Research Brief 4Q02 Update: Semiconductor Capacity Still on Hold Abstract: Semiconductor capacity expansions have gone into a hold mode as soft semiconductor demand drops utilization rates lower. Further significant expansions will wait until demand growth returns in 2003. By Bob Johnson Recommendations Semiconductor equipment manufacturers must prepare for a period of low visibility and unpredictable sales levels as wafer fab management attempts to tailor capacity expansions to uncertain device demand growth trends. Fab management needs to monitor overall capacity trends closely so that they are not caught unprepared if demand for semiconductors surges unexpectedly. Equipment manufacturers must not assume that slow demand for 0.13-micron equipment in late 2002 and early 2003 will delay the initial demand for 90-nm equipment, scheduled for first pilot lines to be operational in late 2003. Automation equipment suppliers should focus near-term efforts on tool automation products because the next new wave of fab construction will not occur until mid-2004 or 2005, with the first production from new fabs in 2005. Publication Date: November 26, 2002
2 4Q02 Update: Semiconductor Capacity Still on Hold Introduction Earlier in 2002, capacity utilization throughout the semiconductor industry appeared trending to return to desirable levels by the end of the year. Then semiconductor demand softened, and projected demand growth for the third and fourth quarters of this year fell to anemic levels. While semiconductor manufacturers responded to the lackluster economic news by cutting back on capital spending in the near term and sliding out planned expansions, sufficient new capacity had been put in place earlier in 2002 to drive utilization rates downward again. Gartner Dataquest estimates that overall fab utilization rates peaked at about 77 percent in the second quarter of 2002 and will remain slightly below that level until demand growth resumes in the second half of 2003. For leading-edge fabs, defined as 0.18 micron and below, utilization peaked at about 94 percent in the second quarter of 2002, but it should drop to the high 80 percent range by year-end and gradually increase to 90 percent through the first half of 2003. The overall capacity picture will respond to this trend by slowing expansions through the remainder of 2002 but resuming growth in 2003. While the wafer fab equipment (WFE) market will not show "robust" performance in this period, equipment will still be sold, and the majority of that will go for incremental capacity increases as needed at fabs throughout the world. These marginal increases will tend to keep overall utilization rates slightly depressed in the first half of 2003. By the second half of 2003, leading-edge utilization should return to above 90 percent, ending the year at a projected 95 percent. This will provide the impetus for strong growth in the equipment industry in 2004 as capacity increases to meet increasing semiconductor demand. Business Trend In the aftermath of the 2001 to 2002 semiconductor market collapse, semiconductor manufacturers have become more cautious with their capital spending plans. By the second half of 2002, most leading-edge semiconductor manufacturers were solidly entrenched in the 0.13-micron process generation, with basic processes developed and debugged to the point at which yields were becoming reasonable. At the same time, the 90- nanometer (nm) generation is still well over a year away for everyone but the most aggressive manufacturers (Intel and Taiwan Semiconductor Manufacturing Co. [TSMC] both plan 90-nm pilot lines by the end of 2003). So the appropriate strategy for capacity building is to wait for demand to materialize, then incrementally ramp capacity to meet demand. This approach is made easier because equipment manufacturers in almost all segments except implant and lithography have very short lead times, so commitments to incrementally increase capacity can be delayed until the last minute. Consequently, we expect worldwide capacity to increase very slowly through the first quarter of 2003, then begin an upward trend that should
3 Figure 1 Worldwide Semiconductor Capacity extend through 2004. Figure 1 summarizes Gartner Dataquest's estimates for total worldwide capacity by wafer size through 2004. MSI 1,400 1,200 1,000 800 600 400 200 0 4Q97 2Q98 4Q98 2Q99 4Q99 2Q00 4Q00 2Q01 4Q01 2Q02 4Q02 2Q03 4Q03 2Q04 4Q04 300 mm 200 mm 150 mm 125 mm 100 mm and Below 111537-00-01 MSI = Millions of square inches Source: Gartner Dataquest (November 2002) Table 1 summarizes the total estimates for empty shell capacity for both 200- and 300-mm fabs that had their initial production in 2001 through 2003. As shown, significant 300-millimeter (mm) capacity began to come on line in the middle of 2001, and it has been growing slowly ever since. By now, most 300-mm fabs have reached minimal production volumes, with typical installed capacities of about 5,000 to 10,000 wafers per month. However, Table 1 only begins to tell the tale. Because the overall capacity declined to an apparent plateau in 2001 and 2002, a relatively large number of fabs still achieved initial production status during that time frame. However, few of the new fabs have reached their fully planned capacity, and the majority of them have significant "empty shell" capacity available. Empty shell capacity refers to the difference between the current installed capacity and the ultimate planned capacity for any fab. It represents the potential of the empty space in existing fab buildings. Taken in the aggregate, it gives us a picture of the amount of capacity that can be brought on line fairly quickly because no additional facilities construction is required.
