To: puborectalis who wrote (2737 ) 9/21/1999 10:59:00 PM From: Immi Read Replies (1) | Respond to of 6531
cs boston on quake brm The last significant earthquake to hit a major semiconductor area was on January 17, 1995, in Kobe, Japan. Like Taiwan, the Kobe earthquake was severe, but most of the fab structures were left intact. Like then, our biggest concern with Taiwan is the damage that supporting infrastructure industries, such as electricity, chemicals, and transportation may have sustained. As of late last night, the main electricity grid for the Hsinchu Science-Based Industrial Park was still down, meaning that the fabs located there were working on auxiliary power. Officials at UMC expect power to be back up on Tuesday. We estimate that roughly 20-25% of wafers in process (WIP) could be at risk, translating into approximately two weeks worth of inventory. Based on our experience with other fabs that have had shutdowns, if there was a major power loss, it is only those wafers that were actually in active process that are likely to be scrapped. Over the past 20 months, the Taiwanese industrial park has been hit by two other power outages, so we believe that solid back-ups are in place. During the last outage in July, emergency power helped minimize the impact, as UMC claimed less than 2% of its WIP had to be flushed. However, without a firm time for power restoration, the passage of time could affect other idle wafers, too. Lower exposure to Taiwan could offer a good investment opportunities We think that the uncertainty surrounding the extent of the Taiwan earthquake could negatively affect most semiconductor stocks. However, we believe that investors may find a good investment opportunity should stocks of companies with relatively low exposure to the Taiwanese foundry business negatively over -react. Contrary to what some investors may think initially, most of the franchise players in our Communications IC universe have little or no foundry work being done in Taiwan. Since many of these companies deal in bipolar, BiCMOS, gallium arsenide, or silicon germanium processes, they either manufacture their wafers internally or at domestic foundries, notably IBM. Within this universe, we believe that investors can break down the opportunities according to exposure: Very little / No exposure: AMCC, CNXT, DS, INTC, RFMD, TXN, VTSS. Fair amount of exposure: PMCS, BRCM Large exposure: CUBE, GALT