Rambus Has Intel in Its Corner, but Still Can Take a Beating
The young memory-chip maker's stock bounces with its giant backer's announcements
September 7, 1999
With some stocks, just as in sports, being a fan can sometimes be more fun than being an owner. Few stocks are more exciting to watch than Rambus (RMBS), but owning it might cause you some sleepless nights. Just look at what happened the week before Labor Day. After the company's major backer, Intel (INTC), reiterated a previous announcement, the stock dropped 11% -- even though there was no change in the news.
The stock can be so bouncy because Rambus is one of those companies that specializes in half-court shots. It has proprietary design and production technology that promise to revolutionize the DRAM memory-chip industry. DRAMs -- for dynamic random-access memory -- are the heart of most PCs' memory system. If these chips are manufactured using Rambus' production techniques, they'll be able to process data more than twice as fast as current DRAM chips can.
TOO POKEY. That's important if DRAMs are to keep up with a recent burst of speed in PC central processing units. Only a few years ago, 133 megahertz was the bees' knees. But Intel just announced that it'll soon launch a Pentium III chip that will zoom along at 700 megahertz. While memory-chip speed hasn't kept up, that didn't matter as long as CPUs stayed below the 500-megahertz level. Now that CPUs are so fast, a PC's performance can noticeably deteriorate when delivering features such as 3D graphics and video because of slow memory chips.
Enter Rambus. Its technology allows memory-chip makers such as Samsung and Micron Technology (MU) to build DRAMs that can handle as much as 1.6 gigabytes of data per second -- more than double the bandwidth of today's memory chips. The new chips, called RDRAMs (for Rambus DRAM) will let next-generation PCs process the massive data streams that will become commonplace as visitors start connecting to the Internet via high-bandwidth devices such as cable or DSL modems. Although Rambus, which has been a public company for only the past year, has no factories, it earns a tidy royalty fee for each RDRAM chip that's produced by third-party manufacturers.
Since it owns 100% of its market for now, Rambus' biggest obstacle is time. It takes at least a year to build a new chip fabrication plant, and memory-chip makers must be absolutely certain that demand will be there before they spend hundreds of millions of dollars to revamp their factories to produce RDRAMs. That's why Intel is so critical to Rambus' success.
"A BINARY COMPANY." As long as Intel is behind the technology, which so far it claims to be, then memory-chip makers will feel confident about the technology and will continue working on the plant renovations they've already started. If all goes according to plan, the new RDRAMs should appear late this year in very high-end machines. By the end of next year, RDRAM assembly lines should be rolling at full capacity, just in time for Intel's new Merced processor, which is optimized for use with RDRAMs. If that happens, then Rambus will have won the game. "This is such a binary company," says S.G. Cowen analyst Drew Peck. "Either it's going to be fantastically successful, or it's going to be a footnote. We won't know until next year."
In the meantime, Rambus' stock will surely endure many ups and downs based on rumors about the relationship between Intel and Rambus. Hence the market's negative reaction to the announcement by Intel executives at an industry conference on Sept. 1 that the chip giant will support a different technology for some of its late '99 models. This technology, called PC 133, also enhances DRAM speed, although memory chips using it run at only about half the speed of a RDRAM chip.
Rambus' stock dived 11% early in the trading session on Sept. 2, before recovering a little to 92 3/8 as the market surged ahead of the Labor Day weekend on Friday, Sept. 3. Oddly, say analysts who follow the stock, the same announcement had been made during a July Intel conference call with no effect on Rambus' shares. The most recent announcement "was a non-event," says Gregory Mischou, who covers the stock for Warburg Dillon Read. "We expected the PC 133 news. But every Intel executive has been loud and clear that they fully are committed to Rambus as the next-generation technology. [PC 133] is just a short-term solution to a short-term problem."
FULL SUPPORT. So why the drop in stock price? It had more to do with the fact that there's a huge short position in the stock, and that the shorts were able to spin the news as significant. These short-sellers, who borrow stock from long-term holders and return them later on the bet that the stock price has gone down in the meantime, profit if Rambus' stock drops. "The hedge funds love shorting this stock," says Cowen's Peck. "There is no news to speak of for a long time, so they are able to manufacture news. The market is looking at the wrong reasons to bid this stock down. All that matters is Intel's commitment to the Rambus technology, and that hasn't changed at all." Indeed, each Intel executive who spoke at the developers conference mentioned the company's support for Rambus.
This isn't to say that every individual investor should start buying Rambus because it's so misunderstood. Few stocks could be more speculative than this one. After all, Rambus had only $37 million in revenue in 1998 but has a market capitalization of nearly $2 billion, making it wildly expensive. Even though the company is slightly profitable today, the real test will be 6 to 12 months from now, when either royalties will be flowing in or the company will be in trouble. Which is why it's easy to be a Rambus fan, while it takes guts to be an owner.
Jaffe writes about the markets for Business Week Online |
