The New Mr. Chips: An Interview with T.J. Rodgers

Speaking for Silicon Valley's upstarts


People who like T.J. Rodgers describe him with words like intense, brash, and cocky. People who don't like him—and there are plenty of them in the semiconductor business—use more colorful language. But all of them have to admit that Rodgers is very successful, both in running his company and in derailing the latest fads in industrial policy.

At 41, Rodgers—a Stanford Ph.D. and, in the Silicon Valley tradition of one high-tech company spawning others, a "graduate" of Advanced Micro Devices—is president and CEO of Cypress Semiconductor Corp. The $200-million company is doing very well, thank you, despite the national fuss about Japanese competition in the chip business. Last December, Rodgers went to Congress to testify against U.S. Memories—a consortium of semiconductor companies designed to get the United States back into the commodity chip business by manufacturing DRAMs (dynamic random-access memory chips, pronounced dee-rams).

U.S. Memories would have gotten special exemption from antitrust laws—an exemption not available to Cypress or other companies—and was hoping for federal loan guarantees as well. But by the time T.J. Rodgers had finished attacking the idea—in Congress and in the press—U.S. Memories was a goner, unable to gain either private financing or political support. America, Rodgers maintains, has been spared a high-tech Amtrak.

REASON Editor Virginia Postrel and her husband, Steven Postrel, a management professor at UCLA, interviewed Rodgers at Cypress's San Jose, California, headquarters.

Reason: In 1983, when you started Cypress, entrepreneurs were hot. Silicon Valley was looked to as the salvation of the country. Do you see a change since then?

T.J. Rodgers: Entrepreneurs are still regarded as generating technology, generating jobs. But the press has equated Silicon Valley with the entrepreneurs who got famous when Silicon Valley first came to their attention. So "Silicon Valley" equals "Intel, Advanced Micro Devices." What's really changed is that the people that run those companies are not entrepreneurs anymore. What real entrepreneurial company can afford to have a director of government relations or an office in Washington, D.C., that doesn't sell chips? What real entrepreneur has time to run to Washington to lobby for subsidies?

The classic example is [the government-financed R&D consortium] Sematech—absolute classic example. We have the big guys who are hurting, primarily self-inflicted wounds. They go to Congress. Declare America is in trouble because they're in trouble. They scream "Jap." They want $100 million a year to save America. And they get it.

The theory is the government puts in 100 million a year and they put in 100 million a year. Now, phase two is how do they channel the government money back to their companies and get away with it? First, there's mandatory sending of employees to Sematech. Well, in '85 and '86 [when the industry was in a slump] that's called a layoff. You can take your turkeys and send them to Sematech and subtract their salaries from the dues. Their newest boondoggle is a program where Sematech members get to use Sematech steppers—which are pieces of capital equipment costing roughly $2 million—for free.

Reason: Back in 1984, when people were talking about industrial policy, the argument that shot it down was the government shouldn't be picking winners.

Rodgers: Yes, yes.

Reason: The new argument is, "OK, we'll have an industry-led industrial policy. We'll allow the captains of industry to get together and essentially decide how the government money should be allocated." Is that what happened with Sematech?

Rodgers: Who's the captain of industry? The captain of industry is the guy who's got the best lobbyist in Washington. I'm absolutely sure saving America is equivalent to saving Intel in [Intel CEO] Andy Grove's mind—and he's a bad guy to pick because he runs a good big company. And I'm quite sure saving America is saving National Semiconductor in [National Semiconductor CEO] Charlie Sporck's mind. It's not at all clear to me that America's better off by using government money to save National Semiconductor.

The captains of industry happen to be the guys who lost it for us. Now they can claim that they are the losers because the Japanese played unfairly. They can claim they are the losers because they can't borrow money. But there are a whole lot of winners in the industry who lived on the same playing field, even at relative disadvantages to these guys.

Reason: For example?

Rodgers: Us. We're seven years old. We're $200 million a year. We're making 23 percent profit. And when I worry about who's sneaking up behind me, the members of Sematech are not on the list. I worry about other companies like us. I worry about Integrated Device Technology. I worry about Performance Semiconductor, which is a private company. I worry about small companies that are very energetic, very competitive, and have got good technology.

Reason: Why, in your mind, did the big companies lose?

