Can AMD avoid the Intel graveyard?
Fortune; New York; Apr 14, 1997; Stewart Alsop;
Issue: / 7
Start Page: / 169-171
ISSN: / 00158259
Subject Terms: / Microprocessors
Competition
Market strategy
Microprocessors
Computer industry
Competition
Classification Codes: / 8650: Electrical, electronics, instrumentation industries
7000: Marketing
9190: US
Geographic Names: / US
Companies: / Advanced Micro Devices IncTicker:AMDDuns:04-863-4059
Advanced Micro Devices Inc
Intel CorpDuns:04-863-4059
Abstract:
A commentary discusses the market prospects for Advanced Micro Devices (AMD). AMD bought Nexgen in early 1996. The plan was to combine Nexgen's intellectual property with AMD's production expertise to create clones that could compete with Intel chips. AMD is now introducing the first product of that merger, the K6 processor, which is just about to be shipped. AMD says the K6 is everything that Intel's Pentium Pro is, and a little bit more. Unfortunately, even if AMD succeeds in selling K6, and gets new customers, and retains old customers, and even doubles its current share of the microprocessor market to about 15% or so, it will not shake Intel's hold on the architecture of the personal computer. Intel is just too far ahead to be challenged.
Copyright Time Incorporated Apr 14, 1997
REMEMBER THAT COLUMN I WROTE last fall, complaining that there wasn't anything left to write about in Silicon Valley, since everybody but Microsoft and Intel was imploding? ( "Please Save Me From My Editors," October 28.)
As I requested, you very generously took your time to write my editors, telling them to give me a break-I was right, you said, it was getting increasingly difficult to find interesting competitive issues to discuss. (I seem to have continued to find new subjects to write about, however.) But I did get one E-mail myself, from a publicrelations guy at AMD, who said: "Watch us over the next few months ... We might have the stuff columns are made from."
Turns out, he was right. AMD does have a story to tell about how it might compete with Intel. In fact, even Wall Street has begun to pay attention: AMD's stock has risen from $13 in September to $44 as this column goes to press.
That kind of good news has not been the norm at AMD, which stands for Advanced Micro Devices. You may remember AMD as the company with an early license to build exact copies of Intel microprocessors. Not surprisingly, Intel and AMD ended up in court a few years ago, arguing about just what that license did and didn't entitle AMD to do. After much to-ing and fro-ing, AMD won the battle, kind of-but it lost the war. The company still makes copies of Intel's earlier microprocessors, but there isn't much of a market for those chips. Meanwhile, Intel has moved on to newer processors that AMD is not licensed to copy.
AMD made news again last year with a Pentium clone called the K5. Since AMD had no right to the Intel design, it built the K5 from scratch to mimic precisely the Pentium's functions. But it delivered the K5 months late. That teed off industry heavyweight Compaq, which had already annoyed Intel by agreeing to buy a lot of the chips. AMD missed its revenue target by $400 million, resulting in three fiscal quarters of losses.
Flops like that help explain one big question in the computer industry, whether any company can compete by making unlicensed Intel clones. In the past five years a few other companies, most notably Cyrix and Nexgen, have taken a shot. Cyrix has had an up-and-down experience. Currently it is having a little rebound, since Compaq just introduced a low-cost home PC with a Cyrix processor. But the company is still tiny in silicon terms, with $183 million in revenues in 1996 (vs. Intel's $21 billion).
As for Nexgen, well, AMD bought the company in early 1996. The plan was to combine Nexgen's intellectual property with AMD's production expertise to create clones that could compete with Intel chips.
Now AMD is introducing the first product of that merger, the K6 processor, which is just about to be shipped. AMD says the K6 is everything that Intel's Pentium Pro is, and a little bit more. The company says the new chip is good enough to make AMD a competitive threat, allowing it to grab a substantial portion of the market for Intelcompatible microprocessors and possibly even giving it a way to lead the industry out from under Intel's thumb.
Sounds like a column to me! Can AMD be the microprocessor company that finally marshals the resources to take on Intel? Can it establish the credibility needed to convince customers that they can rely on AMD as they rely on Intel? The answer is no, by the way. But AMD's pitch is really interesting. I recently spent an hour getting it from the horse's mouth, the horse in this case being AMD's voluble founder and CEO, Jerry Sanders.
