In the past months, the x86 microprocessor landscape has undergone a dramatic shift. In the space of less than a month, chipset supplier Via Technologies Inc. snapped up both Cyrix Corp. and Centaur Technologies, two chip design houses that had struggled in their attempts to address the low end of the PC microprocessor market.
With the fate of both Cyrix and Centaur left somewhat uncertain, the spotlight has shifted to the most recent startup, Rise Technology Co., a company founded to specifically address the low end of the PC chip market. Is Rise the next acquisition target, or will the company indeed rise to seize upon the market uncertainty? EBN senior editor Mark Hachman asked Rise chairman and chief executive David T. Lin for his thoughts.
Q: Can you comment on the x86 microprocessor landscape, and how that has changed since last year at this time?
A: Throughout the last year there were a number of x86 suppliers, all of whom were trying to define how to compete in the low-cost segment of the market. Rise, AMD, National Semiconductor, and IDT were all taking different approaches to the low-cost desktop PC market, while Intel was bringing in the Celeron to compete in this area.
In recent weeks, AMD has announced their thrust to move aggressively into the performance PC segment of the market, with a product [Athlon] designed to give them the higher ASPs they need to survive. Both National Semiconductor and IDT have been acquired by Via .
This leaves Rise as the sole company focused on the low- power and low-cost market segments.
We believe we are uniquely positioned to address these markets due to two factors. The first issue is that in order to be financially viable in this segment, it is necessary to have a very low fixed-cost structure. Rise achieves this through being a fabless semiconductor supplier. Thus, we are able to avoid the billions of dollars in investment and subsequent amortization costs that have plagued our competitor's financial statements. The second issue is that our product is inherently a low-cost design. This is due to the extremely low power consumption of the Rise mP6 architecture, which allows us to reduce our packaging costs by up to 50% compared with our competitors.
Q: After Centaur and Cyrix are acquired by Via Technologies, Rise will remain as the only merchant x86 microprocessor supplier in the low end of the PC market. How do the two acquisitions affect Rise's future?
A: We view this merger as having a positive effect on Rise. It is not clear how the merger of Cyrix and Centaur will be achieved under Via's management, but in any case this task will be challenging. As analyst] Keith Diefendorff [of MicroDesign Resources] said in a recent article you wrote: "If you start to stir the pot together, you're going to come out with something that smells really bad." The current products from Rise are superior in both performance and power consumption to anything being shipped by Cyrix or IDT, and it is a very daunting and challenging task to move forward with their roadmaps during this transition period .
With regard to notebooks and mini-notebooks, Rise's products are unique in the market. The mP6 and mP6 II processors are the only processors that can meet the industry's Mobile 2000 Power guidelines above 300 MHz. Even Intel's mobile processors consume more power than recommended for notebooks and are well above the guidelines for mini-notebooks. AMD's mobile processors, according to their own published specs, consume even higher power and dissipate more heat than the mobile processors from Intel, and neither Cyrix nor IDT have anything in the mobile power range.
Our initial tests, run in standard notebooks, confirm that the mP6 processor is able to deliver up to an hour longer battery life running [software-decoded] DVD. Tests were run on a Clevo notebook, comparing the mP6 366 with the [AMD] K6-2 350. To be specific, only the Rise mP6 and mP6 II, the industry's lowest power x86 cores, can meet the industry Mobile 2000 power standards for heat dissipation and battery life in either a notebook or mini-notebook. So we are confident that we can compete in this segment.
Q: Sources have said Rise has talked to several outside firms concerning its options: namely, an investment in Rise by a third party, or an outright sale to another company. Is this true? If so, can you comment on the nature of the discussions and with which companies the discussions have taken place? And why have these discussions been conducted?
A: Rise, as the only remaining x86 supplier, has been the subject of significant speculation in the press and analyst community since the acquisition of Cyrix and IDT by Via. As we have previously stated, Rise has had investments from third parties for some time, including two foundries, chipset suppliers, graphics companies, and other companies who have wished to remain anonymous. Many of these investments were made prior to Rise even having first silicon, and are certainly not new nor are they due to any recent market activity.
As a supplier of x86-compatible CPUs, Rise continues to have ongoing discussions with various industry participants, including chipset [suppliers], graphics [suppliers], and OEMs. These discussions involve a wide range of issues, including new products under development, and other relevant business discussions. However, we do not have any news to report regarding additional investments or other business arrangements at this time.
Q: If there in fact have been no substantive talks with other firms, how can Rise survive and flourish? I would point out that both IDT and National blamed their respective processor lines for their losses.
A: We can not comment on whether any talks have taken place except to say that there is nothing to report at this time. Regarding the Cyrix and Centaur losses, these losses were attributable to very different problems.
National Semiconductor's losses regarding Cyrix can be directly blamed on their foundry in South Portland. This foundry was never fully loaded, and Cyrix' products were bearing virtually full overhead for the massive depreciation this fab was incurring on the company. Not only were Cyrix' products being produced at an extremely high cost due to fab overhead, but they also were incurring high packaging costs due to the amount of heat dissipated -- the highest heat dissipation in the x86 industry. With high packaging costs and high factory amortization costs, National was the highest cost supplier in the industry with the lowest priced products. The poor economics of this business model are obvious.
Rise has eliminated both of these problems. We are a fabless company and thus do not incur the R&D and amortization costs associated with a fab and proprietary manufacturing process. At the same time, we have significantly reduced our packaging costs by using plastic BGAs rather than the Ceramic PGAs used by our competitors. This allows us to produce products at a low unit cost which is necessary to target a low-priced segment.
In addition, as in every major merger and acquisition, the Cyrix product direction to some degree had been impacted and delayed by National's acquisition. It caused customers to feel uncomfortable. These same transition difficulties are going to be difficult to overcome as Via tries to merge Cyrix into it's own corporate culture and simultaneously merge Cyrix and Centaur.
