Richard, if Stuart sold (or bought) shares, sooner or later he'll have to file a form 4 or something like that. The "disadvantage" of having more than 4.9% of a company's stock (I think I could live with this kind of disadvantage (G)). Stuart stated quite clearly that his RMBS position is a very long term one, thus do not expect him to change his opinion or position unless the major story changes, and the major story is still that INTC wants the CPU to be the speed limiting factor in a PC, not the DRAM, and currently, RMBS is the only game in town. INTC makes their big margins on their leading edge devices, they can load premiums on these and the systems involved can accept premium memory as well. Everyone here is with white knuckles because the current Camino fiasco indicates that RMBS may not reach the low end PC as soon as some wanted. Well, I have some news guys, RMBS was never going to go into the low end until all the cream was scooped from the high end, and that could very well be 18 months or so after introduction of RMBS into the high end. 18 months of "learning curve" in the semi business is a very long time. By then RDRAM might very well be produced with .15 or lower features, reducing (per switching) the heat dissipation and thus operating temperature (of course, they'll go to higher switches density, but that is another story). The current problems with signal integrity are not "intrinsic" IMHO, these are design problems equivalent to babies teething problems. When you go to 800 MHz type of speeds, you got to be very religious about line impedance matching, concerns about degradation of signal rise time (and "fall time") due to parasitic capacitance and line resistance, and since the signals are "read" both on the rise and fall curves, it should be expected that much more stringent design principles will have to be applied. Carl is absolutely right in noting that engineering tolerances which were quite acceptable at 200 MHz will not be satisfactory at 800 MHz, but I do not remember seeing anywhere that tolerances cannot be tightened (Carl, I still remember the days were passive components with 10% tolerances were the premium one (G)). These cost a premium and thus initially, will go only in premium systems, once the "bugs" are ironed out, it will spread down the feeding curve, IMHO.
That said, I still think that the market is going to punish RMBS, whether or not it is RMBS' fault, and thus my interim forecast of the low $60 before we double from there, stands.
Zeev |
