The chip industry is changing from PC-centricity to a more-diffuse focus on the ubiquitous network. When the transition is complete the chip industry of this "Net World Order" will differ from that of the 1990s in some noticeable ways. Ironically, the promising communication applications targeted by numerous chip makers are among the primary contributors to one of the industry's most severe downturns.
The downturn may herald a new and dangerous source of cyclicality for the chip industry. Telecommunications deregulation set the stage in both wireless and wireline telephony for competitive capacity investments, which inevitably give rise to overbuilding and shakeouts. The chip industry will probably face a similar boom-bust cycle with future generations of network-based technology.
Although the problem was "technology push"-bringing product to market before the market was ready-the solution may be a different type of technology push. Chip suppliers can hasten a return to their normal growth by putting more effort into developing the broadband services to soak up excess network capacity. That excess capacity has contributed to the current malaise in telecommunications.
The development of services would appear to be beyond the core competencies of most chip companies. The strategy requires alliances with some combination of software developers and network owners. Chip makers that successfully reach beyond their usual base of customers and suppliers in this way will likely be rewarded with a head start in the big markets of the 21st century.
The downturn, for all its brutality, will not last forever. With that in mind, we chart some trends from the industry's recent past to make predictions about its future at a regional level.
One of the most striking features of the chip industry in the 1990s is the massive weight of a single company, Intel Corp. This can be seen in the chart that shows the tracks of annual shares of chip revenues for the Top 40 producers classified in four regions: North America (primarily the United States), Japan, Europe and non-Japan Asia-Pacific (primarily Taiwan and Korea).
As is well known, U.S.-based companies lost global share in the 1980s as DRAM production shifted to Asia and then recovered share in the 1990s. What is perhaps not recognized is how much the recovery has been based on Intel's tremendous performance. The U.S. share without Intel is almost flat since the late 1980s, while Intel's share of global chip revenues expanded from about 5 percent in the 1980s to almost 20 percent by the end of the 1990s.
By the late 1990s, the U.S. share (including Intel) recovered to roughly what it had been in the early 1980s. Producers in Taiwan and Korea expanded their shares modestly, while Japanese producers experienced a steady revenue decline that appeared to bottom out only recently. Japan's relative decline was caused in large part by the massive entry of the Korean memory producers, just as Japanese producers had taken market share from U.S. DRAM producers a decade earlier.
To more fully understand these global dynamics, we looked at the competitiveness of the regional groups, which we measured as their ability to sell beyond their home regions. The second graph, based on customized Dataquest data on the sales by companies with headquarters in a given region into other regions, shows that the concentration of each group's sales within its home region declined from 1992 to 2000. The decline is lowest for Japan, which was already the most inward-focused at the beginning of the period. Moreover, Japan was the only semiconductor market to decline, falling from 31 percent of the world market in 1992 to 23 percent in 2000.
In other words, Japanese manufacturers as a group rely heavily on a market whose global importance has declined. This apparent loss of competitiveness in overseas markets is clearly a major force driving the retreating global market of Japanese chip makers. In contrast, the global competitiveness of European producers dramatically improved while their own regional market held steady at roughly 20 percent of world semiconductor consumption.
What can these trends tell us about the future? First, it seems unlikely that Intel can continue to maintain its privileged position. Although excellence in coordinating large design projects and in volume manufacturing provided the underpinning of Intel's phenomenal performance, the enormous margins it has earned owe a great deal to its ownership of the X86 standard. Would-be competitors have been forced to ensure compatibility with Intel's instruction set, which left them running a race with a considerable handicap.
But after years of misfires, Advanced Micro Devices has finally put a significant dent in Intel's market share, beginning with the sub-$1,000 PC market that Intel was slow to serve. And even as Intel faces stiffer competition in its primary market, it must divert resources to build a position in the Net World Order where it does not enjoy a standards-based advantage.
Intel has recognized that processors will remain at the core of all Net World Order products, but it has been forced to simultaneously pursue a range of architectures, including the X86, StrongARM and Micro Signal (for digital signal pro-cessing). Instead of focusing its considerable design resources on redundant development of a single chip in a market that it controls, Intel must pursue numerous projects in fragmented markets where it does not have much leverage and where customers have learned the cost of allowing a supplier to control a major standard.
Intel will retain its industry leadership for the foreseeable future on the strength of its entrenched PC position, its success in moving into higher-value server markets and its continued achievement in driving down its per-chip manufacturing cost, but its fortunes are likely to wane along with the PC market, which seems ready to morph into several submarkets that may or may not have Intel inside.
If Intel were to decline during the coming decade, does a resurgent Japan stand ready to capitalize on consumer acceptance of the Internet and regain its global market share? Though we can't rule out such an eventuality, this turn of events seems unlikely.
Japanese chip makers still seem unable to unleash the innovative capacity of their engineers. In the interviews that constitute the bulk of our research on the industry, managers at systems companies criticized several Japanese ASIC suppliers for weak support from sales staffs and unreliable performance.
A look at recent industry initiatives in Japan is revealing. Most of these collaborative efforts, such as 300-mm tool evaluation, the development of extreme-ultraviolet lithography or the launch of an intellectual property trading center, parallel or repeat efforts in the rest of the world. Although these programs are nominally open to foreign concerns, they are in practice almost exclusively Japanese. This inward gaze may be the Japanese chip industry's weakest feature, at least until the Japanese economy emerges at last from its long malaise.
U.S. producers are more likely to face global competition in the Net World Order from European producers, since they have steadily improved their ability to make acquisitions, form alliances and sell in foreign markets along with their own large, integrated market.
Greg Linden (firstname.lastname@example.org.) is a Post-doctoral Fellow at the Center for Work, Technology and Society, University of California at Berkeley. This article relies on research done for the Sloan Semiconductor Program.
Return to 2001 Midyear Forecast