Rodgers: Management. Lousy management. It's very clear. The Japanese manage manufacturing better than we do. We manage research and development on the R side better than they do. They manage research and development on the D side better than we do. And the handwriting's clear: If you don't work on manufacturing management—which equates to high-quality products, low-cost products, fast manufacturing—you're in trouble. We had the message in the early '80s and we didn't do anything about it. And to a large extent the large companies aren't doing anything about it right now.

Reason: Alfred Chandler, the eminent business historian at Harvard, had this quote in Fortune magazine: "Silicon Valley is kind of our entrepreneurial dream—and a disaster. Those companies are very good, but they don't make big investments in production and continued research. The three firms that did—Intel, Motorola, and Texas Instruments—are our best hope today." There's a feeling that a lot of these newer wave companies go off and have their designs made somewhere in Malaysia or Taiwan—that somehow they are stripping the industrial base of America.

Rodgers: Whoa…Whoa…Wait. First of all, the big guys are much more foreign than we are. We're an American company, I don't make anything in Malaysia. You want to find things made in Malaysia you ask Gordon Moore [chairman of Intel, or you ask Charlie Sporck, or you ask Jerry Sanders [CEO of Advanced Micro Devices] and you'll find as many jobs in Malaysia as in the United States. We do all our designing here. We do all our R&D here. We do all our manufacturing here. Which they do not. So if you want to have an American contest, I win hands down.

We invest 26 percent of our sales in R&D. There is no big company that invests 26 percent of their sales. It's more than double what Intel invests…

Reason: Few companies period.

Rodgers: Right. And we do it consistently, while we're reporting 23 percent profit. I don't think that if you took 10 medium-sized semiconductor companies and took some big ones and then took investment in R&D as a percentage of sales, you would find a difference. And do we really think a dollar invested in research and development would be better spent as the billion and first dollar in Texas Instruments' central research lab or the first dollar that Cypress invests in research and development? Where does the productivity come from?

Reason: Someone I talked to in Washington referred to you as "the guy who torpedoed U.S. Memories." Is that a fair characterization?

Rodgers: God, I hope so. U.S. Memories was a preposterously bad business deal. It was concocted by the guys who brought you Sematech. Now the original premise of Sematech was to make dynamic RAMs and compete against the Japanese. Government wouldn't let them do it. The government said rightly, "We're not going to subsidize a competitor who competes against other American companies."

So U.S. Memories was the adjunct, the dynamic RAM company. They made it ostensibly private. But it had the guy who did the site search for Sematech running it. And he negotiated deals—city and state subsidies. Ranging from a minimum of $46 million in San Jose, California, to a maximum of over $300 million in Massachusetts. He would claim that he wasn't subsidized—his lawyers had firmly planted the idea that "subsidized" equals "receiving direct federal government subsidy payments." Well, hell, that's interesting, but that's not what subsidized means to me.

So they get this guy, Sandy Kane, to run U.S. Memories, whose credentials were that he was a government relations director at IBM. This is the guy who's going to go beat the Japanese in manufacturing. He presents a business plan. Think about it as a venture investor: He has no fabs [chip fabrication plants], no customers, a product maybe, no credentials himself to run it, no other founders.

This guy wants $500 million in equity, plus whatever carried equity would go to the employees. Plus another $500 million in debt. So if you think about what venture capitalists want—they want to see a 10-to-1 to 20-to-1 return on their money in three to four years. Take the least stringent of those criteria, and this $500 million company needs to be worth $5 billion in four years. We drop half a billion dollars on Sandy Kane. Some bank, without any security at all, lends him another half a billion dollars. And our government relations director from IBM builds a $5 billion company in 4 years, doing more in 4 years than Intel has done in 20 years. Preposterous bullshit.

The argument is, "Well, we have other motivations to get people in." And there were. Say I run a big company. My stock's in the mud. It's been in the mud for three years. I'm getting raped by the Japanese because now they own a monopoly on DRAMs—which I walked off from voluntarily. Now, I could spend $50 million and get back in the DRAM business, with a high likelihood of failure, because I failed before. I've got $50 million in the bank. I could afford to spend it, but I can't afford to lose it.

So, how in the hell can I do this thing and not claim a loss? Simple, I'll give it to Sandy Kane and let him lose it for me. If you buy less than 20 percent of the stock of a corporation, you don't have to consolidate the losses. So, you buy 19 percent of U.S. Memories. Sandy Kane agrees to transfer you 19 percent of his product at his manufacturing cost plus 10 percent.

