Can anything slow down Intel? Jerry Sanders hopes his latest deal will finally do the trick. On Oct. 20, the chairman of Advanced Micro Devices Inc. (AMD)--Intel's chief competitor--announced a surprise deal to buy tiny rival NexGen Inc. for about $900 million in stock. The big prize: a coming NexGen chip that, for the first time, could rival the latest of Intel Corp.'s microprocessors, the silicon brains of most personal computers. Crows Sanders: "This will be the closest we have ever been to Intel."
Closer, but no cigar. For all of Sanders' optimism, the marriage of convenience shows one thing clearly: how pmwerful Intel has become and how weak its rivals remain. The merger may spur even more consolidation: Rumors suggest that Cyrix Corp., another Intel rival, could be acquired by IBM or Korea's Samsung Group--although they all deny a deal is in the works. Concedes Steve Tobak, Cyrix' marketing vice-president: "People are really getting discouraged about the prospect of an Intel alternative."
A lot is at stake--and not just because PC microprocessors are a $12.6 billion market. Intel microprocessors command up to 80% profit margins--margins chipmakers such as AMD need if they're to stay on the technology treadmill.
So far, despite a decade of trying, the Sunnyvale (Calif.) company has only captured an estimated 8.7% of the market this year; Intel has 85.6% (chart). AMD remains stuck in the lower-margin and rapidly shrinking 486 chip market. Worse, it's losing ground: On Oct. 10, AMD reported third-quarter profits that were off 33%, to $56.2 million, from 1994 on sales that were up only 9%, to $590.4 million. The profit drop was due mostly to Intel's relentless price-cutting on its 486 and flagship Pentium chips. And there's no relief in sight. Says Cowen & Co. analyst Drew Peck: "It's going to be grim for AMD throughout 1996."
Things would be even grimmer for NexGen had it stayed independent. It has the only rival to the Pentium chip on the market. But because its chip has a different arrangement of wires coming out of it, it can't use all the standard components Intel's Pentium can use. As a result, few major PC makers have bought it. On Oct. 24, NexGen said that its cash position had dropped a shocking 83% in just one quarter, to $5.5 million.
Meanwhile, Intel keeps barreling along--happy, says Intel Senior Vice-President Paul S. Otellini, to have one less rival. On Oct. 16, it logged a 41% jump in third-quarter profits to $931 million on sales that were up 46%, to $4.2 billion. And on Nov. 1, it will introduce the Pentium Pro, a faster successor to the Pentium.
"THAT'S BRILLIANT." Just six months ago, it looked as if Sanders might be closing the gap. As clones of Intel's 486 chips raked in record profits, engineers were on track to introduce a Pentium rival by yearend. But troubles were brewing. AMD's new K5 chip, designed to be 30% faster than comparable Pentiums, fell well short. Mike Johnson, AMD's chief designer, admits the young K5 team failed to devise adequate software tests, so the chip ran real programs more slowly than expected.
Month by month, delays mounted. AMD had to take engineers off the next-generation K6 project to help with K5, making it doubtful the K6 could be completed soon enough to challenge Intel.
NexGen was running into trouble, too. Once dubbed "Nevergen" for its nine-year record of empty promises to come out with an Intel alternative, it got on track in September, 1994, by introducing the first Pentium rival--and getting IBM to manufacture it. But sales remained minuscule next to Intel's. Despite raising $48 million in an initial public offering in May, it was running out of cash. By midsummer, it was mulling everything from a secondary stock offering to an investment by Samsung.
Then Marshall Cox, a NexGen board member and former employee of Sanders, got involved. He arranged for Sanders to meet Vinod Dham, NexGen's chief operating officer, and NexGen Chairman and CEO S. Atiq Raza in early September. Sanders was impressed with NexGen's technology--especially a chip in development that promised to match Intel's next-generation Pentium Pro. But he had big doubts: The chip didn't fit into a standard Pentium socket. It needed custom PC designs likely to turn off most large PC makers.
But Dham had a surprise. As soon as he arrived in May, he redesigned the chip so it might even do Intel one better: It could plug into the same socket as a Pentium and thus use standard PC parts--yet it would run as fast as the new Pentium Pro. Sanders' response, recalls Dham: "That's brilliant."
Indeed, in NexGen, AMD gains ammunition to fight Intel. Mainly, it gets a chip to match Intel's best at least a year earlier than it would have on its own. And it is now only a year behind Intel--vs. three years back on the Pentium. Says Linley Gwennap, a former chip designer and editor of newsletter Microprocessor Report: "This really puts the heat on Intel to keep executing."
Problem is, AMD and NexGen still have months of work to tweak the new chip to fit in Pentium sockets and run on AMD's factory process. Sanders makes no excuses: He needed NexGen, and without it, he faced falling behind Intel for good. "We just had to find a way to punch above our weight," he says. But unless Sanders can parry better than he has so far, Intel's punch still may knock him woozy again.