Semiconductor sales for 2009 will be down anywhere up to 25% from about $250 Billion in 2008, and with sales exceeding 2008 levels only in 2012 or 2013. The ASP [average selling price] for all memory makers is in the trash, and for high-margin companies like Linear (LLTC), Maxim (MXIM), Analog Devices (ADI); their sales to automobile companies has evaporated. Currently, the only company making real money is Intel (INTC). In fact, if Intel had not entered the flash business [through the IMFT joint-venture], they would be making even more money.
NVDA is doing fairly well in their niche - GPUs, and AMD should probably think about spinning off ATI to the public. In fact, the latest ATI chips are as good as or better than NVDA in processing power. Among the large memory makers, most of them are, for all practical purposes, dead. In fact, they will get to lose less money - if they just shut down the FABs and stop selling chips - even if they had no change in staffing. Among the memory makers, only SNDK and SSTI have a stream of licensing/royalty revenues - which will fall directly to the bottom-line, but hardly sufficient to stop the haemorrhaging of cash and other assets.
Now, why are they in such a predicament?
a. Technology improvements. The migration to smaller geometries allows for 60 to 70% more storage per unit-area - this is fantastic for the consumers though.
b. Overbuilding of fabs. This remains the #1 scourge of all semiconductor companies. When the cycle swings to the upside, the industry usually goes on a fab-building binge, which hurts them when things go bad.
c. The hiring of engineers who are clueless about chips, especially from Asia. While I might be biased when I say this, I have seen many hundreds of “engineers” who are clueless about what it takes to actually put a chip out in the marketplace.
d. Being too complacent about taking on debt or being forced to “invest” in FABs for guaranteed wafer-starts. Even fabless companies like SanDisk are forced to take equity stakes in FABs - making the fabless model irrelevant for larger chip-makers. In fact, any company that wants guaranteed wafer-starts purchases an equity stake in their respective FAB/FABs [examples: SNDK, BRCM].
e. While idling FABs in lean times is a good idea, it usually fails to serve its purpose [to reduce supply] - since one cannot just shut off $3M a pop machines from AMAT, NVLS, LRCX. Instead, they are on a production “rotation” which might affect yields [negatively, of course].
f. And finally, the most important reason for their failure is the fact that no Product Line Manager ever plans for two consecutive years of price declines of 70% annualized. In other words, no PLM [that I know of] planned that a newly introduced product would be decimated in price in two short years. Most assume a price decline of 40% [and that is aggressive], and possibly, a migration to a more advanced technology after two years [to cut manufacturing costs].
And if this wasn’t bad enough, the Jan book to bill ratio was one of the lowest that I have ever seen - 0.48 [that means that if a company ships a dollar worth of product in Jan 2009, they have on their books only 48 cents in orders]. This means that Gartner (IT) and IDC are wildly optimistic about the future of the semiconductor industry - at least for 2009. A 33% shrinkage of the semiconductor industry will mean a book to bill ratio of 0.67.
Despite all of this gloom, some companies are positioning themselves for success - when the market bounces back. Among the contenders are:
a. Applied Materials.
c. Silicon Storage Technologies
e. Lam Research
f. Novellus (NOVL)
g. Kyocera (KYO)
i. Texas Instruments (TXN)
j. Linear Tech
l. Marvell (MRVL)
n. Altera (ALTR)
p. Lattice Semi
r. Micron (MU)
In Part II, I’ll pick the five best positioned to survive this nuclear winter. In fact, I’ll pick the companies with the least debt and the highest profit margins - since they alone can possibly survive and then thrive when things get better.
>>> Go to Part II