
Last month, the company even sold (and then leased back) its corporate headquarters in Austin, Texas for $164 million as a way to make some quick cash. After years of technical stumbles, Intel now runs circles around AMD in desktop, laptop, and server CPUs, while newcomers like Qualcomm, Samsung, and Nvidia have used their low-power ARM chips to shut AMD out of the burgeoning smartphone and tablet markets. And critics charge that the company still has fundamental structural problems that go beyond technical missteps.
"There's no control on spending—even now, one of the problems is if you take a look at the salary structure," said Atiq Raza, the company's former president, chief technical officer, and chief operating officer, in a conversation with Ars. "[AMD is] a sinking ship, fundamentally. I really am sorry for [current CEO Rory Read]. He's a well-intentioned person but the ship has a huge hole in it. That $164 million is going to go in no time."
Raza calls the company's decline "one of the great fumbled opportunities of our time." How did AMD go from the most successful period in its history to one of its bleakest—and does the company have a fighting chance going forward?
“I couldn't fail”
When Jerry Sanders was 18, he was beaten in a street fight by people who left him unconscious in a trash can. “They fractured my skull, broke my nose—that's why you're photographing from the left,” Sanders told Daniel Marrow of Computerworld (PDF) in 2000. “So my nose is not more crooked than normal, ribs—I mean just a disaster. And they left me to die. They literally left me to die.”
But after three days in a coma, Sanders rallied.
"I once said 'I can die, but I can't fail'," he told the San Francisco Chronicle for a profile. "What I meant was, I was always going to give it my all. I couldn't fail, because failure wasn't an option. I would die before I'd fail."
Sanders went to work for Motorola and Fairchild Semiconductor, but he didn't work long for others. He opened AMD for business on May 1, 1969—when he was just 33 years old. He quickly cemented a reputation for being defiant and flashy, and he kept a poster in his office which read, "Yea, though I walk through the valley of the shadow of death, I shall fear no evil—for I am the meanest son of a bitch in the valley."
“He’s a mix between Don Quixote and Indiana Jones,” former CFO Barton said, “a swashbuckling idealist, not afraid to tilt at windmills and dream the impossible dream. His whole career, his impossible dream was to battle Intel. And he battled very hard for decades.”
AMD began life as a second-source supplier for companies using Intel processors. Companies like IBM didn't want to rely solely on Intel for one of the primary components in their computers, so they licensed AMD to produce versions of processors like the 8088 and 80286. While these CPUs were manufactured by AMD (and, in some cases, AMD was actually able to clock the CPUs higher than their Intel counterparts), almost everything about their designs came from Intel.
Beginning with Intel’s 80386 in 1985, Intel stopped giving AMD access to its designs. AMD had to forge its own way, soon producing 386 and later 486 CPUs that were essentially reverse-engineered versions of Intel’s parts.
In 1990, a Merrill Lynch analyst called AMD “dead,” largely because of the time needed to reverse-engineer the chips, just one of many times an analyst would say this about the company. Intel’s 386 was released in 1985, for instance, but AMD’s didn't appear until 1991; Intel’s 486 was out in 1989, while AMD’s wasn't on the market until 1993. The increased complexity of Intel's CPUs made the reverse-engineering process more difficult: Intel's 486 had 1.18 million transistors, and the original Intel Pentium chip had 3.1 million in 1993. The Pentium line quickly grew more complex, leading to the 5.5 million transistor Pentium Pro in 1995 and the 7.5 million transistor Pentium II in 1997. Compared to the 134,000 transistors in the 286, these were enormously difficult chips to replicate, and though AMD sold millions of its versions through aggressive pricing, the strategy wasn't viable in the long term.
Rolling their own
To stay competitive, AMD created its first original x86 design, a Pentium-equivalent called the K5. This CPU was roughly comparable to the Pentium's performance, but it came to market in 1996, a year after the Pentium Pro.
A tiny Silicon Valley company named NexGen, meanwhile, was doing what AMD couldn't: taking on Intel with a relatively miniscule team of just 60 engineers and a little funding from IBM. The company was headed by Atiq Raza, a brilliant Pakistani engineer educated in the United Kingdom and the United States, who had arrived in the US in 1979 with only $3,000 in cash to his name. (By 2009, Raza put his Palo Alto estate up for sale at $24 million; he ended up not selling it.) Raza wanted to take on Intel, too, but he knew that he couldn't do it at NexGen's current size.
Microsoft CEO Bill Gates showed an interest in NexGen and had a 45-minute meeting with Raza that stretched into a four-hour dinner. Gates suggested that Raza speak to Sanders since AMD owned a chip fab but needed a better product to build in it. Raza said he would like to talk with Sanders but said he didn't know how to reach the AMD CEO, much less get him to pay attention.
