Semiconductors
Rambus Patent Suit Holds Huge Stakes for Company, Chip Industry
By Marcy Burstiner
Senior Writer
2/9/00 9:15 PM ET
URL: thestreet.com
SAN FRANCISCO -- It's warfare in the memory-chip industry, the relatively small group of companies that makes the semiconductors that store data in computers. Last month, industry upstart Rambus (RMBS:Nasdaq) sued giant Hitachi Electronics (HIT:NYSE ADR) in a Delaware court, claiming that Hitachi's memory chips, called Synchronous Dynamic Random Access Memory, or SDRAM, infringe on Rambus patents. But that's just the beginning of the story.
Because memory chips are commodity parts that differ little from one company to the next, all the industry players -- from Hyundai to Micron Technology (MU:NYSE) as well as NEC (NIPNY:Nasdaq), Samsung, Toshiba, Infineon Technologies (the renamed semiconductor unit of German conglomerate Siemens), Fujitsu and Mitsubishi -- could owe royalties if Rambus wins the case, making personal computers far more expensive. It is primarily because the industry loathes paying royalties that it has fought off pressure from chip giant Intel (INTC:Nasdaq) to adopt Rambus' designs. For Intel, Rambus chips would speed up memory, which would increase the performance of the Intel microprocessors that are the brains of most PCs.
The stakes are huge: Since memory-chip makers first started marketing SDRAM memory chips in 1996, computer makers have bought an estimated $33.6 billion worth of them, and expect to buy another $22 billion worth this year, according to the Cahners In-Stat Group. Moreover, stock watchers say an out-of-court settlement in Rambus' favor would send its shares soaring, largely because so many short-sellers (investors who bet a stock will fall) have built up large positions against Rambus.
At least two patent experts believe that Rambus' patents have a good chance of standing up in court. If that happens, or if there is a settlement in Rambus' favor, the company is looking at a huge new revenue stream that could eclipse even the most bullish financial forecasts for the company. Wall Street currently guesses that Rambus' profits will increase 125% in calendar 2000 to $19 million, or 81 cents per share.
Of course, Rambus may lose. Some observers think the suit is a desperate act by a onetime IPO highflier whose market potential is shrinking fast. Whatever happens, it could be a long and costly fight in an industry that has helped build the foundation for the PC revolution.
"I think it is an act of desperation," says longtime industry analyst Nathan Brookwood of semiconductor research firm Insight 64. "It is beginning to look like the window is closing as to Rambus' ability to make it into the high-end PC, desktop or server environment."
It wasn't always this way. Rambus went public in May 1997, and was one of the hottest IPOs in a year that included the initial offering of Amazon.com (AMZN:Nasdaq). Its popularity rested on its financial model. As a designer rather than producer of chips, it would produce high royalty revenue with low costs: Average royalties of 1.5% on a 50% share of a $20 billion memory market would snag Rambus $150 million in royalties a year, on total annual expenses of about $30 million.
But the memory industry balked at paying Rambus. And it took advantage of repeated delays by Intel (INTC:Nasdaq) in rolling out Rambus-supporting chipsets to develop royalty-free alternative technology called double data rate memory, which doubles the rate at which SDRAM transfers data. So confident have some investors been that Rambus would never make it to the mainstream computer market that they are now shorting almost half of Rambus' stock float, or the number of shares available for trading.
But at least two respected experts on chip patents who are not involved in the Rambus-Hitachi suit say they have examined Rambus' patents closely and believe the patents are solid. Richard Belgard, an independent consultant in Silicon Valley noted for his chip-architecture expertise, says he spent about three hours trying to dig up "prior art" -- evidence the technology predated Rambus -- but came up empty. "I couldn't find any semblance of possibilities of ideas that people had come out with," he says.
The suit revolves around four patents. Rambus filed one in 1990, about three years before SDRAM first hit the market, and three more in 1997. Rambus waited so long to sue because the U.S. Patent Office didn't issue them the patents until last year.