4 4Q02 Update: Semiconductor Capacity Still on Hold Table 1 Total Fab Empty Shell Capacity, 2001-2003 Initial Production Year 2001 2002 2003 200-mm Capacity Number of New Fabs 15 10 7 Total Planned Capacity 342,850 257,000 150,000 Total Current Capacity 155,510 98,500 0 200-mm Empty Shell Capacity 187,340 158,500 150,000 300-mm Capacity Number of New Fabs 8 5 13 Total Planned Capacity 150,000 130,000 297,600 Total Current Capacity 50,500 27,000 0 300-mm Empty Shell Capacity 99,500 103,000 297,600 MSI per Quarter 200-mm Empty Shells 28.3 23.9 22.7 300-mm Empty Shells 33.7 34.9 100.9 Notes: Capacity is stated in wafer starts per month. MSI-per-quarter data are provided for a comparison with Figure 1 totals. Source: Gartner Dataquest (November 2002) The total empty shell capacity represented by existing fabs is enough to provide sufficient additional capacity for the industry well into 2004, given Gartner Dataquest's forecast assumptions. This means that we would not expect to see new fabs being announced for greenfield construction during 2003, except for entirely strategic reasons. However, this assumes a nearly perfectly efficient market in terms of allocating production to available capacity, and this just does not agree with reality. Accordingly, we anticipate that there will be a few new fabs announced in 2003, and the next wave of new fab announcements will gather momentum through 2004, when capacity shortages will appear again. The other side of the fab capacity picture is fab utilization. Our projections for fab utilization through 2004 indicate that while utilization rates will not be sufficiently high to drive a full-blown wave of new fab spending until then, they will provide the impetus for incremental capacity expansions to fill the empty shells. Figure 2 shows our projections for both total and leading-edge utilization rates through 2004.
5 Figure 2 Worldwide Fab Utilization Rates Percent 100 90 80 70 60 50 40 30 20 10 0 4Q97 2Q98 4Q98 2Q99 4Q99 2Q00 4Q00 2Q01 4Q01 2Q02 4Q02 2Q03 4Q03 2Q04 4Q04 All Fabs Leading-Edge Fabs 111537-00-02 Source: Gartner Dataquest (November 2002) Gartner Dataquest defines "leading edge" as the two most-current process generations. Thus, through 2003, leading edge includes both 0.18- and 0.13-micron processes, but as 90 nm begins to come on line in 2004, the leading edge shifts so that by year-end 2004, it includes only 130- and 90- nm production. In the near term, utilization rates should decline, in response to both incremental new capacity brought on line in 2002 and a short-term softness in the semiconductor market. By mid-2003, total utilization should be about 80 percent, and the leading edge will cross over the critical 90 percent level, which provides the impetus for continued capacity growth. Gartner Dataquest Perspective Fab operations managers are extremely cost-sensitive, and, given the choice, they will delay fab additions as long as possible to minimize fixed costs and maximize margins. This is exactly what we expect them to do during the next six to 12 months as semiconductor demand slowly returns to a strong growth condition, and the need for more capacity becomes critical. We are seeing a significant number of announced reductions in planned capital spending in response to the softness in utilization rates toward the end of 2002. Such reductions are to be expected because fab management delays capacity expansions, which they don't need at this time.
6 4Q02 Update: Semiconductor Capacity Still on Hold However, new spending initiatives can be announced just as quickly as spending cuts, and semiconductor manufacturers will not be caught short on capacity once semiconductor demand returns. This is one time when we have to be skeptical of announced capital spending plans through 2003 because device demand remains shrouded in uncertainty. It won't take much to bring overall fab utilization back to the point at which new capacity is needed, and then we'll see the purse strings loosen selectively to fill those shells. Key Issue What is the size and projected growth of the semiconductor manufacturing markets and supply chain elements? This document has been published to the following Marketplace codes: SEMC-WW-DP-0219 For More Information... In North America and Latin America: +1-203-316-1111 In Europe, the Middle East and Africa: +44-1784-268819 In Asia/Pacific: +61-7-3405-2582 In Japan: +81-3-3481-3670 Worldwide via gartner.com: www.gartner.com Entire contents 2002 Gartner, Inc. All rights reserved. Reproduction of this publication in any form without prior written permission is forbidden. The information contained herein has been obtained from sources believed to be reliable. Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. Gartner shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof. The reader assumes sole responsibility for the selection of these materials to achieve its intended results. The opinions expressed herein are subject to change without notice. 111537