"The first time I was introduced to Jerry Sanders, he didn't even pause long enough to hear my name—he kept calling me Raja," Raza told Ars. "He did that nonstop for 45 minutes telling me how AMD is going to crush me. I stayed there with a smile on my face, but how do I communicate with such a person? Bill said, 'Jerry should not be underestimated—he has three qualities: he's smart, he's extremely egotistical, and he's completely random.'"
Gates picked up the phone.
"Without Bill making that call, we would not have had that momentum," Raza admitted.
AMD purchased NexGen in 1995 for $615 million. Raza became president and chief operating officer of AMD and was widely viewed to be Sanders’ eventual successor—according to the Wall Street Journal, Sanders even dubbed Raza the "the Michael Jordan of microprocessor design." AMD also picked up NexGen's in-progress CPU, the Nx686. After some difficulties adapting the design for manufacturing in AMD's facilities (rather than the IBM factories for which it had been designed), AMD in 1997 produced a CPU that was competitive with Intel’s Pentium II but which could be inserted into the then-prevalent Socket 7 motherboards used by the original Pentium, allowing for easy in-place upgrades.
This CPU was called the K6, and marked a turning point for AMD—it could compete with Intel in both speed and price.
“The Pentium MMX is now a completely obsolete CPU!” enthusiast site Tom’s Hardware proclaimed in April 1997. “The K6 233 will be priced lower than the Pentium MMX 200 and it's faster than this CPU in almost every respect... The K6 is faster than any current CPU under Windows 95 and will probably be just as fast as the Pentium II at the same clock speed.”
The K6 later morphed into the K6-2, which debuted a year later, and then the K6-III, which lasted for the bulk of 1999. While those transitions might have looked straightforward to the public, things weren’t so smooth behind the scenes.
Life on the ragged edge
Sanders' management style was idiosyncratic, and it created both loyalty and friction. Patrick Moorhead, AMD’s vice president of marketing from 2000 to 2010, still remembers the day he interviewed for the AMD job with Sanders. “I parked my 1990 Toyota Camry next to his Bentley convertible,” Moorhead told Ars. “We talked about making a difference and shaking up the industry and quite frankly I just fell right into that. I absolutely loved that pitch.”
But Raza quickly grew unhappy with his boss' approach to work. Since 1979, Sanders had kept his Beverly Hills office, commuting up to Silicon Valley each week and returning to Southern California Friday through Monday.
"If [Sanders] was constantly engaged, that would be acceptable," Raza said. "He was not engaged. He would be upset if I called him too late in the evening or too early in the morning. I'm a 24/7 kind of guy."
As The New York Times reported in 1989, AMD's corporate culture had also lost the thrifty habits it had at its founding. "Mr. Sanders's increasingly lavish style also became contagious," wrote the paper. "Salesmen who did not wear Rolex watches were likely to find their less-expensive time pieces thrown away by high-level executives who wanted the company to have a certain image. It was an atmosphere, former employees say, that encouraged loose spending throughout the company—a far cry from the early days at Advanced Micro, when Mr. Sanders once rejected a request for an electric pencil sharpener because it was too costly."
Raza also became concerned about the spending, which he came to see as extension of Sanders' personal lifestyle. Shortly after AMD bought NexGen, Raza visited Sanders at his expensive Los Angeles home.
"I said, 'It's a beautiful house,'" Raza said. "He said, 'I got in a competition with Madonna. We had a fight and I outbid Madonna.' I said, 'OK…' And he said, 'I understand what you're thinking. I spend more than I make. I always have spent more than I make.'" Raza said that the comment hit him like a punch in the gut. "And I said, 'I hope you don't do it at AMD,' but he did," he added. (Sanders did not respond to our request for comment.)
Financial problems mounted. As AMD stretched out the life of the K6 line, former CFO Barton recalled how the company flitted from one near-disaster to another every few weeks.
“We would have Monday morning operating meetings and someone would say, ‘We have a glitch. The yields have dropped precipitously and we don’t have anything to sell next week,’” Barton said. “The financial ramifications of that are awful. We have nothing to sell! We’re doomed! Then everybody would get their assignments and somewhere in the next day or two there would be an e-mail or phone call saying, ‘We fixed it. And it’s even better, sales will be higher.’”
Each time this happened, Barton said, Sanders firmly delegated assignments to each department. On numerous occasions, Raza himself only got a few hours’ sleep to isolate the repeated problems and get things back on track. On the financial side, Barton and his team had to make a decision about disclosing each problem to investors or to the Securities and Exchange Commission (SEC).
"We were running quarter after quarter on the ragged edge and had been for decades," he added. "It was a real culture change of ups and downs. “For the rest of my career nothing bothered me [as much] anymore.”
Despite these problems, the K6 and its revisions set the stage for one of AMD's most successful architectures: K7, dubbed the Athlon.
