2003 outlook a little better

The first half of 2003 brought weak sales for both semiconductor manufacturers and their suppliers. Early signs of a recovery from the industry's ongoing downturn faded under the pressure of global uncertainty about war in Iraq and the mysterious SARS virus. Still, both Dataquest analyst Dean Freeman and Bill McClean of IC Insights expect the second half of the year to show strong improvement. Personal computers purchased during the last major upgrade cycle, in 1999, are now well behind the technology curve. After two years of low business investment, both analysts expect demand for upgrades to recover in the fourth quarter of this year. Freeman expects solid but unspectacular growth of about 8% for the year for both chips and equipment. McClean was a little more optimistic, expecting 15% growth in both chip and equipment sales. Both forecasters, speaking at the SEMI New England Breakfast Forum last week (Burlington, MA), expected strong growth in 2004 and 2005. As in most cycles, a surge of electronic systems demand is likely to drive overinvestment in fabs, leading to another downturn in 2006. (See Thin Film Wiki: Chip Market Dynamics.)

McClean noted that semiconductor material sales surpassed equipment sales for the first time in 2002, a sign that unit sales of chips remain strong. Yet electronics equipment suppliers face devastating pricing pressure. Inflation rates have fallen steadily for the last several decades, from 5.7% per year in the 1980s, to 2.7% per year in the 1990s, to only 1.9% annually in the opening years of this decade. Some observers, notably Federal Reserve Bank chairman Alan Greenspan, worry that deflation is a very real possibility. It is difficult or impossible for electronics suppliers to raise prices in this environment. The resulting pressures on margins stress the entire industry supply chain.

For several years, chip suppliers were able to grow more rapidly than electronic equipment suppliers because the semiconductor content of devices was increasing. Semiconductor content now appears to have stabilized at around 18-20%, McClean said. Similarly, capital expenditures have averaged 20% of chip sales for years. Though equipment suppliers are being asked to shoulder a larger fraction of industry research and development costs, rising overall costs leave little room to increase equipment's share of the total fab budget. Photomasks, raw materials, and the operation of the cleanroom itself compete with capital equipment for scarce dollars. Both McClean and Freeman expect equipment growth to parallel overall chip growth.

Back to top

At the same time, Freeman observed, equipment suppliers are facing lengthy delays between investment and income. The semiconductor industry's technology Roadmap calls for rapid introduction of a range of new technologies, from EUV lithography to low-k dielectrics. In the past, new technologies have actually reached fab floors two or three technology generations later than their first appearance on the Roadmap predicted. For example, the 1997 edition of the Roadmap predicted that materials with dielectric constants less than 3.0 would be needed for the 180-nm technology generation in 1999. Current predictions expect those materials to enter production with the 90-nm generation in 2004, a five year difference. The gap between the first production use of leading edge technology and industry wide acceptance can be as long as three years. Equipment suppliers therefore face a six to eight-year gap between their initial target availability dates--when research dollars must be spent--and actual volume production, which allows them to recoup their investments. Particularly given the industry's cyclical nature, only the largest companies are likely to maintain continuous efforts for such long periods.

Overly ambitious fab construction at industry peaks leads to overly conservative investment during downturns. Global capacity shrank 2% in 2002, McClean said, as semiconductor manufacturers mothballed equipment to cut costs. At the same time, every company wants to have capacity in place when the recovery comes, knowing that those who do not will be left behind. For instance, Japan's extremely cautious capital spending since the mid-1990s has seen that nation's share of the overall semiconductor market drop from 40% in 1994 to 22% in 2002. The foundry industry, in contrast, launched itself into prominence with aggressive capital purchases. Even in 2002, facing painfully low utilization rates, Taiwanese foundries spent 40% of their revenues to build new capacity. Mainland Chinese foundries are following the same road, and have been one of the few bright spots for equipment suppliers in the last few years.

Still, McClean warns that such aggressive spending cannot continue indefinitely. Eventually, both Taiwanese and emerging mainland Chinese foundries must turn to banks and capital markets to finance further expansion. When they do, they most convince investors that they can maintain profitability and deliver good returns. Over the long term foundries, like other fabs, are likely to spend about 20% of revenues on capital expansion.

Such aggressive foundry purchasing practices may have contributed to the severity of the current downturn. Building excess capacity to gain market share has been common practice in the DRAM sector for years. The resulting dramatic swings between overcapacity and undercapacity have been largely confined to that sector, while periodic collapses in memory prices help logic and microprocessor suppliers by reducing the overall cost of more advanced systems. Similar swings in the foundry market can have a much more serious impact. Any chip designer can turn to foundries for manufacturing. Foundry overcapacity therefore depresses margins throughout the industry.

To summarize, the near-term outlook for semiconductor manufacturers and suppliers is certainly better than it has been for the last two years. Nonetheless, neither Freeman nor McClean expects a return of 2000's 80% equipment market growth. Instead, the industry is likely to see an extended period of flat prices and tight profit margins.