The equipment industry: Are we bouncing along the bottom?
08/01/1997
The equipment industry: Are we bouncing along the bottom?
Carl Johnson, Infrastructure, Irving, Texas
Every new beginning requires an ending. The havoc caused by over-capacity in the memory markets will come to a close some time in 1998. Whether this occurs in the beginning of the year or near the end does not really matter. The industry is detailing the blueprints for future expansion. We are optimistic, but one must acknowledge the relationship between the growth of worldwide semiconductor revenues and the revenue growth for capital equipment expenditures. Until the semiconductor industry begins to expand in earnest, we expect the equipment business to bounce along the bottom. Many have called 1997 the year of recovery and clearly there are specific areas of strength. Most front end equipment purchases have been for "strategic" reasons, such as filling out production lines and moving to smaller feature sizes. In terms of capacity buys, we believe the capital spending environment can be likened to a state of hang fire - ready, aim, wait.
As Fig. 1 illustrates, the boom of 1993-96 sent capital spending for semiconductor equipment to unprecedented levels. The excess capacity created by this period of "over-spending" has weighed heavily on the equipment industry for the past 18 months. In the past, device manufacturing companies devoted 15-18% of their revenues to semiconductor capital equipment. In 1996, the fraction of revenues devoted to semiconductor capital equipment exceeded 25% - a level that was clearly unsustainable. Though we expect equipment purchases to pick up toward the end of the decade, we have maintained that the
recovery will likely take longer to evolve than many in the industry anticipate.
Every device manufacturer knows that smaller feature sizes and larger wafers are the stepping stones to the next round of profitability. Until the end of 1997, it is unlikely that a full equipment tool set will be available to accommodate these demands. The desire to move to this new level of manufacturing technology is beginning to become more clear. The regional groups, I300I, J300, and the European Group, have all stated that demonstration models will be completed by the end of this year. Plans to build 300-mm fabs have surfaced over the past few months at an amazing rate. According to a recent news item released by Japan`s Nihon Keizai Shimbun, seven of the top ten semiconductor companies in the world will construct production-ready 300-mm fabs by the end of the decade.
Infrastructure sees the efforts of Japanese front end wafer processing equipment vendors at the forefront of 300-mm developments. This intense effort is aimed at gaining back lost market share. A good deal of the market share loss can be attributed to the lingering economic weakness of the Japanese economy. The loss of business has not crippled vendors` ability to work cohesively, evidenced by the intense effort we have seen from the device manufacturers and the equipment vendors involved in J300 and SELETE. Fortunately, SELETE and I300I have now reached a level of understanding that will allow 300 mm to progress along the development path at a more rapid pace.
In any recovery phase, the equipment suppliers will be forced to support the new wafer sizes, and at the same time, address the continued expansion of 200-mm fabs. History has shown that when these shifts from one cycle to another occur, the profits of the chip manufacturers have dropped. As we progress through the summer months, it will be interesting to see if this transition affects the profit lines of the largest sector of semiconductors, the PC market.
The most exciting part of the recovery revolves around the development of new devices that will drive total semiconductor revenues higher. The demand for DSP technologies, the insatiable demand for logic, the embedded memory market, and a widening of bandwidth for the information highway should lead the industry to a new level of profitability.
 |
Figure 1. Semiconductor capital spending ratio.
Test, assembly, and packaging - the strongest segment
Corporate announcements and statistics provided in SEMI`s monthly Express Report and VLSI Research Inc.`s Industry Pulse, continue to suggest smart order growth for the companies supplying tools for test, assembly, and packaging (TAP). This is due to the continuing increase in demand for chip unit volume, new tooling requirements, and the TAP characteristic of shorter delivery lead-times. We believe there has been a great under-investment in these areas, and to accommodate the full contingent of production-worthy front-end equipment/capacity, semiconductor manufacturing companies are committing a higher level of their capital expenditures to the purchase of tools serving the back-end. Figure 2 details capital spending for semiconductor equipment by segment.
Assembly equipment is typically delivered within only six weeks of receipt of an order. This makes it easy for chipmakers to turn off and on the supply relatively quickly, enables them to buy real-time, and minimizes the risk of over-capacity building. Additionally, many new, smaller packages (e.g. the popular micro-BGA) look like they will drive some tooling changes and growth for this sector.
Test equipment has 6-12 week lead times, which makes it a little less "real-time" than assembly, but still a lot shorter than fab equipment and able to recover faster. New tooling is also a driving factor here with demand for faster processors, faster and bigger memories, higher functional density, and "system on a chip."
All TAP equipment capacity is driven by unit volume demand. Throughout this last "slowdown" we saw memory prices drop while unit volumes continued to grow. Thus the raw demand for TAP equipment is still healthy and will be so as long as the pervasive demand for integrated circuits remains strong.
Semiconductor revenues are just about back where they were in 1995 and, barring a general economic slowdown, are on the road to recovery. Even though we are optimistic, we have to ask this question: Is a 20% compound annual growth in semiconductor sales all that believable? Temper our enthusiasm a bit - let`s assume the numbers we show on our chart are too optimistic. Certainly, we will hit the $300 billion revenue level, but the question is when. Total semiconductor revenues near the $260 billion level (20%/year) instead of $300 billion, will likely push total purchases of semiconductor capital equipment toward $50 billion by the year 2000.
No matter how we slice the numbers, the need for additional semiconductor manufacturing capacity will continue to grow throughout this decade, but the pace of this growth will be moderate compared with the go-go years of 1993-95. Furthermore, 1997 will be a moderate year; we will likely only see a significant pick-up in 1998.
 |
Figure 2. Semiconductor capital spending by segment
.
Carl Johnson is president, co-founder, and executive editor of Infrastructure. He also writes a weekly semiconductor column for CMP?s Tech Investor and a monthly column for Microsoft Investor. He can be reached at Infrastructure, P.O. Box 167369, Irving, TX 75016-7369; ph 972/492-7208, fax 972/402-8774; http://www.infras.com, e-mail: [email protected]