AMD's and Cyrix's problem is that they do not have the $ to develop x86 designs in the timeframe that Intel has to date. The K6 is the closest that AMD has been, but Q4 I expect socket 7 to come under significant pressure from Slot 1. Therefore your comment of gatescode is true - Intel depends on MSFT and at the moment MSFT depends on Intel (it would be very difficult for MSFT to ever change from x86 to a completely new processor - look at the difficulty NT has had under competing architectures (only DEC Alpha remains)).
Rambus will make money in the short term - for at least one generation of PCI chip set - this represents anywhere from 6-12 months of time. The Rambus RAM will be used on a 100 MHz memory bus, which should provide some improvement over current 66-75MHz designs. Although DRAM manufacturers have licensed the technology, they do not enjoy paying royalties in what they view as a commodity business. Therefore there will be a push to get competing DRAM architectures for future x86 processor designs. In the end, I don't think that systems that use RAMBUS vs some other new DRAM design will have much in the way of cost differences when production volumes are achieved (initially they will be at a premium - so don't look to be able to buy before late 98, early 99 I would guess).
IBMs attempt to pass their PS/2 and MCA was unsuccessful because their design was a closed one, and licensing was limited. I remember them stating that in order to get a license one would have to pay 1% (I believe that was the #) of all revenues from PCs ever sold using the ISA architecture - retroactive to 1981. The industry balked at this, brought out EISA for servers (which had moderate success and has only dropped out of use in servers during the last year), and stayed with the ISA design for the desktop. When Intel released the PCI standard as an open standard, it not only wiped out the VL-Bus standard over a few months, but it also wiped out EISA, NuBus and in the embedded market is starting to gain acceptance with uPCI. To date I would say Intel has been exceedingly successful in getting its way, and as long as it continues with the development of open standards it should continue to maintain its market share.
Slot 1 has some licensing issues associated with it, and hopefully the will in time allow competition from the likes of AMD on this front as well. |
