How Does This Affect DDR and SDRAM?
One of the concerns expressed has been that royalty payments of 3% to 5% to DDR would give DRDRAM much more of a competitive situation, and would likely spell the doom of DDR. The fact is that normal market forces have caused SDRAM prices to swing from as low as $4 per chip to as much as $20 per chip in the past year alone. Even by eliminating the unusual circumstances that occurred in the Fall of 1999, prices have gone as high as $10, and may do so again in the near future. Based upon this, even a 5% royalty would have very little effect on prices in the long term, and chances are that the larger manufacturers will be able to negotiate favorable terms.
In addition, the DRAM memory industry has been in a consolidation mode during the past few years, with mergers or buyouts such as LGS and Hyundai, Micron and TI, and most recently NEC and Hitachi. This trend is likely to continue as large conglomerates let go of their unprofitable businesses to focus on more lucrative markets (such as Flash memories). This situation will likely result in more controlled production and more stable (and higher) prices, even without additional royalty payments.
For the near future, Micron is committed to SDRAM and to DDR as the follow on to SDRAM. They like the commodity memory market, and have done exceptionally well with it in the past few years, announcing record profits the last few quarters. Converting to DRDRAM production is expensive, and is still very problematic as evidenced by the yield issues reported by such companies as Samsung and NEC. Regardless of whether there are royalty payments to be paid, it would appear that Micron will be pushing DDR and SDRAM as long as the market is willing to buy it.
What this means is that regardless of whether Rambus ends up owning the technology or not, it will likely have very little effect on which DRAM becomes the dominant player on the desktop. There are still DRDRAM proponents, and there are DDR proponents – both of which have their own agendas and power base, but it is the consumer that actually decides. Ultimately, just as with every other technology that has appeared, it will be price/performance that determines which becomes the standard over the next few years.
Some believe Rambus’ long-term objective is to block or hinder the newly formed DRAM alliance from developing a new packet-data DRAM for the PC market, buying time for acceptance of their own designs. It turns out that packet-data technology pre-dates Rambus by several years, and is in fact based upon an IEEE standard developed in 1986, which was also used in the now-defunct SLDRAM design.
Therefore, unless Rambus is able to make SDRAM and DDR SDRAM economically unviable due to huge royalties, development and marketing will continue on just as it has. Most sources indicate that DDR based systems will be made available later this year, regardless of the outcome of Rambus licensing issues, and it will be at prices similar to SDRAM. In fact, sources within Micron have indicated that they believe price parity with SDRAM is essential to the acceptance of DDR.
What causes the most concern is the fact that Rambus is completely unwilling to engage in the standard practice of cross-licensing, because they have no need to do so. Generally, when a company licenses their IP to another, they will give favorable terms if that company also owns useful patents. This helps to keep royalty costs down. Since Rambus has no products, they need no patents from others, putting them into a position of strength when negotiating licenses.
Discuss (15 comments)