Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: HarrisonS who wrote (5749)	7/17/1998 9:13:00 AM
From: REH	Respond to of 93625

A pretty good description of Rambus from a December article in Forbes magazine - old but interesting! Rambus: Moore's little helper By Rita Koselka YOU KNOW MOORE'S LAW-the assertion that the power of processor chips doubles every 18 months. The law also applies to memory chips. There's a weak link here, though. It's not so easy to get the two kinds of chips to double the speed at which they talk to each other. In to solve this problem is Rambus, a seven-year-old company that went public last spring to wild applause. Its stock has quintupled to $56, giving the little company-$26 million in revenues and $2 million in net income in its fiscal year ended Sept. 30-a market value of $1.3 billion. That's a value of $9 million per employee, even more than Microsoft's. Just what does Rambus have that makes it so valuable? One very clever idea about routing signals between memory chips and logic chips like memory controllers. Such manufacturers of DRAM chips as NEC, Toshiba and Samsung are paying Rambus royalties of some 1.5% to adopt the idea. On the logic chip side, Intel, Texas Instruments and Cirrus Logic also are paying a royalty. Intel says PC manufacturers will start shipping machines using Intel chips and Rambus technology beginning in 1999. If all goes according to plan, someday Rambus will be collecting royalties on maybe half the world's memory chip production. That royalty stream could be worth several hundred Faster recall million dollars a year. The prospect makes the company's valuation seem almost rational. We say "almost" because in the Santa Clara Valley anything is possible. Rambus was born during a 1989 dinner-table conversation between Mark Horowitz, now 40, and Michael Farmwald, 43. Farmwald, an entrepreneur who had already founded and sold a small chip company, is an idea guy, constantly bubbling with new opportunities. Horowitz, a professor of electrical engineering at Stanford, is the type of guy that a friend like Farmwald goes to for a reality check. Farmwald's idea had to do with signal pathways on a personal computer motherboard. Chip speed was bursting ahead, he said, while the links between chips were not keeping pace. Why? Very simply, there are three messages that need to travel between a memory chip and the logic chip a few inches away: the address where the impulse is to go, the data bits themselves and the instruction that says what should be done with that data-store it, replace it, retrieve it. These three messages traditionally run on different paths at different speeds, to be recombined on the memory chip. This process paces the transmission at the slowest or longest route among the messages and requires careful coordination at the chip to put the right threesome together. Farmwald and Horowitz came up with a different method: Package the three messages for transmission down a single electronic line. This shift in methodology would require some significant design changes in both the memory chips and logic chips. But if it could be pulled off it would speed up computers immensely. Over the next six months the two raised some $2 million for 50% of the company from venture capitalists. Horowitz took a leave from Stanford and hunkered down with two senior engineers and one of his former graduate students, all lured from Xerox Palo Alto Research Center. The easy part was designing changes in chip layouts and getting some patents. The hard part was persuading chipmakers to try the new system out. In 1991 they hired Geoff Tate, a veteran chip man at Advanced Micro Devices, as their chief executive. Rambus' big break came in 1995, in the videogame market. Nintendo was desperate to leapfrog the popular PlayStation made by rival Sony. A new memory technology was just the trick. The Nintendo64 game player uses Rambus designs with custom-made logic chips from NEC. The Nintendo game player costs only $150, but it processes graphic data at something pretty close to supercomputer speeds, ferrying the data between memory controller and memory chips at a speed of 500 million bytes per second. Nintendo's product is smaller and faster than Sony's PlayStation and was the hit of the 1996 Christmas season. That score gave Rambus some credibility with chip companies in the far larger market for multimedia and graphic boards in general purpose computers. Gateway 2000 started using Rambus technology in its Destination PCs in June. Micron Electronics started shipping it in its multimedia PCs in June, too. And Dell has been shipping Rambus technology in its workstations since July. In October Rambus announced its latest-generation technology, which will enable memory chips to receive and send data at up to 1.6 billion bytes per second. Could something go wrong? Sure. Chip companies could end-run the Rambus system. The big thing holding them back is the fairly low royalty rate. Says Tate: "We knew we had to make it cheap enough that chip makers would rather deal with us than start afresh." Another uncertainty: What happens in a decade or so, when microprocessor, memory controller and memory are all consolidated on one tiny chip containing billions of transistors? But then there's no way to anticipate technology ten years out. So you might well wonder at that extravagant market valuation-but then this is Silicon Valley, where almost anything can happen.