Cellular telephony is this year’s big story for the semiconductor industry. More accurately stated, the story is about the fact that handset and cellular infrastructure manufacturers, which consume 20% of all semiconductor devices, are now vital to much of the semiconductor industry. With annual sales of cell phones at 500 million phones per year, this is the second largest single market (in dollars) for semiconductor products.
Of more importance than the numbers however, is the fact that the cellular market is driving the semiconductor industry – its growth, innovation, manufacturing, packaging, sales and investments by the venture capital industry – in much the same way that the personal computer did in the 1980s.
VC firms are aware of this trend. Fabless semiconductor startups targeting the cellular industry attracted the largest amount of venture funding last year, and are likely to continue to lead new startups during 2004.
Of course, when you’re looking at overall demand for chips, you can’t overlook the personal computer industry, which consumes 40% of semiconductor products, according to the Semiconductor Industry Association. Executives across the chip industry have their fingers crossed in hopes that the upswing in PC sales during the second half of 2003 and the concomitant improvement in product pricing for both logic and memory devices will prove to be more than seasonal.
Nor can you overlook important secondary markets for semiconductor devices in “embedded” uses (security, manufacturing, appliances, homes, logistics and automobiles) or the broader consumer electronics market, which utilizes another 7% of the semiconductor industry’s output in several fast growing uses (LCD televisions, game machines, digital cameras and DVD players). Both of these secondary markets continue to deliver on the promise that they will someday surpass consumption of semiconductor devices in personal computers.
But cellular telephony is particularly important to the semiconductor industry in the New Year for several reasons. Chief among them is that cellular phones demand products across an exceptionally wide spectrum of semiconductor companies. For example three-quarters of the semiconductor devices in a cell phone are analog, says Brian Halla, CEO and president of National Semiconductor. This supports strong growth at publicly traded companies such as National, LSI Logic and Agilent. On the digital semiconductor side, processor manufacturers, such as Texas Instruments, Qualcomm, Motorola and ARM, benefit from sales to cellular manufacturers. Producers of low-power SRAM memories, such as Cypress Semiconductor and AMD, also benefit from sales to cell phone makers. Likewise, publicly traded fabless semiconductor companies like Broadcom, Vitesse, Conexant and Cirrus Logic prosper due to high competitive barriers to entry for their products sold to the cellular market.
Add in the host of other firms that make semiconductor products for power sources, base stations, signal conditioners, network processors, network switches, routers and other devices used in cellular infrastructure and you see that almost every semiconductor manufacturer benefits in some way from the ongoing growth of cellular telephony.
To be certain, the cellular market is brutally competitive. Unlike the market for personal computers, which is reduced to competition based on cost and to one- to two-year replacement cycles, cell phone makers must differentiate products due to new standards, new applications (camera phones top the list in this category), and design improvements. New product cycles for cell phones range from six to eight months, barely enough time for sales and marketing organizations to keep up. But it is these same short product cycles and demands for new product features that drive much of the innovation and startups in the semiconductor industry.
As a focal point for design innovation, cellular telephony makes stringent demands upon semiconductor makers. High-quality screens, once considered a gimmick, are rapidly becoming de rigueur in all handsets. Camera add-ons, another once-frivolous feature, are similarly moving toward standardization.
Applications such as games, which were ballyhooed in 2000, have finally begun to appear in significant numbers. Demands for improved reception and sound quality have given rise to expectations for MP3 sound for handsets. Each of these features requires a subset of new semiconductor products that is constrained by power consumption, size or the need to integrate several individual components. Out of the froth of such competing needs arises an almost unlimited opportunity for new semiconductor products.
It comes as no surprise then that the venture industry made over 120 new or follow-on investments in fabless semiconductor companies during 2003, placing nearly $1.5 billion of venture capital in such companies. Of concern to observers, however, is the fact that many of last year’s investments went toward trend- or standards-based startups – just as they have over the last three years – adding to the overpopulation of startups for 802.xx wireless LAN products, imaging and network processor companies.
The coming year will see more investments in companies that offer novel technologies and/or companies that design ever-more complex system on a chip (SOC) products that are designed in conjunction with large customers and which contain significant proprietary intellectual property through both hardware and software. The final quarter of 2003 alone saw one third of VC investments placed in SOC companies. Among the startups that pulled in fresh capital were Europe Technologies, RF Magic, Frontier Silicon, picoChip, Zeevo, Layer N Networks and StarCore.
Of great interest to industry observers is the roster of leading VC investors in fabless startups: Apax Partners, Atlas Ventures, CDIB, CMEA Ventures, 3i, InterWest, JPMorgan Partners, Sequoia Capital, Technology Crossover Ventures, TPG Ventures and Vantagepoint. These firms are likely to lead the pack during the upcoming year as well.
At the same time, VCs are increasingly aware that they can’t invest much more than $50 million per fabless semiconductor startup if they are to achieve healthy returns, says John Michaelson, President of Needham Asset Management. With no IPOs for fabless companies on the immediate horizon, 2004 is unlikely to produce a large wave of such offerings. However, with an increase in stock prices at companies like Agilent, Intel, Cisco, Conexant, Motorola, Philips, STMicroelectronics and TI, acquisitions of startups are expected to flourish in 2004, as these companies return to the market to acquire new technology or to make strategic investments.
There are at least three big questions for venture capitalists looking out over 2004. First, can one profit from the creation of new fabless startups that provide novel products to handset manufacturers? Secondly, can VCs build companies with products or intellectual property franchises in a niche that they can defend? Third, how does one deal with the continuing shift of semiconductor labor (programming, design/layout, fabs and test/assembly) to India and China? Pierre Lamonde, a general partner with Sequoia Capital, says that the trend is irreversible, but at least for now real innovation in chip architecture remains in the United States.