Copyright Economist Newspaper Group, Incorporated Dec 5, 1998IT IS a fair bet that more than half of the Ics bought this Christmas in America for less than $1,ooo will have 
AMD rather than Intel inside. Not content with this seasonal miracle, Advanced Micro Devices, the perpetual also-ran of the computer processor business, is bidding to loosen Intel's grip on the more lucrative high end of the market too. It could well succeed.
The clearest sign yet of the markets' new confidence in AMD was a credit-rating upgrade last week by Tom Kurlack of Merrill Lynch, doyen of Wall Street's semiconductor analysts. Mr Kurlack, a famous bear who, in 199o, pronounced AMD to all intents and purposes dead, was not the first to revise his views. Since August, shares in the 29-year-old Silicon Valley firm have more than doubled in value.
For most of its existence, AMD has lived in the shadow of the deal that it did with Intel in 1982. To power its early Pcs, IBM had decided to buy Intel's new x86 chips, but wanted a second supplier to keep Intel under control. Under the terms of the agreement, Intel got the contract, but had to share its intellectual property with the smaller AMD. Intel broke the arrangement, AMD sued, and thus began nearly a decade of bitter legal battles between the two companies. Intel's Andy Grove and AMD's Jerry Sanders became the Jarndyce and Jarndyce of the computing world.
Although Mr Sanders was largely vindicated, the conflict distorted AMD'S business, absorbed untold management energy and undermined investor confidence. By selling cheap, reverse-engineered Intel clones, AMD staggered on, sometimes quite successfully, especially if Intel was late to market with a new product. But despite the support of computer makers groaning under Intel's dominance, trying to hitch a ride on the back of a grumpy 800lb gorilla was proving a precarious form of existence.
Eventually, under a settlement in 1995, AMD gave up any rights to Intel microcode. It was confident that its home-grown K5 would give Intel's Pentium a run for its money, while a new $1.8 billion fabrication plant ("fab") in Texas would meet demand and match Intel's manufacturing expertise. It did not. Design faults put the K5 more than two years behind the Pentium, and the Austin plant lay largely idle.
What saved AMD was an inspired acquisition. In 1996, it paid $625m for a company called NexGen that had a great design for a sixth-generation processor, needing only a few tweaks to make it capable of volume production. It also had a gifted engineering team whose leader, Atiq Raza, shared Mr Sanders's burning desire to take on Intel.
Mr Raza's K6 processor, launched last year, has been a smash hit. Early production problems prevented AMD from meeting demand, but output this year should top 12m. It is the K6 that has made the profitable sub-$i,ooo Pc possible, turning it into the fastest-growing part of the market, accounting for 25% of all sales.
Compaq, Hewlett-Packard and IBM are selling as many AMD-based Pcs as they can crank out. Consumers seem to be indifferent as to whether the "intel inside" logo is there or not-which, given the billions of dollars that Intel has spent creating what it hoped would be one of the world's most valuable brands, is perhaps surprising. Fortunately for Intel, corporate buyers have proved more loyal. It may be that they fear (wrongly) that networking Pcs that run on different processors is risky.
Can AMD maintain its momentum? To stay on track, it must double output and successfully launch its high-performance K7 processor in the first half of next year. Microprocessor Report, the industry bible, says that the K7 knocks spots off Intel's forthcoming Katmai. Yet any delays or production glitches would be disastrous. Next year should be a bumper year for PC sales, because millennium-bug worries will mean the junking of old hardware; but the industry fears a bust soon after. AMD therefore needs to make hay while it can.
Its ambitious target is to double its share of the PC market to 3o%. With a new fab in Dresden, Germany, coming on stream, it should be capable of churning out 4om units by 2001. PC makers will be eager to help it get there, to keep up the pressure on overmighty Intel. But a fast maturing PC market and a wounded Intel will give AMD little respite.
My Research
