Send in the clones
Before "Intel inside" took over the world and the age of the chip giant's 90 percent market share in microprocessors had dawned, there were two other companies making Intel-like microprocessors, Advanced Micro Devices and Cyrix. When AMD announced recently that it had solved its yield problems, the stars were finally in formation for its return.
The whole strategy of these two companies was to produce 100 percent Intel-compatible clone chips that were 25 percent to 50 percent lower in price, hoping to garner share in the markets vacated by Intel as it moved forward. Then, in 1994, AMD started to suffer delays in its production of the K5, and the company was in serious trouble. Meanwhile, Intel's Pentium chip enjoyed massive success, creating the stranglehold on the PC industry that we see today.
Then the whiz-kids of the Pentium designer team from Intel, Vinod Dham and Atiq Reza, formed NexGen to design and sell x86-compatible microprocessors. At a time when AMD was struggling for survival, it made a gutsy move, scrapping internal K6 plans in favor of purchasing NexGen and adopting its chip as the K6.
Now fast-forward to more recent times. Plagued with low yields, AMD has struggled to manufacture its chips. PC manufacturers want security of supply in volumes, and AMD completely missed the K5-Pentium clone cycle. Faced with no alternative to the seemingly inexhaustible supply of Intel chips, OEMs jumped wholesale on the "Intel inside" bandwagon, giving it a dominant market share in microprocessors and helping to create the Wintel duopoly. In the midst of a ramp-up of the next cycle with the Pentium II, it seemed, for a time, that history was repeating itself as AMD continued to face yield problems for the K6 processor.
But change does occur. Right out of the blue, during the Christmas season of 1997, the sub-$1,000 PC arrived. The "segment 0," as Intel calls it, changed the landscape of the PC market once and for all. Beginning with the consumer segment, the sub-$1,000 PC is expected to reach up to 30 percent of the market. Intel initially ignored the trend toward cheaper PCs, responding rather belatedly with its Celeron offering, which was announced only a week ago. The Celeron is nothing but a stripped-down version of its Pentium II without its on-chip level 2 cache, a piece whose absence sorely limits its performance. The Mendocino, a more tailored version of the Celeron chip, is due at the end of the year.
This situation presents a real opportunity for the clonemakers in general, and for AMD in particular. Given the fact that the quality of the NexGen designer team and chips have never been in question, and given the window of opportunity that Intel left open, demand for cost-effective Intel-compatible microprocessors is not an issue.
The whole strategy of AMD has been to sell Intel-like performance at price points that are 25 percent to 50 percent lower. We now have a unique situation in which Intel actually has a lower-performance chip than AMD, in a large volume segment of the market. Moreover, half of the AMD chips due to arrive next quarter will perform at speeds of 300 MHz or higher, further increasing the performance gap with Intel's Celeron until Mendocino arrives at the end of the year.
When AMD announced recently that it had solved its yield problems, the stars were finally in formation for its return. AMD currently is in the midst of the steepest ramp-up of microprocessors in history, from 1.5 million units this March quarter to an estimated 11.5 million units by the end of the year. Remember that AMD had a market share of about 30 percent in microprocessors in the not-too-distant past, while the combined current non-Intel share is below 10 percent.
So the upside potential for AMD is huge. Indeed, in the words of the AMD management: "We are back!"
Dan Niles is a managing director at BancAmerica Robertson Stephens covering the PC and semiconductor industry.