Rips in the Road Map
By Ed Sperling and Jeff Chappell -- EDN, September 1, 2005
Rick Hill, chairman and CEO of Novellus Systems, sat down with Electronic News to talk about changes in the capital equipment business, economic and technical challenges on the Moore’s Law road map and his company’s competitive position. What follows are excerpts of that interview.
Electronic News: What’s changed in your business over the past year?
Hill: The semiconductor industry is moving from an industrial-centered business to a consumer-centered business. As such, there’s a little more steady demand as opposed to dramatic peaks and valleys. It’s still cyclical, but it’s dampened somewhat both on the upside and on the downside. It’s highly cost-conscious, and that’s rippling down into the equipment industry.
Electronic News: That’s because of the influence of the consumer market?
Hill: Yes. In the consumer market, you have to continue to drive down the cost. So to compete in the equipment market, you’ve got to be delivering value to your customer. Clearly the mission of Novellus is to deliver value. The growth opportunity outside a market is your ability to expand your product portfolio. When you look at where we are today with our product portfolio, we are positioned better than any other equipment vendor to outgrow the market. We’ve got PVD [plasma vapor deposition] coming on. We’ve gone from having six to 18 customers, so there’s an opportunity to grow market share. We’ve got a CMP [chemical mechanical planarization] business and we have a new product out. We have the opportunity to grow market share there. We have the [photoresist] strip business, which historically was strong in Europe and the United States. We’re growing market share in Japan and China. We think, when you look at our opportunities in an equipment market that’s probably growing 5 [percent] to 7 percent, we’re competing in a segment that’s growing about 10 percent -- that’s the interconnect plus the front-end transistor. Plus we have more opportunity to grow market share, so we should be able to grow beyond that.
Electronic News: Do you see new competition from places that didn’t compete in these businesses in the past?
Hill: Clearly, China has its eyes set on getting into the equipment business. I think that’s quite a few years off, however. In the semiconductor business, the ability to ship a product worldwide is fairly easy. It’s at such a level that human-to-human interaction isn’t the key. It’s spec sheet-to-spec sheet interaction. Their value to weight is very high. With equipment, to get a market large enough to support a business it needs to be global. You can’t be an equipment company and sustain it, given the level of R&D, in a single region. Korea, for a long time, has wanted to get into the equipment business but hasn’t been able to because it can’t just support the market in Korea.
Electronic News: Do you think China will build it or buy it, though?
Hill: That thought has entered my mind. It’s possible.
Electronic News: So your competitors may not increase, right? They can simply change.
Hill: There’s no question about that. And in this business, you’ve got to have a competitive edge. That’s what we focus on. Those competitive edges come from the standpoint of innovative new technology, which is pretty defensible. It also comes from a service network to support all your products worldwide. You can’t introduce a product and tell a customer, ‘I’ll be in Asia in three years.’ You’ve immediately got to have the infrastructure. So there are some barriers to entry. But could they buy Applied Materials? Yes. It would be the same people, but it would be Chinese investors.
Electronic News: But is the edge really technology these days? Isn’t it also how you run your business?
Hill: It’s still products. If you have good products that give value to the customer, whether it’s an iPod or a CVD [chemical vapor deposition] system, and you can service it with the full value chain that the customer needs, you’ll be successful with it.
Electronic News: Most analysts are predicting a downturn in capital equipment next year as chips ramp up.
Hill: If they ramp capacity and think it’s going to grow, they’ll be back in the market ordering more capacity. But the rate they’re adding capacity is in much more rational segments, which track unit growth. As a result, I don’t think any slowdown in the future will be for very long. We may see six-month slowdowns, but I don’t think we’re going to see a two- or three-year slowdown anymore. This year I think it will be ‘in like a lion, out like a lamb,’ as long as consumer spending stays where it is.
Electronic News: How much of your sales cater to the bleeding edge? Does it matter if it’s 130 nanometers versus 65?
Hill: Not really. We keep getting a stronger market position at every node. We weren’t around as early as Applied, which made market penetration in given nodes. As those nodes expand, they’re not going to come to us and change out the equipment. If 0.35 expands, Applied will do better because we didn’t have equipment. At 0.15, they’ll get a little bit bigger than us, but not quite as much. At 90 nanometers, they’ll get a little more, but again not quite as much. At 65, we’ll get more. But those are cycles to come.
Electronic News: Do you think there will be fewer players as time goes on, as well?
Hill: Yes.
Electronic News: Any predictions?
Hill: For any given product line, you can have two competitors and good profitability. With three competitors, you start to get irrationality. Any more than that and you have chaos.
Electronic News: We used to think about a ‘Big Three’ in business segments. What’s changed?
Hill: It’s a function of how companies are operating. Everyone wants to cut down the number of suppliers. I don’t want 10 purchasing people talking to 100 vendors. I want two purchasing people talking to four vendors. They’re trying to reduce operating expenses, not manufacturing. The manufacturing is in China now.
Electronic News: How much of your business is in Asia now?
Hill: From the standpoint of end sales, 70 percent.