IDT's issues were very different. IDT started with a 486-like single-pipeline core that, due to its small die size, was supposed to be low-cost and low-power consumption. However, they were unable to produce a product that could run below 3.3V and shipped standard products at 3.52V. Since the competition was in the range of 2.2V (except Cyrix at 2.9V), IDT's supposed advantage in power consumption never materialized.
The open question for Centaur, under Via, is whether they will be able to successfully deliver the Winchip4, which attempts to correct the shortcomings of the previous Winchips in terms of multimedia performance and MHz. However, the Winchip4 is also a serious step away from Centaur's previous design mentality which was designing for small die size. Preliminary estimates are that the Winchip4 die size is larger than Intel's Coppermine, Rise's mP6 II and Tiger, AMD's K6-2, and the Cyrix Mojave processor, if it is still being brought to market. At the same time, power consumption estimates are in the range of 14 to 19 Watts, which is two to three times greater than the Rise mP6 II.
We believe that by continuing to focus on the low-cost and energy-efficient products optimized for consumer multimedia, that we will be able to achieve significant penetration into the low-cost portable and desktop markets, as well as emerging consumer markets that require the low-power consumption of the mP6 and mP6 II processors.
Q: But Rise's market share is much smaller than either Centaur or Cyrix. Why?
A: Rise has been in the market for less time than either Cyrix or Centaur so it should not be surprising that our market share has not caught up to theirs yet. However, there is another issue that we should address. Last quarter, National Semiconductor announced that they were writing off their inventory of Cyrix processors and would sell them at whatever they could get for them. Subsequently, over a million Cyrix processors were put onto the market at liquidation prices. During that same quarter, AMD announced that they had overbuilt inventories for the K6-2 by more than two million units, representing over one-third of their total production for the quarter.
Based on the thrust by AMD and Cyrix to clear inventories, pushing three million excess units into the market, Rise decided to scale back production of our 0.25-micron mP6 products and instead put our resources into development of our 0.18-micron products. This ability to scale production to the needs of the market illustrates the value of being a fabless company, as opposed to AMD and Cyrix, which are paying depreciation on their factories whether they are building product or not.
Rise is currently sampling 0.18-micron, 2-V mP6 products to a variety of customers and we expect to ramp production of these products later this year as previously announced. With power consumption below 4 watts typical, these products fit well below the power consumption limits for mini-notebooks as well as notebooks, and we are aggressively pursuing design wins for Super7 mobile solutions using the Acer Labs Aladdin V chipset and the Trident CyberBlade chipset. We also see strong opportunities and customer interest for low-cost desktop systems and information appliances using chipsets such as the [Silicon Integrated Systems] 540, which allows the entire motherboard to be implemented in just two chips producing the lowest cost DVD-capable system on the market.
Q: Is integrating a chipset or graphics core with a microprocessor the inevitable direction of the low-end market, as Via executives think?
A: Providing adequate functionality at the lowest cost possible is required to be successful in the low-end of the market, whether that is desktop systems or information appliances. However, blind integration is not necessarily the correct approach-it is very important to balance the desire for integration with the feature sets required by the market. With an integrated solution, if any of the features that are integrated are obsolete, it makes the entire solution obsolete. At the same time, if there are chipset features missing, this will cause a vendor the need to buy a separate chip, which eliminates the value of the integrated solution.
What is critical at this stage of the market is to partition the solution to maximize flexibility and features while minimizing cost and chip count. It is also important to look at the state of the process and manufacturing technology before deciding what to integrate-sometimes it is cheaper to build a two-chip solution than to build a one-chip solution, if that one-chip solution has a die size that is too large.
Rise is looking very closely at the merits and drawbacks of various integration approaches, and is in discussion with our partners on what may make the best sense in the future. The Rise cores themselves, due to the inherent low-power consumption of the architecture, make them a good candidate for integration if we decide to take this approach in the future.
Q: Do other opportunities exist outside of the PC market?
A: The low power consumption of the Rise processors, and our use of small BGA packages rather than larger and more expensive ceramic PGAs, are very attractive in certain segments of the information appliance and embedded markets. Rise currently has the only high performance x86 core that can run without a CPU fan. This makes it attractive for set top boxes and other consumer devices. The ability to run software DVD, soft audio, and soft modem solutions has also generated significant interest in customers looking to deliver low-cost multimedia internet capabilities to the consumer.
We have seen strong interest from companies in devices such as DVD game players and other consumer devices in addition to the PC desktop and notebook markets. With the Rise processor and an integrated chipset, these devices can be built with as few as two chips, delivering modern multimedia CPU performance at an attractive price.
Q: Does Rise's roadmap remain unchanged, as presented at WinHEC and the Microprocessor Forum? If it has been altered, how so?
A: Currently Rise is continuing with our plan, but at the same time we also realize the x86 market dynamics are always changing fast. We are always carefully evaluating our plan, market dynamics, and different business opportunity. Our product strategy is to build and improve our CPU core vertically and apply these CPU cores horizontally to different applications or markets. With this strategy, it allows us to dynamically respond to market needs, and we need to be able to do so as a smaller company. If there is enough change of plan in the future, we would let you know.
Q: Is Rise's foundry the UMC-ST Microelectronics joint venture, as is thought?
A: Unfortunately, we are unable to comment on our foundry partnership, except to reconfirm that we have an agreement with a foundry partner who has a patent cross-license with Intel, allowing Rise to ship processors without violating Intel's intellectual property. At this point, only AMD and Rise are in this position, as Cyrix and Centaur do not enjoy this type of arrangement. This foundry partner is able to provide access to leading edge 0.18-micron technology, where we are currently doing all of our manufacturing.