And one of two things happens: Sandy Kane is successful. He runs a zero profit company. I transfer my losses to him. I take his product and sell it. I've got the DRAM page back in my catalogue, and I'm home free. Or, Sandy Kane craters. In which case, some time three years from now, I write a letter to the shareholders and say, "Well, we made profit last quarter. We're going to make profit next quarter. This quarter we have an extraordinary writeoff, $50 million. Sandy Kane screwed up. Not my fault. Too bad. Nice try. We tried to help America. One-time event." Neatest little scam you've ever seen. That's U.S. Memories, in a nutshell.

Reason: You talked about the importance of buying your way back into the DRAM market. Do you think it's important for American companies to be major producers of DRAMs?

Rodgers: It would be nice if the big guys hadn't walked off on it. Gordon Moore, the chairman of Intel, admitted that Intel deliberately walked out of the DRAM business to make more profit. Two years later they made a profit of $600 million in microprocessors. Which is where they squirreled their money instead of DRAMs.

Reason: So maybe it was a good decision.

Rodgers: Sure. And I think DRAMs are a crappy business. The reason they're a crappy business, is that it's the highest volume integrated circuit. So it attracts governments, not companies. Right after the Japanese have created a monopoly on it, the Koreans are going to destroy their monopoly, and the Indians or Chinese will follow. There will be no free competition. There will be subsidized losses.

Reason: Part of the party line you get from people like the Semiconductor Industry Association is that it's only if you make DRAMs that you can keep up with the progress in putting higher density—more and more memory—on a chip. If you lose the capacity to design and make new generations of DRAMs you lose everything. Is that accurate?

Rodgers: No, it's bullshit. DRAMs and fundamental technology are very well matched. But there are three other vehicles that are as good as DRAMs for driving technology. They are ROMs [read-only memory], EPROMs [erasable programmable read-only memory], and static RAMs [random-access memory]. Now, one of the best foundries in the entire world is the Sharp foundry in Japan. And the main vehicle for driving the Sharp foundry in Japan is the ROM, the read-only memory. Sharp makes most of the ROMs for Nintendo games, gigantic quantities.

The EPROM is also used by many Japanese companies as a technology driver. It's used by Advanced Micro Devices and by Intel. The EPROM is a single-transistor memory. That single transistor is very special—it stores charge and holds it when the power is turned off.

And the last one, our technology vehicle—at Cypress, Motorola, Integrated Device Technologies—is the static RAM, SRAM. SRAMs are four times more expensive per bit than DRAMs, and they're four times faster. Typical computer systems have a fraction of memory in SRAMs for quick access and then the bulk of the memory in DRAMs for cheap price. The SRAMs are a very good technology driver.

Reason: Charles Ferguson of MIT, among others, has criticized Silicon Valley in particular and U.S. industry in general for being too entrepreneurial. He refers to "chronic entrepreneurialism"—people constantly leaving their employers to start new companies. You disagree with him. Why?

Rodgers: Well, where do you draw the line? I mean where does chronic entrepreneurialism start? Right after Andy Grove gets done telling me about how Cypress weakens Intel, is he going to tell these stories about the immaculate conception of Intel? They ripped off Fairchild [Semiconductor Corp.], right? The old companies, the old boys, the reformed hookers, were very much more cavalier in their appropriation of technology from their founders than we are today. Ferguson talks about the weakening effect of losing people, but you know, Rod Canion, the president of Compaq—is he better off being a department manager at TI?

Reason: You've said that there's no quick fix for competitiveness problems.

Rodgers: That's right.

Reason: Is there a slow fix?

Rodgers: Sure. Work. We have a high work ethic here at Cypress. We run the company on the basis of revenue per employee. Managers who want to hire more people have to come in and demonstrate increased workload—growth—and increased productivity—output per employee—and then they have to show that their workload increase, even with the increased productivity, still means that they have to grow the number of employees. And only then do they get to hire more people.

Reason: So at Cypress you're fighting this battle by working. But are there changes that need to take place, either at the governmental level, or educationally…

Rodgers: You're talking about the tide rises for all companies, as opposed to what a company can do to be competitive?

Reason: What needs to change outside the companies.

Rodgers: Sure, I can name a bunch of things. Education. The government is only good at throwing money at something and running away. The government cannot manage. So what can you throw money at that you can get some value for your money? And I claim that our universities are pretty efficient. You can send a $10 million check to Stanford and say, "We want this used for integrated circuit research and development. We'd like a proposal from you, before we validate this check, on how many more Ph.D.s or whatever you want turned out." You would see more chaired professors, more Ph.D.s, more courses, more stuff on integrated circuits, and the efficiency would be pretty high.

