There's been a lot of discussion in the market and coverage in the press about the current crisis in the DRAM market. Certain competitors of Hynix have been doing their best to paint a picture of Hynix as the villain who single-handedly has caused the collapse of DRAM prices.
One has to ask, however, is Hynix the real villain, or is it Hynix's competitors, who are aggressively building DRAM capacity while calling for Hynix to get out of the DRAM business?
DRAM has always been one of the most challenging markets in the semiconductor industry for a variety of reasons, including its cyclicality, the commodity nature of the market, and the intense competition among the various players. Today, the entire semiconductor industry is suffering the worst downturn in its history and the DRAM market is at one of its lowest points.
This has prompted certain DRAM companies to argue that if a particular DRAM company would just quietly fold up its business and go away, everything would be rosy for the survivors. In addition to it being highly self-serving, this argument is based on false premises and would probably, in the long run, end up hurting the OEMs purchasing DRAMs, and, subsequently, consumers. There are approaches, however, that DRAM companies can take to help reduce the extreme market swings.
First, let's look at the premise that there are just too many companies in the market. In 1989, there were 12 companies around the world producing DRAMs. At that time the total number of DRAM pieces produced was approximately 11 million a year.
Today, there are only five major manufacturers, a reduction of more than 50%, while the number of pieces produced has grown to 3.9 billion, a jump of almost 35,000%. Quite clearly the premise that there are too many players doesn't hold water when compared with market growth. In fact, reducing the number of DRAM makers would probably result in shortages in the next upturn, where demand is expected to grow by a minimum of 50% year over year starting in 2002.
While the bursting of the Internet bubble has certainly played a role in the recent collapse of DRAM prices, an even greater cause was the rapid buildup of DRAM manufacturing capacity on the part of a couple of DRAM companies. When Hynix acquired LG Semicon, at the urging of the Korean government, certain competitors began to fear that they would lose their position in the DRAM industry in the face of this new entity. In reaction, they rapidly expanded DRAM manufacturing capacity by approximately 50% over the last two years, and are continuing to expand that capacity despite the down market and falling DRAM prices.
For example, Samsung plans to invest $3 billion in capacity expansion in 2002-the same amount it invested in 2001, according to J.P. Morgan Securities Asia. Most other DRAM suppliers have lowered their 2002 capacity investments. Interestingly, these are the same companies calling for Hynix's demise.
These DRAM players are not dissimilar to a couple of farmers who, seeing that the price of potatoes is high, plant as many potatoes as they possibly can. Then, when the glut of potatoes forces prices into the basement, they start complaining that it's the fault of the competing farmers over in the next county. It never occurs to them that the problem is one they helped create. Instead, they start complaining to various government agencies and officials about the awful farmers the next county over, all the while continuing to plant more potatoes to get a larger market share.
It's worth noting that the DRAM manufacturers who are crying foul against Hynix have both increased their market share by double digits during this downturn.
What is needed to mitigate the extreme cyclicality that the DRAM industry is currently experiencing is more rational and responsible behavior on the part of all of the players. Hynix, for example, has taken certain steps it feels will help balance the demand and supply situation in the DRAM market. For instance, we have diversified our manufacturing capacity to support the production of non-DRAM products, including SRAM and flash, as well as offering foundry services.
This approach has enabled Hynix to reduce DRAM production by approximately 23% over the past two years, while cost-efficiently converting some older fabs to more profitable business ventures in other markets. In addition, the company has idled some fabs, such as its facility in Eugene, Ore., to upgrade its processes to prepare for the next upcycle. It has also either sold or retired other old fabs to further reduce its capacity.
All these activities not only help to increase Hynix's competitiveness in the market, but also address the very capacity issue about which Hynix's competitors are complaining vociferously, even as they increase their own manufacturing.
Farhad Tabrizi is vice president of worldwide marketing for Hynix Semiconductor Inc., San Jose.