Athlon ascendant, and Sanders’ search for a successor
The K7 architecture was successful in all the right ways for AMD. It performed well when stacked up to the Intel chips of the day, and its clock speed also scaled upward readily from 500MHz at its introduction all the way up to 2.33GHz by the end of its run. This allowed AMD to take the fight to Intel during the heady days of the MHz race, where ever-increasing clock speeds served the same marketing purpose that ever-inflating CPU core counts do in processors today. The K7 also scored important symbolic wins—AMD was able to beat Intel to market with a 1GHz processor, for example.
However, just weeks after the K7 debuted on June 23, 1999, Raza left AMD amid rumors that he had a major falling-out with Sanders. Analysts were left scratching their heads.
"It's certainly going to reduce confidence in the company's ability to compete," Linley Gwennap, chief analyst at Cahners MicroDesign Resources, told the Los Angeles Times in 1999. "Atiq was the one guy in upper management that really understood the technology behind the K6 and K7.”
As Raza tells the story today, his boss insisted on building a fab in Dresden, Germany, over Raza's objections. (That fab, which still operates today as part of AMD spin-off GlobalFoundries, was completed in the spring of 2000.)
"The trouble in the entire economic model was that AMD did not have enough capital to be able to fund fabs the way they were funding fabs," Raza said. "The point at which I had my final conflict was that [Sanders] started the process of building a new fab with borrowed money prematurely. We didn't need a fab for at least another year. If we had done it a year later, we would have accumulated enough profits to afford the fab in Germany. He laid the foundation for a fundamentally inefficient capital structure that AMD never recovered from. I told him: don't do it. I put the [purchase orders] on hold. He didn't tell me and accelerated the entire process."
Both Raza and Barton recalled, independently of one another, one of Sanders' mantras: "Real men have fabs." Raza called this comment "simultaneously a sexist remark and the most stupid thing you can say," and he saw the fab decision as one of Sanders' "significant acts of irresponsibility." After he quit, Raza never spoke to Sanders again.
Despite Raza's departure, the K7 certainly made a splash: AMD's net sales soared from $2.5 billion in 1998 to $4.6 billion in 2000, and the company rented out the San Jose arena to celebrate the boom. But Intel continued to challenge AMD with chips like the Pentium III, and it soon swept the world with its slick “Intel Inside” marketing campaign.
With Raza gone, Sanders needed a new heir apparent. His gaze fell on Hector Ruiz, then-head of Motorola’s Semiconductor Products Sector. The two knew each other at a professional distance as a result of a 1998 cross-licensing deal giving Motorola access to AMD’s Flash memory for cellphones and other portable devices, while Motorola would help AMD get into copper-based (e.g., faster) microprocessors.
Sanders seemed to identify personally with Ruiz’s challenging upbringing and drive to succeed—Ruiz grew up in a Mexican border town but crossed the border each day to attend a high school in south Texas, where he ended up valedictorian of his high school class. Late in the summer of 1999, Sanders invited Ruiz out for dinner at an Austin steakhouse.
As Ruiz tells the story in his forthcoming book, Slingshot: AMD's Fight to Free an Industry from the Ruthless Grip of Intel, Sanders opened by explaining that his job was better than Ruiz's.
Ruiz took the job; he was announced as AMD’s new president and chief operating officer in January 2000. Just over two years later, Sanders retired and Ruiz was named CEO. When Ruiz took over the reins, he set about cleaning house—keeping some veterans but also getting rid of top executives (CFO Fran Barton left the company in 2001) and bringing in new talent. (Ars spoke with Ruiz in two phone interviews in February and March 2013.)
Ultimately, Ruiz found that the company wasn’t really a CPU maker anymore; rather, it was a Flash memory maker that happened to also sell microprocessors. AMD’s record profits in 2000 were due largely to its ability to sell memory to gadget makers, so Ruiz set about re-directing AMD.
This proved more difficult than expected; despite Sanders' swashbuckling ways, Ruiz found that "Sanders had unconsciously imbued AMD with a 'second-best' mentality." As Ruiz put it in his book, "Sanders was something of a paradox. On the one hand, he was full of energy and would never accept that AMD could fail. But on the other hand, he never seemed willing to put the long-term effort into developing a strategic plan. As a result we had a mediocre customer plan, a hit-or-miss reputation, and no global strategy at all... Simply put, no one inside AMD believed AMD could beat Intel in the marketplace or even mount a serious threat—not even the head of the company."
But by the time he took over as CEO, AMD’s market share hovered around 15 percent, more or less where it had been during the previous half-decade. Ruiz tried to make AMD into a "premium" chipmaker who could sell to companies like IBM and Dell, not just to the "value" segment of the PC market. AMD had a chance to pounce, and it did so in a high-end market that AMD needed to bolster its profit margins: servers. The company's K8 architecture was designed from the start to power servers, and its Opteron chips were one of its resounding successes.