Two: The budget deficit has to stop. It is the root of all evil. I heard a great quote the other day: "We always hire Democratic congressmen who promise to give us from the government all the things we want. And we always hire Republican presidents to make sure we don't have to pay for it." It's a great quote. It's exactly right.

Reason: Is your proposed solution to the budget deficit to raise taxes?

Rodgers: No. They have to shut down some of their spending. They just have to do it. I think we have the opportunity of my lifetime with this reduction of tension with the Soviet Union if we can have the courage to use the peace dividend just to get rid of the budget deficit, not to go with porkbarrel projects.

The government borrowing hurts companies like mine and all technology companies for several reasons. Equity is expensive money. Debt's cheap money. If I have to sell 10 percent of my company to get some money, as opposed to just borrowing it and paying interest, I'm very much better off borrowing it. In Japan, I can borrow it for 6 percent. Here I can borrow it for 12, because I'm competing with Sam, the biggest borrower in the history of the universe.

Next thing is our government policy of the cheap dollar. Go to 1984. Represent Japan and America by a $1 billion Nippon Electric Corp. and a $1 billion Intel, respectively. We each had 50 percent market share. Five years later in '89, the yen is valued up by a factor of two. Intel's still $1 billion, assuming they didn't change in real size. Nippon Electric Corp.'s now a $2 billion company. Headline: "U.S. Market Share in Semiconductors Drops from 50 to 33 Percent. Sematech, U.S. Memories Needed to Save the Country."

The whole goddamned thing, you can explain with currency. It turns out if you use the 1989 exchange rate, we had a 28 percent market share in 1984 and we've been climbing from 28 percent to 35 percent since 1984. Our currency has been devalued partly because it needed to be devalued, but it's been pushed farther than the real equilibrium. When you cut Intel in half by cutting the value of the dollar in half, you cut its profit in half. Its PE at best stays constant, so you cut its share price in half. So you've removed our ability to access capital markets. We have to stop crunching our dollar to favor the rustbelt industries.

There's one more point, which falls under trade policy. Japanese story of getting into jet engines—Clyde Prestowitz told this story at a conference I was at. Here we own—we own—jet engines. You got Pratt-Whitney, you got what's the other one?

Reason: GE.

Rodgers: GE. Then you've got Rolls-Royce. But the Japanese don't know how to do it. Gigantic trade secret. So all of a sudden, you read an announcement that GE's going to make a 90,000-pound jet engine. Think about it. The SR-71 Blackbird—that's a Mach-3 airplane, fastest airplane that was ever made, flies at 105,000 feet—has two 30,000-pound engines in it. We're talking about a 90,000-pound engine. So you think it's an ultramilitary secret, nobody gets at it. Then you read the next day that the Japanese are going to have 15 percent of the manufactured content of that jet engine. And you go, "Do these guys have a brain tumor? They are letting the Japanese in on the most advanced jet engine in the world. What the hell is wrong?"

Well, Prestowitz explained how that happened. MITI says to Japan Airlines, "You will do what we want and here's what we want." Japan Airlines goes to Boeing and says, "You know us. We buy more 747s than anybody. We can buy 747s, we can buy Airbuses, and we're not sure right now. We want to make sure that whatever 747s we buy from you in the future have a little bit of Japanese content in the jet engines. Boeing, do you hear what we say?" Boeing goes to General Electric and says, "You know us. We're your biggest buyer of engines. Now we can buy them from you, we can buy them from Rolls-Royce…" And the Japanese are in. [Editor's note: Boeing says no Japanese company is involved with engine technology on any of its planes.]

Now you can argue that that's crooked. I can argue that that doesn't matter. But the point is, those guys play hard ball, on issues like that, and we don't. We need to say, "You ship us $10 billion of goods, well, we got this little problem. You don't have American content. And starting June 1991, we're going to have a small surcharge—200 percent of their value coming in—unless you have 20 percent American content. And it's not fair, and we're not intending to be fair. So that's it. That's the new rules." And we could start doing things to the Japanese, like they do to us. Which would raise the tide for an entire industry, are totally non-free market, in terms of the world. But with regard to the United States, help our industry.

