Driving the RAM-bus
The Bull Argument

By Tom Jacobs (TMF Tom9)

Rambus, Inc. (Nasdaq: RMBS), the Mountain View, Calif.-based memory chip technology designer, is like China: Few outsiders understand it, but everyone has an opinion. Some chide that Rambus has an "unproven IP [intellectual property] model," while others sneer that its technology is "no good, doesn't work." While Rambus endures Everests of FUD -- fear, uncertainty, and doubt -- and remains the subject of passionate debate, its stock has exploded (see this chart). Is there any upside left? Yes, a lot, but not without risk. Rambus is emphatically not your grandmother's Ma Bell. Can you handle a 20% move in a day or 600% in a month? You have been warned. Here's why I'm a Rambus Bull even at today's prices.

Rambus 101
First, Rambus is a deadly serious technology design company. Its mission states a staffing goal of no fewer than 80% engineers. Yet the average investor can understand enough to decide whether to become an owner. Anyone who uses a computer can get it.

Think of a computer as a having separate heads that need to talk to each other. The first head runs the show: It controls the computer's operating system and performs the calculations. Call it the microprocessor or the CPU (central processing unit). The CPU must get instructions and data from the second head, the memory, called DRAM (dynamic random access memory).

Since 1980, the CPU head has evolved faster than the DRAM head, speeding up from 5 MHz (million cycles per second) to over 700 MHz (and beyond), while the second head, the DRAM, has advanced to only 100 MHz. One advance in DRAM -- SDRAM, or synchronous DRAM -- has reached 133 MHz, and most of us have computers these days with SDRAM. Founders of Rambus caught on to this evolutionary discrepancy and developed technology to address it. That technology, RDRAM (for Rambus dynamic RAM), speeds things up so that the CPU can make "multiple concurrent requests" to the memory, increasing efficiency. Think increased bandwidth, reduced bottleneck. This is a good thing.

That's the basic idea behind Rambus, but if you want more details, the company's 10-Q and 10-K and website provide enormous amounts of layperson-safe information. I've profited enormously from posts on the Rambus discussion board, such as this recent technology and patent post by Fool info-god jasonxsmith and this company and market discussion by Fool aschwebel.

How Rambus makes money
Rambus doesn't actually make RDRAM, though its engineers developed and patented the technology and are busily advancing it -- even for use in switches and routers for optical networking. Rambus's patents are key: They are legal 20-year monopolies on the invention backed by the U.S. Patent and Trademark Office. If anyone wants to use your "art" -- as they say in patent-world parlance -- they pay you. If they want to use it, they can license it from you and pay you a royalty. For example, this is what happens with music; radio stations pay ASCAP for a license to broadcast music, and ASCAP distributes the royalties to musicians (hence the battle over Napster and sidestepping royalties to artists). Rambus filed for its first patents in 1990 and today has a large and growing patent portfolio to protect its design inventions. It's earned the label of "intellectual property" (IP) company.

Rambus licenses its technology to memory makers. It even provides engineers to help a licensee implement it. The memory makers then sell to original equipment manufacturers. The company then earns a royalty payment generally based on unit sales of the product containing Rambus's stuff. You can check the earlier links for more information, but we're generally talking 2% to 5% per chip, in a DRAM market estimated to grow from $20-something billion today to between $60 billion and $100 billion in the next three to five years. Not peanuts. Rambus's success in this market depends on market acceptance of RDRAM versus SDRAM and any other DRAM competition.

Semiconductor Rule Maker Intel (Nasdaq: INTC) has decreed that the market will accept Rambus. Intel signed a 1996 agreement with Rambus that commits the giant to use RDRAM in controller chips and to meet certain production and sales targets. The deal includes such carrots for Intel as 4% ownership of Rambus and a seat on the board. Since going public in 1997, Rambus watched its stock trade sideways -- especially during last year's delay of the first Intel product using RDRAM. When this glitch was resolved and Intel's support for Rambus reconfirmed, memory makers made RDRAM available, and Sony's (NYSE: SNE) PlayStation2 -- with Rambus inside -- sold gazillions in Japan earlier this year. (Watch for its U.S. invasion this fall.) When this happened, some FUD evaporated, shorts covered, and the stock zoomed from a split-adjusted $18.44 to $111.27 -- multiplying just under six times -- in 28 trading days! Takes your breath away, doesn't it?

There IS competing speed-up technology competition for RDRAM, with the acronym DDR SDRAM (DDR = double data rate). Depending on whom you believe, DDR is either just around the corner, delayed, or never appearing (for a DDR primer, see this post). Rambus quite naturally would like RDRAM to precede (it has) and defeat (it will) DDR to clear the market for its own future advances. For months, RDRAM armies and DDR armies -- Rambus investors and Advanced Micro Devices (NYSE: AMD) investors -- have lobbed grenades at each other over which is better, cheaper, and so on. Sorry to tell you, but it doesn't matter.

