Rambus has transformed DRAM-making from a commodity business to an IP-heavy business.

Kausik Rajgopal

IF you follow the markets and have not heard of Rambus recently, you have not been paying attention. And if you have not, you should. The Rambus business model, its patent fights, and its victories and setbacks are likely to set the tone for IP-heavy bus iness models in the new economy.

The little stock roared in February-March this year, at one point gaining 400 per cent in value over a stretch of four weeks. After coming back to earth in the Nasdaq ``correction,'' Rambus remains one of the most hotly debated stocks.

What Rambus does is license its DRAM (Dynamic Random Access Memory) technology to multiple chipmakers. In order for the microprocessor, which is the heart of any computer, to function, it must dynamically and randomly access the memory of the computer. T he DRAM contains both instructions for solving problems and the related data, which it then supplies to the microprocessor.

What Rambus technology does is speed up the rate at which the microprocessor can access the DRAM. This is especially important because microprocessor technology (driven by Intel) has increased exponentially since the early 1980s, while DRAM technology ad vances have been comparatively slower.

Before we get to Rambus, it is important to understand why that was the case. Part of it is purely technological; the price to performance ratio for the microprocessor simply had more room for increasing speeds.

But part of it is also because the DRAM business was a commodity business in the mid-1980s, when Intel got out of it. At that time, Japanese manufacturers (NEC, Oki, etc) were squeezing Intel's market share and margins. Intel, which knew much more about design than about manufacturing, found that it was playing a losing game. It did not matter that Intel started out as a DRAM company, making the 1103 in 1970-71. By the mid-1980s, DRAM was a commodity business, not an intellectual property (IP) heavy bus iness.

But Rambus changed all that. With a mission of maintaining at least an 80 per cent engineering headcount, the company itself is an interesting story: It was founded by two Stanford electrical engineers and was around for several years before its IPO in 1 997. The key event that launched it: Intel signed a 1996 agreement with Rambus that commits the giant to use RDRAM (Rambus DRAM) in controller chips and to meet certain production and sales targets. Intel dominates the business, and an agreement with Int el dictates a ``standard'' for the industry.

Rambus merely developed and patented the technology; it does not deal with the hassle of manufacturing. It has patents, legal 20-year monopolies backed by the US Government that allow it to gain royalties on every DRAM. There is competition for DRAM, mos t notably from SDRAM (Synchronous Dynamic Random Access Memory) and DDR SDRAM (DDR is double data rate).

But here come the Rambus patent lawyers, and let's listen carefully to what they are now saying. According to Rambus, SDRAM and DDR SDRAM also depend on patented Rambus technologies, and therefore Rambus should get royalties for every SDRAM and DDR SDRAM as well!

The math is simple. Royalty rates run from 1 per cent for SDRAM to 3 per cent for DDR. There is some confusion about overall market size. Estimates range from a bottom of $70 billion to a high of $120 billion. Assuming an average 2 per cent royalty rate, that gives Rambus royalty revenue of $1.4 to $2.4 billion. And remember, Rambus sits in the middle of a low-cost intellectual property web, and reaps the benefits of defendable patents.

There is obviously uncertainty around whether the patents are defendable. But the track record is compelling: This year alone, Toshiba, Hitachi, and Oki have agreed to give royalties to Rambus for RDRAM, SDRAM and DDR SDRAM. They are, respectively, the s ixth, ninth and tenth largest DRAM makers in the world. The others are not caving in so easily. Micron, another large DRAM maker, recently sued Rambus, claiming that the Rambus patents did not cover their products.

Without diving into the merits of the technical case (which even qualified patent lawyers can disagree about), here is why I am bullish about Rambus: Toshiba, Hitachi and Oki are not small companies. They have a lot of money, and they can afford to hire some smart lawyers. Assuming that these highly-paid lawyers did their job in trying to determine if their companies could avoid paying Rambus, we must be forced to conclude that either Rambus had a watertight case, or that the legal costs were so high th at paying the royalties was preferable. Either way, the Micron suit seems like a last-ditch effort to bring Rambus to the negotiating table than a serious threat to avoid paying royalties.

IP-heavy business models such as Rambus are likely to be the harbinger of the ``new economy'' company much more than transitory dotcoms.

(The author is a consultant to start-ups on business plan writing, partnerships and marketing strategy. He can be contacted at ulticat@stanfordalumini.org)

(Please e-mail us at bleditor@thehindu.co.in if you have queries on computer usage or if you find an interesting way of using a computer.)