Via's ballooning problems

AS OF DECEMBER 24th, and as we reported at the time, Taiwanese semiconductor firm Via filed new allegations with Taiwan's Fair Trade Commission claiming Intel was using unfair competitive practices.

This, in and of itself, is nothing new -- Via first complained about Intel's alleged unfair practices as the result of an issue that ballooned into a problem at last year's Computex.

Whether the charges are ultimately upheld or not, Via is definitely turning up the heat on Intel -- which may be a sign of how much its lack of a Pentium 4 licence, and Intel's response to it, has hurt its business.

While the P4 market grew tremendously in 2002, Via's piece of the pie has fallen dramatically, with the last available figures suggesting the company holds less than 10% of the P4 chipset market.

This is quite a different picture than the one a defiant Via painted a year ago when the company took top honours as the largest third-party P4 chipset provider, outshipping rival SiS by almost three to one.

The tables have turned -- SIS is now the single largest third-party chipset provider, and holds almost 20% of the market. Although Via has continued to update and revise its P4X line of chipsets and has a competitive feature set with the P4X400, neither motherboard manufacturers nor end users have been quick to embrace the Via-built designs.

Even Via's launch of its own platform sales department seems to have done little to reverse this trend, with the company's gross revenues slumping by 95% through the second quarter of this year. While part of Via's current problems can be traced to the current IT recession, there's another side to the story worth digging into.

1999 and 2000 were banner years for Via, as the company found itself perfectly positioned to take advantage of a major stumble on Intel's part.

While the 440BX chipset was perfectly capable of running at a 133 MHz FSB (and could theoretically have been outfitted with a 1/4 divider and issued officially at that speed), Intel had chosen to focus its next-generation efforts on i820 and the RDRAM platform.

The high cost of RDRAM significantly slowed i820 adoption, however, and consumer confidence in the platform collapsed when Intel's MTH (Memory Translator Hub) designed to allow the use of low-cost SDRAM in i820 boards utterly failed when brought to market due to inherent technical flaws.

With the cost of RDRAM stuck in the stratosphere, a giant gap opened in Intel's product line, and Via lost no time in taking advantage of it. The Via Apollo Pro 133 chipset offered support for a 133 MHz FSB and memory bus and was designed around the far-cheaper PC133 SDRAM standard.

With computer sales surging and the original Athlon, and its KX133 chipset gaining popularity, Via made money and gained market share hand over fist.

Intel, moreover, had no choice but to watch the upstart manufacturer run roughshod all over its chipset business.

AMD's Athlon represented a tremendous threat in terms of performance and had a nasty habit of undercutting equivalent P3 systems in terms of price. Without Via's low-cost Apollo Pro chipsets, Intel would have had no platform on which to base a competitively-priced product, which would also have left the entire market completely open to its Sunnyvale rival.

As it was, the Athlon was gaining market share quickly -- leaving the door completely open would have been suicide.

When Intel was forced to delay its i815 chipset, Via dominated the chipset market even longer than anticipated, claiming almost 50% of all sales by the end of 2000. The company was riding high on its own success, and several starry-eyed analysts predicted the Taiwanese company might even have a shot at unseating Intel from its position as the largest worldwide chipset provider.

Santa Clara, however, had other plans. While it was forced to rely on a rival for support during a critical misstep, it laid plans to ensure such an event would not happen again. Intel announced early in the P4's development that third-party companies wishing to build chipsets for the Pentium 4 would be required to license the technology to do so.

Rather than licensing the technology, Via chose to take the position that prior agreements transferred by corporate purchases meant it had the right to produce a Pentium 4 chipset, licence-fee free.

Intel, of course, thought otherwise. When Via persisted with its plans to produce a P4 chipset without acquiring a license to do so, Intel publicly declared its intent to take Via to court over the matter, while simultaneously putting private pressure on motherboard manufacturers to avoid the P4X266 chipset altogether.

The combination of public and private pressure worked well, and effectively forced Via's P4X266 (and subsequent iterations) into the shadows. While some motherboard manufacturers have produced Via-based P4 solutions, such boards were never promoted highly, even when the P4X266 offered the only available DDR solution for the Pentium 4.

The effect on Via's chipset sales was catastrophic, and the company has fallen from holding almost 50% of the P3 chipset market in 2000 to less than 10% of the P4 chipset market in 2002.

While Via's AthlonXP chipsets have been far more successful than their P4 counterparts, AMD simply doesn't command the vast market share Intel does, which makes success in the AthlonXP market a thin blanket indeed with which to cushion such a massive loss.

While Via as a corporation is in no danger of being dried up and blown away by these events, the company could conceivably find its chipset presence in the US severely threatened if it cannot find a way to boost its own visibility. Nvidia's new Nforce2 chipset threatens on one front, a resurgent SIS on another, and increased licensing pressure from Intel on a third.

The facts thus far might seem to point rather damningly at Intel -- and some questionable business practices -- for having cost Via a giant part of its business. While these allegations may be true, there's another aspect to the situation that's generally not addressed.

Via has, rightly or wrongly, acquired something of a reputation for chipset problems with a number of end users. We've heard about sound problems, data corruption problems, lacklustre chipset releases, and a system seemingly designed to push users towards promising chipsets - followed by a new release almost immediately after.

First we saw the launch of KT333 scarcely two months after KT266A. KT333 boards had no sooner hit maturity than we were greeted by the arrival of the KT400 -- but without any 400MHz DDR to back it up.

The KT400 chipset has proven itself to be a fine chipset, but it scarcely overclocks beyond its 166MHz FSB, while the KT333 was known to push well beyond 190MHz. Now the ominous spectre of a "KT400A" looms on the horizon, leaving some to wonder why Via can't get it right the first time round.

While a KT400A chipset with increased performance and a new feature set would be welcomed by the community, it would also raise the question of why KT400 couldn't have been delayed and released as the KT400A.

The MSI KT3Ultra2 proved that a KT333 board could officially support a 333 MHz FSB, and it's not as if Via has had competition in the AMD chipset market, with nForce2 boards only now beginning to appear. The main "advantages" of our current KT400 are all-but useless -- AGP 8X and Serial ATA are merely buzzwords, offering little performance increase and are barely useful.

While Via may well have a solid case against Intel -- it's not hard to believe -- end users must not be forgotten. When the Athlon was released Santa Clara put similar pressure on motherboard manufacturers not to release AMD products, but skyrocketing AMD demand and P3 constraints gave the motherboard manufacturers ample reason to begin manufacturing and promoting AMD boards.

There's no sign that P4 end-users have demanded a Via solution from motherboard manufacturers, nor does a sudden swell of pro-Via board support seem likely any time soon.

The Taiwanese company can address the free trade issue in a court of law, but end-user perception can only be changed by hard work and consistent results. If the company wants to regain its lost market share and remain relevant in the US, it's going to have to find a way to address both of these issues -- and soon.

Intel's already demonstrated its willingness to play dirty where Via is concerned--and Nvidia isn't exactly known for kid gloves and soft treatment either. µ