Monday, December 17, 2001
VIA just would not give into the P4 license. It is looking for a fair license feee because VIA think that Intel can also benefit from the major 3rd party DDR chipset.
Can P4X266 extend the life span of PC133?
In the past, Intel was number-one in global chipset market accounting for 70% of all chipset sales. Because the mistake it made in supporting Rambus, itís popular lineup was down to only one, the 810E chipset that supports the OEM manufacturers. At the same time, VIA was pushing PC133 and DDR structure as extension to SDRAM. Cost and selling price were all matched up with consumer expectations. That caused VIA to spread like a bush fire to quickly occupy 40% of the market.
Intel, after reviewing its mistake is motivated to regain its chipset market leadership quickly. This time, Intel came prepared. It has just won back its title on CPU clock speed through its new Pentium 4 microprocessors. It created the first opportunity for business to upgrade their computers. It further uses its patents and its 845 chipset to establish a stronger marketing position. VIA, being the best chipset house besides Intel, might have the opportunity to seize the ďBig BrotherĒ title in the next round.
From the chipset manufacturer point of view, although obtaining the P4 license is important. However, a high dollar license would also hinder the long-term competitiveness of the company. At this time that the PIII is winding down, AMD does not have sufficient production capacity, all chipset manufacturers saw that P4 with DDR is the market turning opportunity. Other P4 chipset manufacturers are all at the same start but see that the expensive licensing is a ticket to prosperity and is an opportunity to gain market share. While this might be an acceptable scenario to the second tier chipset companies, it is not acceptable to the first tier VIA. The high licensing fee would not only hinder its market position, it would also put it under Intelís control. This would later affect the company in the CPU market.
The other question is ďDoes VIA need P4 license?Ē According to VIA, all it has to do is to label the chips ďVIA-S3Ē signifying that this is a chip made out of the VIA/S3 joint venture. Since S3 does have cross-license protection from Intel, VIA would get an indirect license to produce. On the other hand, VIA also express that it wants to be an Intel P4 partner. Therefore, it is important that it gets licensing approval and in good relationship with Intel.
In anyway, Intel would reap the most benefit. Technology license would help P4 to reach mainstream at the earlier date. Intel would solidify its first position in the CPU market. While it is gaining profit from royalty, Intel can also make good profit on its own chipset. Intel should have learned a lesson from the last battle. It should taken plenty time in settling these license dispute. Settlement should be just a matter time.
As the fact, Intel lost its chipset market to VIA last year. It is still like a sprinter on its back. Intelís total domination of the entire market might never be back again. When compared the microprocessor market to the chipset market, microprocessor would be far more important. This year Intel would have plenty of capacity to support its CPU production. Together with the chipset support from ATI, Ali and SIS, it should be less dependent on VIA chipset. Although it cannot ignore the support from VIA chipsets, the urgency to license VIA is greatly reduced.
According to Intelís latest roadmap, Intel will keep its mainstream PC at the price about U$800. It has also adjusted its CPU prices accordingly. One of its strategies is to accelerate the replacement of P3. This included dropping the P4 price drastically to fit the $800 mainstream PC market. It has planned to totally replace the P3 with the new P4. It planned to have totally replaced the 1.2GHz P3 with the 1.4GHz P4 at the third quarter of 2001. It will totally stop P3 production on Q4 and push the 1.5GHz P4 on a broad basis. The P3 would become the engine for notebook computers. Price wise, typical 1.8GHz P4 after the price adjustment of 8/26 and 10/26 would be reduced 54% and 12%. This rate of price dropping has never happened before. In the example of the 1.4GHz P4, the 8/26 adjustment would put it under the price of the 1.2GHz P3 microprocessor at U$135. Intelís strategy is to attract upgrade by slashing prices. This leads the market to abandon the P3 and to embrace the new P4. At the same time it can increase its technological lead on AMD. It can then use its microprocessor influence to get back the lost chipset market.
When Intel first introduce Pentium 4 processor, it was designed mainly to use Rambus memory. It insisted that only Rambus could bring out the high performance of the P4. Unfortunately, this combination had over priced the market and did not replace SDRAM to become the mainstream. After a lot of storm and fuss, Rambus still could not move into the mainstream market. Intel then turned their thrust to the 845 chipset that provides SDRAM solution to the P4. It was expected to fulfill the PC requirement of the second half of this year. However, when compared the 1.6GB memory bandwidth to the 1GB bandwidth provided by PC133 SDRAM, the solution is not attractive. This has lessened the appetite of the end user. Under the market pressure, system integrators quickly turned to the 2.1GB DDR structure hoping to find the best price/performance compromise.
Using the 128MB as a base, we find that Rambus is U$15 higher than the others. On the 256MB, price differentiation is as much as U$42. Considering price difference between DDR and SDRAM, the range stays at U$1 to U$2 no matter it is 128MB or 256MB. Although Samsung and Rambus promised to shrink the price difference to U$20 by yearend. Micron also claimed that it would price DDR the same as SDRAM by then. Overall, we can see that Rambus might have price/value advantage over the SDRAM but definitely would not overcome DDR in the near future.
In summary, Rambus manufacturing equipment costs is 50% more, packaging cost is 30% higher, material cost is 20% higher. All these factors allow DDR to have short to mid-term advantage in the competition. Intelís plan is to sell 18 millions of P4 processors by yearend. Based on the high price of Rambus platform and the low performance of its 845 SDRAM platform, it would be very difficult. When look at Intelís drive for microprocessor generation change over, itís not difficult to predict the path they will take. However, Intel is bounded by contract with Rambus that they cannot sell their DDR chipset before Q1 next year. This has given an opportunity to the Taiwan chipset manufacturers to expand their market share.
Click Here for Part-6
By: Johnson Wang
Copyright © 2001 CST, Inc. All Rights Reserved