News and New Products

Facing the Walls

By Ed Sperling -- Electronic News, 8/4/2006

Rick Hill, chairman and CEO at Novellus, sat down with Electronic News to talk about the future of the capital equipment business, the next big things in technology and what will hit the wall at 32 nanometers. What follows are excerpts of that interview.

Electronic News: Will 450mm wafers ever become a reality?
Hill: Not while I’m still working.

Electronic News: Why?
Hill: First of all, the industry can’t afford it. If you look at 300mm, it’s been so productive for the semiconductor industry and it’s been anathema for the equipment business. It shrinks our volume, and yet our R&D costs are going up. And the competitive nature of it has gone up such that the margins have shrunk. Going to 450 would be the same thing. Unless it’s funded by the semiconductor industry, it won’t happen.

Electronic News: When you look out at the industry, what is the biggest opportunity?
Hill: The single biggest opportunity is flash memory. Imagine that three years from now, every PC that goes out the door has 32 gigabytes of flash in it. You can’t sell a PC without 32 gigabytes of flash. That’s fast start-up memory. If you imagine that, what would happen to demand for semiconductors? It would go through the roof.

Electronic News: But wouldn’t you want 250 gigabytes?
Hill: You’re not going to get to 250 gigabytes, and you don’t really need that. NAND flash won’t replace a 400 gigabyte drive. However, there’s no reason to have a stall at boot-up, and 95 percent of your application programs don’t need to take time to install. You can use hard disk space just for data and backup. We’ve just seen Sony and Toshiba come out with this already. It’s very high-priced right now. But this will be a hot seller, having the NAND drive. There’s going to be such demand for the instant-on that it’s going to become a requirement in every PC you buy. And it’s not just laptops. We’ll see it pervasive in mainframes in three years. But it’s probably not 250 gigabytes. It’s a matter of economics and density. I don’t see that kind of density three years from now. But I think 32 gigabytes will be economical. When you think about how much you really use, 32 gigabytes covers 95 percent of the population.

Electronic News: How much will that save in terms of power?
Hill: I think that’s a key issue. I would say you would double the battery life—if you don’t have a disk drive along with it. And I only highlighted a PC. What about if all video cameras, rather than going to tape, went to flash? What’s the impact? It’s huge. As you see consumer electronics go to flash, that builds a picture of what the lead horses see in this industry. Intel isn’t stupid. Neither are Samsung, Hynix and Micron. They see it, and they’re racing to the trough. The lead horse has a nice view, and so do the first tier companies.

Electronic News: What else can be done with this technology?
Hill: Energy consumption is a big factor, so portability becomes much more available in a multitude of applications. Access time becomes significantly faster through interleaving and other mechanisms. So you have speed and power. We’ve only talked about one application, too. With a PC, you’re getting to the point where your PDA, your laptop and your cell phone are merging. What’s making that possible is flash. But those are just some of the things going on. The ability for a store to have an RFID tag mandated on every grocery that you have in the store—when that comes off the shelf, you have a system that knows how many others have come off the shelf, how many have gone through the register and how many have walked out in a bag from shoplifters. The low cost of semiconductors will mean trillions of thin, flexible, cheap semiconductors that are going to cost a penny. Imagine the ability to have an active device that’s broadcasting.

Electronic News: What does all this mean for the capital equipment business?
Hill: Ultimately it’s all equipment-dependent.

Electronic News: But is there enough demand to overcome the downturn that analysts are predicting in this segment?
Hill: I said this year would be, ‘In like a lion, out like a bobcat.’ That’s not too far off. This is going to be NAND-driven. It’s also going to be DRAM-driven as Vista comes to market. This will be a replenishment cycle. What people don’t fully realize in the electronics industry is that the depth of penetration of integrated circuits is accelerating with the advent of flash memory because it has so many pervasive uses. We’ve all seen the curve that says the electronics industry is ‘this big.’ That mix is changing. It accelerated when microprocessors came in, then it flattened out a bit. I think with NAND flash it takes another tick up. You’re replacing many, many more electro-mechanical devices with solid state. Even though the overall electronics industry grows 4.3 percent—it’s pretty much tied to the gross world product growth—a larger percentage will be semiconductor driven. That’s the good news. The bad news is that the cost has to come down for that to happen. You have to drive the [average selling prices] down. If you took every device produced today and divided it into semiconductor revenue, you’d come out with a number of about $1.60 and $1.80 per device. That’s driven volume up. But to keep growing, that number has to keep coming down.

Electronic News: That might explain the sub-$1,000 notebook.
Hill: Bingo. But it has to continue, and the question is, ‘How does that continue now that we have a wall of technology?’ We’ve got 65 nanometer proven in production, we’ve got some companies committing to go to 32 or 45 nanometers. I think that going to 32 [nanometers] is going to be tough. But somehow the cost has to go down. It has to come down through productivity.

Electronic News: Why is the 32-nanometer process unfeasible?
Hill: Because of lithography. Unless you can get the productivity up, it’s difficult. I think mix-and-match is what the industry is going to do at 32 nanometers. You’re going to see immersion technology, but it’s going to be at critical layers like transistor alignment. Then you’re going to go away from it when you look at the whole interconnect because you can’t afford it.

Electronic News: Doesn’t that also extend across the device to include what’s on a chip and what isn’t?
Hill: Ultimately with these commodities, it’s density-driven. There are opportunities to change the value equation. If you could split the power in half and cut your packaging costs in half, you could still stay down that cost curve. That’s just another avenue available to the semiconductor manufacturer. Until this point it was easier just to shrink it. But as we get to 32 nanometers, that shrinking isn’t allowed. Now we have to look at all the possibilities to lower the cost. You can lower the capital investment by increasing the productivity of the tools. You can have some design partitioning that allows you to reduce the packaging cost. A major cost of the device today is the package. How do you cut that cost in half? It may be 60 percent of the device cost.

