Jon,
I apologize for the confusion. You might need to be a subscriber to access the article. I'm not sure. Here's the meat of it:
I think that the days when the personal computer determined the speed and course of pretty much the entire chip industry are coming to a close. We're in the midst of a transition to a decade when products like cellular telephones and digital video-disc (DVD) players will dominate demand for semiconductors. It's time to broaden your technology portfolio to include the companies that will benefit from this trend.
... But it took a presentation I heard last week at the New York Society of Security Analyst's semiconductor conference from Michael Lins, a senior analyst at Cowen & Co., to pull the pieces together for me.
Lins divides the recent history of the semiconductor industry into two pieces. First, there's the period when chip demand was dominated by the needs of the personal-computer industry for memory, processing, display, control, storage, etc. During those years, global sales of semiconductors rose at an annual rate of 21%, and the personal-computer industry accounted for half of that.
That period, which began in 1986, started to draw to a close last year. We're now in a transitional phase that will last until 2000, he feels, when the relative importance of the personal-computer industry's demand for chips starts to decline. .... But it's clear that we're at the beginning of a very big trend. For example, Lins projects that the new digital video-disc players alone will consume $3.5 billion's worth of semiconductors annually by 2000.
And there are so many different industries and products going digital. Photography has long been an "analog" industry. ... Alexander Graham Bell invented an analog telephone ... Following the path blazed by the recording industry, which moved from analog records and tape to digital compact discs, both of these industries are increasing using digital "0s" and "1s" to represent scenes and sounds. That requires more silicon to turn analog signals into digital ones, to process the digital signals, to store them, to control and synchronize their speed, and then to reconvert the digital signal back into the analog waves and photons that our ears and eyes can process.
The digital data delivery tail will be wagging the dog. That dominance, Lins projects, is good for at least 10 years.
An investor who can find a company that is scoring design win after design win knows with some certainty that higher revenues are on the way
Take apart a digital video-disc player, for example, and you'll find a micro-controller, digital signal processors, video signal processors, audio signal processors, chips for converting mixed signals, and chips for converting digital back to analog signals. You'd find the same jungle of chips if you opened up a digital wireless phone. That will be a $7 billion market for semiconductors by 2000. Seem far-fetched? Nokia (NOK/A) and Ericsson (ERICY) are reporting growth of 100% a year in their digital handset businesses.
The increasing volume of digital signals creates a need for other digital products. ... Companies are installing fiber-optic cable as fast as they can, and as fast as they can afford, ... But until that network is in place, chip companies will make big money from products that expand the carrying capacity of the current network. Switches, routers, local area networks, digital printers, more disc drives, new kinds of portable storage -- all those require more chips. Semiconductors that help jam more signals into existing pipelines will record sales of $800 million a year by 2000. This part of the semiconductor industry will be rowing at least twice as fast as that connected with personal computers by the year 2000, says Lins. The digital-data delivery tail will be wagging the dog. That dominance, he projects, is good for at least 10 years.
I don't yet have all the particulars figured out, but it's clear to me that picking stocks in the two parts of the semiconductor industry requires different kinds of analysis. The personal-computer chip industry is characterized by rapid obsolescence ...
In contrast, many of the chips in the fastest-growing part of the semiconductor industry will have long lives. It's not uncommon for analog semiconductors -- chips that read analog conditions such as temperature and turn them into digital signals -- to have life cycles of five years or longer. Chips of this sort have much longer gestation cycles too. Getting from design to producing samples for customers to try may take two years; from sampling to significant revenue could be another two years. But since the competition takes place at the design stage, when chip makers are trying to convince the Sonys (SNE) and Motorolas (MOT) of the world to use their new chip in the latest camcorder or cellular phone, revenue growth is actually more predictable. An investor who can find a company that is scoring design win after design win knows with some certainty that higher revenues are on the way.
If capital is a less crucial determinant of success in this part of the industry than among the personal-computer chip makers, intellectual property is more indicative of a company's likely prosperity. Chips in this part of the industry are often customized for a single customer. Over time, successful companies build up libraries of building-block designs that they've used in the past and that can be quickly re-used in a new setting. The better those building blocks, the more of them a company owns, the better its system for putting them together for a new customer, the more future business a chip maker is likely to garner.
Lin ended his talk by listing the companies that he most likes as a way to play this trend: Analog Devices (ADI), Burr-Brown (BBRC), Linear Technology (LLTC), LSI Logic, Maxim Integrated Products (MXIM), National Semiconductor (NSM), Texas Instruments and TriQuint Semiconductor (TQNT). ...
But I was especially intrigued by his mention of TriQuint. ...
"The true way to play gallium arsenide is to buy a basket of all three (talking about VTSS, ANAD, and TQNT)," he said. The chips, which are much faster than comparable silicon-based semiconductors but also more expensive and harder to produce, have found their first real mass application in the digital telecommunications industry. These three companies could easily wind up selling 100 million units a year to the makers of wireless phones alone, and the market could grow to $2 billion by 2000 from about $500 million now. The chips are so difficult to produce, and the manufacturing process requires so much tweaking, that these three companies, which have built up a base of production knowledge, have a significant edge on any new players. Unlike the other two stocks, TriQuint hasn't run up much in price since early June. With 17 recent design wins by Lin's account, the company is about to see significant growth. While it's hard to call any stock trading at a price-to-earnings ratio of 42 cheap, this does seem to be a good time to buy shares.
I'd certainly buy Anadigics on any weakness. Lin says these chips are so hard to manufacture that companies blow up periodically, so you'll have another chance to buy it. Same with Vitesse, if you failed to buy it when I wrote about it in June. But the bottom line is that I love to get in at the beginning of a trend -- particularly one that could run for the next ten years. |