Reason: And the consumer…

Rodgers: That's the one problem. The fact is that if we do things like that we hurt our own consumers. But I guess when I look at us losing it and think about the American consumer, my vision of the American consumer is a first grader who weighs 220 pounds and is sucking on a lollipop. I just don't mind putting the kid a little bit through boot camp. And I think if we judiciously do some things to protect strategic industries in this country and if they happen to have an impact on the consumer, I think that's a good exchange. If the Japanese take it all, the consumer will lose anyway.

Reason: Who gets to pick which industries are strategic?

Rodgers: There is a place where I trust a science advisory group to the president…

Reason: Sounds like Sematech to me. A bunch of experts get together, civilian scientists who do a lot of work for the Pentagon, and they say, "Well, you know, we've really got to help these guys."

Rodgers: It's not Sematech in the sense that it's not picking winners and losers within a sector. It is basically raising the tide for an entire sector. They've been doing it for years with supercomputers. So that one limited thing is what I would do. The purpose isn't to screw consumers. I'm not looking for policies—the good example is the one Dr. [Thomas Gale] Moore, at the Hoover Institution, gave on cars. We go jack up the price with a tariff on Japanese cars…

Reason: It would be better if it were a tariff. It's a quota.

Rodgers: A Honda goes from $20,000 to $23,000. The price of an Oldsmobile goes from $20,000 to $23,000. And the Oldsmobile now lives under an umbrella that prevents him from getting competitive. And the American consumer gets screwed by both Honda and Oldsmobile, as opposed to just Oldsmobile.

Reason: Isn't that what the agreement Clyde Prestowitz rigged up to restrict imports of Japanese memory chips did?

Rodgers: It's a disaster. I am not a supporter of that particular deal. The Japanese broke the law. We should have bashed them for it. But after they forced all our companies out—or our companies voluntarily exited a low-profit market, whichever way you want to look at it—we forced them to jack up their prices. We forced our American computer companies to pay billions of extra dollars to the culprits who screwed us to begin with. It was a disaster.

On the other hand, the Japanese changed their way of doing business. We pulled out our six-gun and shot six holes in our foot. But they went, "Holy shit. The gun actually came out of the holster. This is bad news." And there was an absolute, measurable change in the Japanese. Nippon Electric Corp. is making supercomputers. All the memory in the supercomputer will be American. We're getting that. It's like taking some medicine that almost kills you and kills the germs. It's real close as to whether it's worth it. With all the bad side effects, it did wake up the Japanese and did cause them to change the way they're doing business. We, Cypress, will benefit from that.

I don't mind doing things that are "unfair," things to a foreign country that I wouldn't tolerate between companies within the United States, or between the government and companies in the United States. But I don't have much meat on that yet. Because frankly for a long time I rejected all those arguments at face value. And most of the time, when you see the grand scheme that's going to make things better, it only takes you about five seconds to find out who's going to get screwed. I am sensitive to the fact that with every restriction we talk about you can see a clearcut feedback back into hurting somebody in the United States.

Reason: Often another producer fighting foreign competition.

Rodgers: Sure. And sometimes, as in the DRAM pricing thing, to give benefit to an industry which no longer exists.

Reason: What do you see as the proper role of business and entrepreneurs in politics? You were very disparaging of the people who have the governmental relations offices in Washington.

Rodgers: I'll run a good company and pay taxes—and we pay a lot of taxes in this company. I guess we have to be available to testify and give our opinion, but I don't really see a need for lobbying groups. There are only two reasons for them: One is to get subsidies, and the other one is to prevent the government from sticking it to you some way, so it's defensive. Neither of those have any interest for me.

Reason: Are there particular people in public life that you admire?

Rodgers: I was very impressed with all the congressmen I met. They're very smart guys. It's a lot like when you start studying quality at your own company—you find out bad quality is your fault. Really and truly your fault. And most of your workers are pretty smart and try to do what the company tells them to do. When they screw up, it's mostly because they were given confusing and incomplete instructions. When I look at Washington, I think that the system is really more the problem. In other words the management is equivalent to the system; the players, the workers, really are pretty good people in general.

So do I have any heroes? My last hero in Washington is Ronald Reagan. I like Reagan because of his leadership, not because of his management. He's probably a pretty lousy manager. And he didn't work very hard. But he was a good leader. A great leader. And he brought enough respect back to the presidency to get two terms out of it. And that, that to me is an important accomplishment.