It is a problem. Will there be demand for the output of the new 300mm wafer .25 line fabs? Intel has to reverse the decline of PC sales, and this is certainly possible (and AMD, Cyrix, SGS-Thomson are willing to do their part). Reports are that PC's are selling like hot noodles in China (2 million this year), and those Taiwan fabs are very close and will produce very cheap pieces. Most of the new fabs will be able to shift from memory to logic rather simply. Of course the demand for imbedded mpu's is increasing very rapidly, and while this market is not by any means closed to x86 processors, it is likely to grow most rapidly for RISC chips like Hitachi's.
A possibly factor is the limitations of the settlement which allows AMD (according to AMD) to use Intel's (early?) x86 technology and any further extensions (including MMX technology) until 2000. Most of the early patents must have expired and can't be further enforced even by agreement. Does it stop then? Is there any limit on foundry operation in unlicensed foundries. IBM can foundry for Cyrix, because IBM is licensed. Interesting problem, but not worth speculating about without
the data.
One important point is that it is not in AMD's interest to seize market share. With their capacity and cost of production Intel can hold as much of the market as it wishes by setting prices. Intel will always drive prices in the x86 market, and any attempt by AMD to seize market share (with an outperforming chip)will result in disaster for AMD and a modest fall in margin for Intel. Read AMD's plan as requesting permission, rather than making threats. Of course, if the K7 is a Merced buster, all bets are off. If it is merely another super x86 chip it won't help all that much. |