Sanders admits that past critics of AMD have had a point-AMD has always been just slightly behind Intel because it had to wait for Intel to do something, and then re-create it, either by license or by reverse engineering. This time, he says, everything is different.
First, he says, AMD is ready to compete as a business. It has cleaned up its act in its other divisions, which make an array of chips for products like laser printers and wireless phones. According to Sanders, AMD is now making money in everything except microprocessors. Indeed, Sanders says, there are many people on Wall Street who wish AMD would stop beating its head against Intel and get out of the business of making microprocessors for PCs. But Sanders says he isn't ready to retire!
In addition, through creative financing, AMD has lined up the manufacturing capacity around the world to make more than 20 million chips a year. This is important because a PC maker's worst nightmare is first to get Intel angry by buying chips from AMD, and then find that AMD can't fill the order. (Remember Compaq?)
Second, AMD now has a competitive product. The K6 is its equivalent of Intel's Pentium Pro. But the K6 is a single-chip processor, while the Pentium Pro consists of multiple chips. This matters to manufacturers making lots of PCs, because installing a single chip is cheaper than working with a multiple-chip package. Intel has been developing an improved, single-chip version of the Pentium Pro, called Pentium II. But it is unlikely to deliver Pentium II until sometime after AMD delivers the K6.
So Sanders believes he's got his company poised to compete. He also believes that Intel has made two crucial mistakes. The first error was the Pentium Pro, which may not be different enough from the original Pentium to justify a higher price. Customers (both computer manufacturers and end users) have not figured out why they want a Pentium Pro. So Pentium Pros haven't upstaged Pentiums as quickly as Intel might have liked. That's given AMD and other chip vendors a bigger window of opportunity in which to catch up to the Pentium, which is still the chip that powers the vast majority of most new PCs sold.
Intel's second mistake, says Sanders, is its design for Pentium II. Pentium II will require a different kind of motherboard socket than the ones today's Pentiums plug into. That means that manufacturers wanting to use Intel chips will have to get redesigned motherboards. AMD, on the other hand, will offer the same kind of performance as Intel in a chip that will work with existing PC designs.
So there you have it. Sanders is making a proposition to PC manufacturers that seems to make sense: Come to us and you'll get an inexpensive, single-chip equivalent to the Pentium Pro. You won't have to create a new socket. And you'll get your chip from a company with the global capacity to supply even the largest customers.
In any other business, I would say that's the kind of story that would lead to measurable market share! As noted above, it is certainly a story that Wall Street has swallowed-hook, line, and sinker.
Unfortunately, even if AMD succeeds in selling K6, and gets new customers, and retains old customers, and even doubles its current share of the microprocessor market to about 15% or so-even if all that happens, AMD still won't shake Intel's hold on the architecture of the personal computer.
Why? Basically, Intel's just too far ahead to be challenged. Intel has another generation to come after the Pentium Pro, and another one after that. So even if Intel did mess up the Pentium Pro-at this point it's too early to telland even if it turns out the company shouldn't have changed sockets for Pentium II, Intel will use its billions and billions of dollars to get things right with the next generation of microprocessors. And the big computer manufacturers, which rely on Intel for the overwhelming majority of their processor chips, will give their major partner plenty of time and will work with Intel as they plan future generations of chips and PCs. AMD will be back where it started, a little chip company trailing behind in Intel's wake.
AMD will have served to shake Intel up a bit, perhaps making it a little more humble about how well it understands its customers. AMD will have sold a few hundred million or even a few billion dollars' worth of microprocessors and made its shareholders happy for a while. AMD will have given chip analysts and columnists something to write about. But Intel will still be king of the hill.
Unless, of course, AMD can do it twice in a row. It will beat Intel to market with a single-chip version of a Pentium Protype chip. If it can do the same with K7, its alternative to Intel's next-generation chip (code-named Merced, or P7), maybe people will start taking AMD seriously. Maybe AMD can get between 20% and 30% of the market. Maybe then computer companies will begin to think of AMD as a junior partner, the alternative to Intel they've always wanted. If that happens, watch out! Hey, who knows? Maybe I'll have to write another column about it!
[Author note]
STEWART ALSOP is a partner with New Enterprise Associates, a venture capital firm. Neither he nor his partnership has financial interests in the companies mentioned. Alsop may be reached at stewart_alsop@fortunemail. com
Reproduced with permission of the copyright owner. Further reproduction or distribution is prohibited without permission.