From 7/27, but a good read while awaiting AMAT's earnings:
Susan Billat: Upcycle still strong
7/27/2000 Susan H. Billat is a Managing Director with Robertson Stephens (Palo Alto, California, USA), having joined the firm in 1996 from Ultratech Stepper, where she was Senior Vice President of Marketing. She has more than 22 years of experience in the electronics and semiconductor equipment industries. Ms. Billat holds a Bachelor’s degree and a Master’s degree in Physics from the Georgia Institute of Technology.
The Wall Street Transcript, a Semiconductor Online content partner, interviewed Ms. Billat in conjunction with the recent Robertson Stephens Semiconductor Conference. The following is an excerpt from that conversation.
TWST: What will the most important drivers [for semiconductor equipment] be over the next six to 18 months? Are there any that will become more or less significant as we move through the cycle?
Ms. Billat: Yes, there are some that will become more or less significant. But first, I think the largest driver in the industry is going to be the need for leading-edge capacity. Right now, you can’t open a newspaper without reading about parts shortages . There is simply not enough semiconductor capacity out there. The industry hasn’t gone through a major building cycle of new fabs since the 1995-1996 time frame. And, as is typical with the cyclical industries, the industry followed a period of overbuilding by a period of underbuilding. So now chipmakers are scrambling to add capacity in the face of multiple drivers of the semiconductor industry.
Right now, probably the most visible driver is the wireless communications side. There is also Internet connectivity and consumer electronics, which are growing at a very rapid clip, and PCs, although they don’t enjoy the kind of growth they used to have. PCs are, in fact, growing and are a positive force in the market. Now what is changing in importance? Well, there are two things that I think that are in particular growing in importance. One of them is the move to copper interconnect and low-k dielectrics. New materials are very difficult to introduce into semiconductor processing. We saw a lot of publicity about copper as many chip companies plunged into the copper and low-k processing arena. And then, they needed time to work on the yield problems and the process integration problems.
We’re now starting to see companies deploy copper more and more and as a result one of the things that will be interesting to watch over the next year is the rate of copper adoption and its companion, low-k dielectrics (which can be used separately or together) and how rapid that adoption is. However, it does require a lot of changes in processing equipment.
One specific example is oxide etch. Oxide etches occur many times during the fabrication of a silicon wafer but have not been really critical. With a copper process, the oxide is used to define the channel where the metal will go because of the nature of copper processing. All of a sudden, oxide etch is a leading-edge battleground for equipment makers, as opposed to not having been a key focal point in the changing and processing. So copper has broad implications.
Another trend that I think is gaining in importance is the adoption of 300-mm wafers. Unlike copper and low-k dielectrics and many other process changes that are technology and technology performance-driven, the move from 200- to 300-mm wafers is strictly a matter of economics. The equivalent 300-mm process tool costs about 30%-40% more than the 200-mm tool. The chip maker is able to realize about 2.5 times the number of chips per wafer. It’s very compelling. The difficulty is that a new 300- mm wafer fab will likely cost $2.5 billion. So if you look at the revenues of chip companies, it is not obvious that a lot of them can afford to spend that kind of money on one new factory.
We think this will drive the foundry model and continue to drive chip makers to outsource their chip manufacturing to the foundries. And, of course, the largest and second largest foundries in the world, Taiwan Semiconductor Manufacturing Company (TSMC) and United Microelectronics have already announced their 300-mm plans. TSMC’s 300-mm wafer fab is under construction right now. The foundries will stand there with open arms as chip makers decide whether to build their own 300-mm fab or outsource.
Intel has been very vocal on the subject of 300-mm wafers. It, I believe, is leading the pack and pushing chip equipment makers hard to have high productivity 300-mm tools available. And, of course, Intel is the largest single purchaser of capital equipment. But as an interesting side note, if you look at the top four capital equipment purchasers, two of the four are those two foundries I mentioned, TSMC and United Microelectronics. So foundries are very big players here.
The 300-mm fab requires an entirely new tool set. We believe, for instance, in the early running, Applied Materials, the world’s largest equipment company, is seeing a higher percentage of design wins with 300-mm fabs than they saw with 200-mm fabs. If that trend were to continue, it would certainly be good news for them.
TWST: Do you have any concerns about the sustainability of the uptrend or the length of the cycle? Are there any developments that could disrupt the positive trends that you’ve discussed?
Ms. Billat: The industry, historically, goes through overbuilding and underbuilding cycles. This cycle is a very major one, in my opinion. It is not going to be short-lived. But eventually, I think the industry is very likely to, once again, overbuild. The question is, at what point might that happen and who might the culprits be?
<b.I don’t think it’s going to happen this year. I actually don’t think it’s going to happen through most of next year. In part, there’s a governor on the growth of those lithography tools which are in short supply. These tools have multimillion-dollar lenses and the specialized glass that is used in those lenses takes a long time to manufacture. You have to actually melt the glass and cool it and do what’s known as an anneal, very slowly. Certainly chip makers do want more lithography tools, which are an essential part of processing and represent about 30% of the total budget spent on front-end capital equipment. Those tools are simply not available at the rate that chip makers would like. That alone slows down the pace of growth. Other components are also in short supply due to the steep building ramp the industry is seeing, so virtually all equipment makers are capacity constrained.
If every equipment maker had every tool that every chip maker wanted, we would probably have a much bigger year this year and a sharper drop-off, but that’s not possible. And even with that, my personal estimate is that we will see over 50% growth in front-end capital equipment this year and nearly as high growth in the back end, which is assembly and test.
Eventually though, we will run out of capacity. I have a checklist of things to look for, one of the first being: Are equipment makers accounting for too high a percentage of the revenue of chip makers? Historically, the highest the chipmakers in aggregate have ever spent on equipment was 31%, which I believe was in 1996, and that precipitated a major downturn, because chip companies were overspending.
TWST: Are there any [recent IPOs] that you’d like to highlight that Robertson Stephens has brought public?
Ms. Billat: There are a couple of companies. One of my favorites was Rudolph Technologies, the maker of the metrology equipment. They are a play on the copper market and also on the capacity editions the industry is seeing. They make equipment that measures film thickness, and they have a proprietary technology that enables them to measure the thickness of copper wafers, the thickness of copper on production wafers directly rather than having to process a test wafer, which is time consuming and expensive. We think that Rudolph will continue to do very well with the adoption rate of copper, but they also make measurement equipment for transparent films. I can’t think of a fab in the world that doesn’t have at least one piece of Rudolph equipment in it. So all the new fabs that are being built benefit Rudolph as well. Another company that we brought public, that I think is very exciting is Cabot Microelectronics. Cabot is a maker of slurries. Slurries are a wet chemical, kind of a gritty substance, in a suspension, just like a slurry you might use in old-style manufacturing. They are used in the chemical mechanical planerization (CMP) process. VLSI Research estimates that CMP will be the fastest growing single segment of semiconductor equipment in 2001 and I believe this year as well. And all of those tools drink slurries. So Cabot, which enjoys greater than 80% market share of the slurry market by our estimates, we think is very well-positioned. And of course, the industry is moving to 300-mm wafers also. The area of a 300-mm wafer is more than twice the size of a 200-mm wafer and should require a proportionally higher amount of slurry that should help continue to drive demand for Cabot. So Cabot is a favorite of ours.
Edited by Katherine Derbyshire
Managing Editor, Semiconductor Online |
