Teflon, et al, about Rambus. I'm going to play the devil's advocate for once on a semiconductor/PC or server related issue. Every day or so lately, Intel is getting downgraded or equivalent by an analyst. Main problem they see is lack of normal, historic interest in the latest and greatest new processor chip, the PIII. The lower price ones, Celeron, apparently are just fine for the vast majority of users, Intel's ASPs will therefore fall, and their stock will not be worth as much. So, what's the furor over Rambus? It's claim to fame is that it's supposed to make the fastest (PIII) PCs even faster, as the older, slower DRAM memory available today has become a bottleneck, not allowing the 550 MHz CPU to complete as many million instructions as it should. But, if those 550s aren't selling anyway, who needs the faster memory made by Rambus?
Not to worry, Rambus fans might say, Intel has announced design support for the Rambus interface for their low cost PC-on-a-chip (PCOAC), due sometime next year. This one should sell like hotcakes, because it really does have most of the goodies needed to make up a PC, all on one chip. It'll make for cheap, simple PCs like never before. But, because it's intended for PCs with very low selling prices, who needs the high priced Rambus DRAM in it, to pull the PC's price back up? So, Rambus would say, the DRAM vendors, mostly in Asia that actually make the DRAMs, cost reduce the hell out of it and start a price war, bringing the Rambus DRAM price down. Then, it'll fit in better in the low cost PC. But, the DRAM vendors still have to pay a royalty to Rambus for the privilege of building and selling the chips, and they're really going to love that, right?
Another announced usage for Rambus, in Nintendos and Sony Playstations, has to also be a price war situation. I don't think price wars are pretty for anyone.
Bottom line is I think Rambus may be two or three years behind its time. Back then, every Megahertz out of Intel was accepted eagerly and gleefully. Now, bandwidth coming into the PC is what everybody knows we need most. As far as Intel goes, they'll do all right. As Drew Peck, usually an Intel bear, said on CNBC yesterday, they need to watch cost and diversify. He said they were, in his mind, doing these by appointing a CEO that came out of manufacturing (the cost part), and he mentioned their purchase of Level 1, a network chip maker. I can add Intel's announcements of going into the Internet server farm business, and several other network related purchases, as major efforts to diversify. Peck said Intel stock should start to be more attractive as in investment later this year, because of these steps. Rambus, OTOH, is definitely a one-trick pony. I'm not sure if that trick sells tickets to the circus.
Last point, Intel is firmly behind Rambus, every step of the way, which makes for a big quandary to me. Do I believe Intel, whom I believe in as much as I do in any company I know, on the Rambus thing? Or, do I believe all of the previous premises? I haven't bought any Rambus stock, FWIW. Because it is such a quandary, I probably won't buy Rambus, while, at the same time, I won't bet against them because of the Intel backing. I guess that's a neutral rating.
Tony |