Savvy technology investors know that when Intel, the gorilla in the semiconductor space, decrees that All Roads Lead to Rambus, it doesn't matter whether DDR is better. It didn't matter whether Betamax was better than VHS, Apple's (Nasdaq: AAPL) operating system neater than Microsoft (Nasdaq: MSFT) Windows, Edsel finnier than -- well, OK. It matters only whether you win. And Intel has decreed that Rambus will win.

All of this is familiar to followers of technology investing based on Moore, Johnson, and Kippola's The Gorilla Game, in which, very loosely, a discontinuous technology innovation imposes pain on the status quo, gains acceptance and experiences hypergrowth. When I asked the Rambus board for some help with the bull argument, Fool nilocp made the Gorilla Game argument in this post. It beats anything I could ever say, so go there, and if not sated, then to the Fool's Gorilla Game discussion board.

But hold on to your screen, because a new development makes even hypothetical discussions of RDRAM technology's superiority, acceptance, price, or yields a waste of time. This change means not only that Rambus is a good investment at today's prices, but that the company could be seriously undervalued for the long run.

How Rambus will make a LOT of money
Before, I wrote: "Rambus's success depends on market acceptance of RDRAM versus SDRAM and any other DRAM competition."

But this is old news. Now, read this: Rambus asserts that currently dominant SDRAM and the promised DDR SDRAM use technology that depends on Rambus's patents, and any failure to pay Rambus royalties on SDRAM and DDR is illegal patent infringement. Read it again.

This is a big, fat, hairy, monstrous deal: Rambus is negotiating with DRAM makers (and sometimes suing them) to receive a royalty on every SDRAM, DDR SDRAM, and RDRAM chip. It is claiming that its patents cover 99% of the DRAM produced today and likely to be produced in the next few years. If Rambus succeeds, then RDRAM's technical superiority, price disadvantage, or fashion awareness matter not because Rambus gets paid for practically everything DRAM. Do the numbers: Rambus is reported to charge a 1% royalty for SDRAM and 3% for DDR. Estimate 2% of a $60 billion to $100 billion DRAM annual market in three to five years, and you have $1.2 to $2.0 billion for Rambus. This is why debt-free, leanly staffed Rambus could be called undervalued at its current $7 billion market cap, without accounting for any Rambus advances in its technology for the computer market or any royalties for its technology used in networking products.

This is no pipe dream. The Rambus-rules-all-DRAM world is forming. Toshiba, Hitachi, and Oki -- respectively the sixth, ninth and tenth largest DRAM makers -- have agreed to pay royalties to Rambus for their RDRAM, SDRAM, and DDR chips. The June Toshiba and Hitachi agreements catapulted the stock again, from $37.66 to $117.36 in 21 trading days. Imagine the effect of a few more SDRAM/DDR agreements, say, with numbers one to four DRAM makers Samsung, Hyundai, Micron, or NEC. I believe that this is a matter of time, and when it happens, momentum-based responses to these news events will present special challenges for the long-term investor. In the meantime, sing along with Fool goaltender95 here.

To be fair, it is possible to imagine DRAM worlds that Rambus does not rule. Rambus had to sue Hitachi to get a deal, and it's recently sued number five DRAM maker Infineon. This and at least six other possible lawsuits take time, are expensive, and create more risk. Rambus must, like Qualcomm (Nasdaq: QCOM), defend its patents aggressively and risk the whims of judges. Long-term investors must weigh the probabilities. Fool 5SigmaEvent has done so and reached this intriguing conclusion in a recent post.

Not only might lawsuits cloud the company's prospects, but Intel could change its mind about RDRAM, determine that a 4% share of Rambus and a board seat are peanuts for the chip behemoth. As Fool jasonxsmith points out, it might also matter if RDRAM doesn't work or is too expensive and if DDR is not a permanent solution to the two-headed evolutionary problem. Someone would need to come up with a solution, and Rambus could suffer. Pigs could fly too.

Cognizant of the risks and understanding the range of rewards, investing only money I can afford to lose and don't need in the next three to five years, and not buying a single solitary share on margin, I'm a Rambus bull.

The Bear Argument »

 This Week's Duel

  • Introduction
  • The Bull Argument
  • The Bear Argument
  • The Bull Rebuttal
  • The Bear Rebuttal
  • Vote Results
  • Flashback: Berkshire Hathaway

     Related Links

  • Rambus Discussion Board
  • Rambus Snapshot