1996 was one of the worst years on record for DRAM vendors, and therefore for their equipment suppliers. Yet, in 1996, DRAM bit volume soared. Volume of chips does not correlate to equipment purchases.
Future increases in bit volume will come not only from increased output in current production lines, but from the increases from 64Mb to 256Mb and beyond. These changes require new equipment.
No, but the people paying for the equipment care a lot. 1997 and 1998 demonstrated that the money to build those $2 billion fabs has to come from somewhere.
To stay competitive, new fabs are going to have to be built, or old ones will need retooling. It seems that the pressures from 300mm, copper, and feature shrinks are building to the point where this is increasingly true, and discretion in these purchases will become more and more inelastic.
Also, the PC is not dead, but I think other types of devices are becoming a larger segment of the chip market. Foundries are booked solid, so there is room for growth in fabs. Of course, the industry will overshoot this demand and set up the next bust cycle... :)
Just my .02
Doug |
