** SEMI EQUIPMENT INVESTORS -- MUST READ **
Last week, I attended a three day conference in NYC called SEMInvest. Most of the semiconductor capital equipment companies made presentations during the three day event. You will not find what follows anywhere else (even CNBC) and you should file this information away in the back of your mind.
During Applied Material's company presentation at SEMInvest, Joe Bronson admitted that "we don't know how to call cycles at all." He said their forecasting had been disappointing. When asked by this writer if this upturn would be different from the boom-bust cycle of the 1996-1997 in that the chipmakers may have learned something, he referred back to this comment. He just doesn't know. Scott Kulicke said regarding forecasts that "all tend to forecast but never believe them." George Chamillard, CEO of Teradyne, said he was "cynical" of forecasts. He said one should enjoy the cycle when it turns but hunker down quickly when it turns against you.
But don't argue about the limits of forecasting to an analyst at the conference. Forecasting is their job. Despite the fact that Joe Bronson's visibility is six months at Applied Materials and the company's admitted shortcoming in forecasting, analysts -- at least ones who spoke openly on the topic -- are forecasting a multi-year upcycle in the sector. All during the conference, this was the underlying assumption to the high valuations in this sector. Absolutely no analyst I heard speak or spoke to personally disagreed with this proposition. Gunnar Miller, famed analyst at Goldman Sacks, noted the sector's high relative valuation is due to the fact that investors are discounting a multi-year cycle peaking in 2001-2002. Tia Min Pang of Cowen agreed with the cycle length forecast and further noted that capacity problems would drive greenfield fabs later in the cycle. I spoke to a junior analyst at Montogomery and he believed in the multi-year cycle scenario as did Ed White of Lehman Brothers. And so I believe that the analysts are giving this line to their clients and, as a result, this thinking is pushing these stocks up in price.
And so the high relative historical valuations assigned to these stocks (large caps mostly) are seen as justified in the institutional investment community because it is accepted that this cycle will last longer and create higher earnings due to the leverage achieved during the downturn. This year has already been fully discounted as flat. Nobody seems to care about 1999. It is the equivalent to sitting in the waiting room. Investors have already begun to discount the year 2000 and beyond. And when one is making forecasts that far in advance, anyone can assign earnings numbers to those years to justify current pricing. Investors should be concerned if they see anything on the horizon that threatens the likelihood of the multi-year scenario. One such thing would be a capital spending boom which is unsustainable above 20-21% of chip revenues.
So what else would undermine the analyst's confidence? How about DRAM pricing? Nope. After getting tired of watching hundreds of empty suits sit and lap this prognosis up without batting an eye, I questioned a panel of six analysts on this topic. During the AMAT conference call, I commented to the panel that much was made about DRAM pricing being above $10 for 64 megabites. Good point I thought since 38% of AMAT's revenue was from DRAM equipment purchases this past quarter. I asked if DRAM pricing would be a concern for them in the cycle going forward. One analyst commented that there is a need to drive DRAM costs down by investment so implicit in his response was little concern about DRAM pricing. Jay Dehana noted to me that DRAM producers had cut their spending heavily in the last downcycle and that they would have to make it up. So implicit in his response was a lack of concern as well. No other analyst responded.
Thank goodness for Joe Bronson's integrity. He said he had heard this panel's discussion of the DRAM pricing issue and decided to bring it up during his company's presentation later in the day. He said that DRAM pricing bears watching. Tell that to the analyst community, Joe. He also said that his company's "forecast" was based upon DRAM pricing staying above $9 for 64 megabits and he further noted that most DRAM producers were profitable above $7.50. So while the analysts may not have been concerned, Applied Materials was not ignoring this factor in its forecasting model. So if DRAM pricing should decline, investors should take note even if the analysts do not. Joe Bronson said the next four to six weeks were critical as LG Semicon will go back into DRAM production after a period of inactivity due to a labor strike. Bronson also noted that their forecast is based on a 14% increase in PC growth this year and worldwide economic stability. No analyst even mentioned these factors as concerning them.
So now we know why the valuations are so high in the analyst mind. And it is up to us to remain vigilant against their complacency because if their accepted assumptions are undermined institutions will leave these stocks in a heartbeat. We do not want to carry their baggage should this occur. |