A second patent consultant, who asks that his name not be used for fear of being subpoenaed in the suit, says he too believes Rambus could defend its patents in court. "From what I've looked at, they look quite powerful, and [apply] well to today's SDRAM."
The memory industry will likely fight back hard and try to prove that its technology predates Rambus. In the process, Hitachi could end up invalidating Rambus' exclusive rights to its own designs. "I've got to believe that in the collective patent portfolios and expertise there is enough stuff to invalidate most if not all of the claims relevant to the suit," Brookwood says. "The fact that they have a patent is an advantage. We will now find out just how thorough and airtight it is."
If Rambus wins, it won't be the first time the courts force the memory industry to pay royalties. In 1986, Texas Instruments (TXN:NYSE) sued nine memory-chip makers and won. The companies paid TI more than $1 billion in royalties over the next six years.
For their part, Rambus executives say if they weren't certain their patents will hold up in court they would not have filed suit. Avo Kanadjian, vice president of worldwide marketing for Rambus, says his company would prefer to reach an amicable settlement with Hitachi and all other memory-chip makers. A Rambus spokesperson says the company is seeking only reasonable licensing fees and royalties and is not seeking to harm financially Hitachi or any memory company.
Patent experts say Rambus is taking a big risk because the search for evidence that SDRAM predates Rambus could produce evidence that invalidates Rambus' rights to its own designs. Even mentions of similar designs or applications found in magazines, textbooks or student papers could do the trick. "The stuff kids do in college are wonderful for that," says the patent expert who spoke on condition of anonymity. "A judge then says this was done before, and he punches the patent out."
Brookwood says all of the memory-chip makers will very likely pour resources into the search. "Rambus clearly is motivating all of the folks making conventional memory and forcing them into an alliance," he says. "I've got to believe that in their collective patent portfolios there is enough stuff to invalidate most, if not all of the claims relevant to the case."
Peter Clark, CEO of Hitachi Semiconductor America, declined to comment on the suit, saying the company will argue in the courts and not in the press. He did say that the suit was unexpected. "I was very surprised by it," he says. "Hitachi and Hitachi Semiconductor America are very confident in our position."
Privately, representatives from several memory companies say they will pore through every source they can find for evidence to invalidate Rambus patent claims. An extensive search, one representative says, will ultimately trace the technology to the grandfathers of hardware, Hewlett-Packard (HWP:NYSE) and IBM (IBM:NYSE).
But the fight may boil down to an expensive game of chicken. Rambus has about $6 million in cash and $113 million in total assets, compared with about $10 billion in cash for Hitachi, just one company in the potential anti-Rambus alliance. Still, the memory industry is just coming off a multiyear recession, as is the Japanese economy. For the fiscal year ended March 1999, Hitachi lost $3 billion, and its investors may be loathe to pay high legal costs for a suit they're not certain they can win.
One memory industry consultant says he's confident any initial court victory by Rambus would be overturned on appeal. In a 1991 suit claiming infringement of designs for memory modules, for example, a jury in Virginia ruled in favor of patent holder Wang Laboratories against NEC and Toshiba. But the U.S. Court of Appeals in Virginia overturned the ruling two years later.
One stock broker who has long advised clients to short Rambus stock says the suit will backfire on Rambus. "Barring an absolutely huge landmark settlement or victory in court, which is five years out, they are toast," he says, speaking on condition of anonymity. "It galvanizes the industry away from them."
Still, he acknowledges that none of the companies involved may want a long costly court battle and that any settlement for Rambus, no matter the size, could have a dramatic upside effect on the stock.
For now, investors betting on Rambus' success say they are staying put. "If Rambus technology becomes mainstream it becomes a huge win for them," says Michael Watner, a portfolio manager at New York money-management firm Douglas Lane & Associates, which holds Rambus in a high-risk fund. "We won't sell that stock until we become convinced it won't become mainstream."
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
¸ 2000 TheStreet.com, All Rights Reserved. |