Electronic News: Is one option moving off silicon onto a different material?
Hill: Yes. It will be a different material, or a lower-cost material. Overall, I’m bullish the industry isn’t going away. It’s tougher than I’ve ever seen it, though.

Electronic News: But isn’t part of the issue that engineers need to think outside of just technology to a combination of technology and business issues?
Hill: I think more and more that is the case. It’s not intuitive.

Electronic News: Do you see a shakeout coming because of that?
Hill: I don’t know who will shake out, but clearly human resources will be the differentiator, as they always have been in technology. The guys with the most creative ideas are the ones who will win.

Electronic News: Do you see any change in where your equipment is going? Is it all still China?
Hill: From a manufacturing standpoint, China is still the hot spot. The limiter in China has been capital, and that is a dilemma right now. Investors don’t see the semiconductor industry as the high-growth area it’s capable of becoming, so therefore money isn’t flocking in. People who are investing have positive cash flow. That’s rational deployment of capital. We’ve seen the semiconductor industry transfer to regions based upon subsidization. It happened in Japan, Taiwan, Korea and Singapore. Now it’s starting to emerge in China. But there are pressures on China not to subsidize. The reason is that, unlike all these other countries, it didn’t matter if they subsidized. It couldn’t have a real effect on the world economy. With China, it can. It’s so big it can have a significant impact on the economies of all the other nations. If they start printing money, there’s a problem. There’s tremendous pressure to get them not to do that. As a result, it’s thwarted some of the entrepreneurs in China who anticipated the government would subsidize the industry. It slowed the development from what we would have expected.

Electronic News: Are you seeing any significant growth elsewhere?
Hill: The only other place everyone talks about is India, which everyone talks about as a looming, bright star. When you have that many people, there has to be an opportunity.

Electronic News: But isn’t that opportunity more for the end market than manufacturing?
Hill: Yes, exactly. Manufacturing is more land and resource consumption, which gets down to a population density issue. In India, the population density is 10 times the United States. With that kind of density, where do you put plants?

Electronic News: Are you seeing any nanotechnology applications starting to creep into this industry?
Hill: Nanotechnology is a hype word. There are a lot of applications that are fairly exciting in the medical arena. I haven’t figured out how we take deposition, CMP [chemical mechanical planarization] or electrofill and put them into those markets. It’s a long way off to do it in what I would call an organized fashion. To put down some paint and have all the transistors line up and then put down some wires—we’re a long way from that.

Electronic News: Are you involved at all in the Albany Nanotech project?
Hill: Not yet. Whether or not the technologies materialize into commercially viable entities is the key. We have to pick the ponies that are going to win. There are so many paths you can go down technologically, but there is not enough money to choose all of them. Are you going to put it in Albany Nanotech or IMEC or are you going to use it internally? You can’t do all of them. In the 1990s you could. Because of the growth of the industry, and frankly a lot less choices, you could be sure they were going to pay off. Today you’ve got more choices, with a higher percentage that will never pay off. That makes the process of choosing where you’re going to invest extremely critical.

Electronic News: Has your R&D as a percentage of revenue stayed the same?
Hill: It’s come way down because the revenue has gone up.

Electronic News: Is it getting more efficient in terms of how you use it?
Hill: We’re doing that, too. Internally, we’re getting the revenue dollars up while trying to do more with the same amount of money.

Electronic News: In your product line, what’s the biggest growth opportunity?
Hill: In the near term, it’s PVD/CVD [deposition technology)]. I see growth in PVD in terms of market share. I see growth in CMP because the industry needs an alternative to the people that are there. We’ve got a technology that is very good and that can lower the cost. Those are the key areas. In electrofill, we dominate the market so we have to defend that. There is going to be some expansion in that market because copper will go into memory. There’s growth there.

Electronic News: You mentioned immersion as one technology that will be used at advanced nodes. Will Novellus develop products in that market?
Hill: No. That’s like a recipe for burning dough. I’m not sure who’s going to make money on that, but it certainly isn’t going to be the suppliers.

Electronic News: Let’s talk about low-k materials at 32 nanometers. Will they work?
Hill: When you look at thicknesses of dielectrics at 32 nanometers, it’s roughly 100 atomic layers. If the base material was glass—and for simplicity’s sake say the dielectric constant of glass is 4.0 and the dielectric constant of air is 1—I need to make that 50 percent air. So there would be only 50 atomic layers of glass and 50 atomic layers that are oxygen. It’s obviously not those layers, but there would be some kind of compound in which porosity exists. That’s the only kind of way you get down there. Those are the limits. That’s why I think getting productivity by continuing to shrink is not highly probable.

Electronic News: So will the next big thing be technology or business or something else?
Hill: I think the next big things will be architectural and equipment, from a resource and allocation standpoint, and then grabbing unique business opportunities.

Electronic News: Will this decrease or increase the number of companies competing in this space?
Hill: It will increase the number of small players because they’ll be fast enough to go after the incremental opportunities. It will consolidate the medium players because the bigger guys will become more efficient and tougher to compete with.

Electronic News: One last question—Is U.S. government regulation of what can be shipped and what can’t be shipped to certain countries more of a burden than it used to be for companies such as Novellus?
Hill: I think the government is naïve to think that other governments don’t have the technology. Our customers in Asia can get that equipment from other countries. So you put American business at a disadvantage by doing that. They can get it from France, Japan, and Korea. We always talk to the government and tell them our views, but we try to stay out of the government business.