Electronic News: Where is the remaining 30 percent?
Hill: Between the U.S. and Europe.
Electronic News: Is that number decreasing?
Hill: Yes. The amount of expansion in Asia versus Europe and the U.S. is much higher.
Electronic News: Is that expansion mostly in China?
Hill: So far. There is some talk about building a facility in India, but you’ve got to have power and water -- and reliably.
Electronic News: China had water problems, too, didn’t it?
Hill: Yes, and they put a huge amount of money into Pudong [Shanghai]. As a result, you see a bunch of factories there.
Electronic News: Back on the subject of competitiveness, has your R&D as a percentage of revenue changed over the past year?
Hill: It’s tended to drift downward as our revenues have increased. The dollars have stayed pretty constant.
Electronic News: Are you able to keep it constant as the technology becomes more complex?
Hill: If anything, because of all we’ve invested it will go down as a percentage because of the all the revenue expansion opportunities.
Electronic News: What’s the next step where your investment will go up substantially?
Hill: I suppose it will be 450 millimeters.
Electronic News: You’ve said in the past that you’ll be retired by the time that happens.
Hill: I will.
Electronic News: The only one seriously talking about 450 millimeter wafers is Intel. Can they carry that step by themselves?
Hill: I don’t think so.
Electronic News: How about some of the DRAM makers? Is it more cost-effective for them to migrate to the next size?
Hill: No. The minute you go to the next wafer size you lose re-use.
Electronic News: Looking out at the semiconductor industry, what types of things do you think will drive growth?
Hill: I think flash will drive an enormous number of new applications. Flash is an enabling chip. It’s not DRAM, which followed the computer industry. By having this non-volatile memory at a low price, there are so many other things we can create for productivity purposes and quality of life that we couldn’t do before. That will be a major driver, and for that to happen they’ve got to figure out a way to lower the cost. They’ve got to get more density and lower price. That will expand the market exponentially.
Electronic News: How about 65 nanometers?
Hill: There’s no question we can make it, but can we do it at equivalent yields and costs? It isn’t clear to me that we can. When you talk about a 650 Angstrom trench, and let’s suppose that you want to control the width of that trench by plus or minus 1 percent. That means roughly 6.5 Angstroms. The diameter of an atom is roughly 6 Angstroms. So you’re talking about the capability to control one atomic layer of variance. I don’t know how you do that repeatedly.
Electronic News: Does it become cost prohibitive to manufacture perfect chips?
Hill: All the accuracies stack up on you and you lose it. At 1 percent accuracy, you don’t have enough material to make it work.
Electronic News: Does that mean Moore’s Law hit the edge?
Hill: No, it’s the shrinkage. There are other ways. One of the greatest productivity enhancements in the industry has been copper. It reduced the number of layers that a device needed, reduced the total amount of energy that was consumed because you went from a vacuum process to an atmospheric process, and it shrunk the PVD market. That still keeps you on Moore’s Law.
Electronic News: But when does Moore’s Law, as an economic statement, become unviable?
Hill: I think at 32 [nanometers], it’s out of gas.
Electronic News: How about stacked packaging? Does that offer a way around the 65 nanometer crunch?
Hill: If you could do it economically and make it thicker without taking up more area, it would still be on Moore’s Law. But aligning billions of connections is the problem. If you have a billion transistors, you have 3 billion connections -- source, drain and gate. The minute you want to physically stack two things together, registration becomes an issue. Thermo coefficients of materials becomes a huge issue, too. They may both be the same material, but when you put glue between them it’s a dissimilar material. At 65 nanometers, the thermo coefficients of these materials may change dimensions by microns. That’s the dilemma with the stacked package.
Electronic News: How about new materials such as low k and high k. Are they viable?
Hill: You solve problems in different ways. That’s where innovation comes in. You’re not dealing at 65 nanometers. The ultimate way to get to increasing density would be to connect critical dimensions, namely 65 nanometers to 65 nanometers. We don’t know how to do that. But if we come up seven or eight layers and the top level is 0.35 microns, we start to be able to have the capability to mate 0.35 microns. That’s where the tradeoff is. If you look at materials today, we can identify materials, match coefficients of expansion and make that work. But if you’re talking about taking down the level at which we can connect, I’m not sure we can get there.
Electronic News: How about current leakage at 6 atoms of insulation or less? Is that viable?
Hill: That’s why you’ve got SOI and things like that starting to emerge. That’s a big issue. More and more of your power is being lost. Think about getting down to 32 nanometers and how many electrons represent a one and how many represent a zero. You’re starting to get to 10 electrons versus zero electrons. We’re at 2,000 today. There’s quite a charge with 2,000 electrons. But take it down to 20, and in your best of circumstances, when you want to get to zero, you won’t bleed off all 20 electrons. What really is the difference between a one and a zero is a lot less than 0.7 volts. Now you’ve got noise immunity issues, too.
Electronic News: So what you’re saying is that at 65 nanometers the road map stops?
Hill: At 65, we stop because beyond that is economically questionable. It’s technologically feasible, but not economically feasible.
