Listening to a variety analyst conferences this last couple of weeks I heard a variety of executives mention shortages of the components they needed to manufacture their own products. Some simply had to scurry to get parts, others lost sales because they were unable to deliver product.
Extreme Networks, EXTR, for instance, was unable to manufacture its latest and greatest network switching devices in sufficient quantity. They had let their inventories get too low.
Integrated Silicon Solutions, ISSI, reported that there was a shortage of wafers (the starting point for making their memory devices), but that they actually sold more than expected because their rivals in the SRAM and specialty DRAM market were not able to make deliveries.
NVIDIA, NVDA, had the most serious problem I heard. They make graphics processors for computers and mobile devices. NVIDIA's third quarter revenues grew 16% sequentially, but they believe they could have shipped $50 million to $100 million more in product if they had product to ship. The problem was mainly in the newest, high-end graphics chips. TSMC, the foundry that makes them for NVIDIA [and makes much of the world's semiconductors], could not make the chips fast enough. Partly this was due to low process yields on the latest, most miniaturized technology.
AMD's graphics processor segment probably picked up some of the slack, but they are also using TSMC and can't get enough of their newest product either. However, they apparently did a better job anticipating demand than NVIDIA did.
Dot Hill, HILL, was able to get enough disks to ship their storage units to HP, NetApp, and other customers, but reported that they had to work hard to get enough disks and could not guarantee they could get as many disks as they will need in the fourth quarter if demand continues to ramp.
There are probably many more examples, and of course there are always companies products aren't competitive and so are not worried about shortages. But the general pattern is clear. From bottom to top, inventories were taken too low earlier this year, and manufacturing capacity was cut back too much.
Much of the increase in end-user demand is coming from China, but there is a buildup in demand in the United States off the lows of late 2008 to early 2009.
I am expecting robust growth in most segments of the global semiconductor, networking, and computer equipment industry during 2010.
Even capital equipment makers like Applied Materials, AMAT, are looking at a return of demand. It might seem that since on the whole demand is still lower than in 2008, foundries would not need to be buying new semiconductor manufacturing equipment yet. But just as new, better, higher capacity products are always being designed, the machinery for making them has to be upgraded. If someone wants 33nm parts, you can't make them with 90 nm tools. So it now makes sense to buy newer, higher-technology equipment rather than simply restarting production lines that make parts that are no longer in demand.
Disclaimer: I own shares of AMD, HILL, and AMAT. I do not own shares of the other companies mentioned (as of today). - William P. Meyers