Electronic Business Interview: T.J. Rodgers: Programmability, Politics and Profitability | Cypress Semiconductor
Electronic Business Interview: T.J. Rodgers: Programmability, Politics and Profitability
Cypress Semiconductor Corp Founder, President, and CEO TJ Rodgers recently spoke with Electronic Business about the company's shift toward programmable products and away from Moore's Law, how politics have impacted Cypress, and how the company maintains profitability in this rocky economic environment. What follows are excerpts of that conversation.
Electronic Business: Let's talk about the programmability shift in Cypress' core business and the management decisions behind that transition.
Rodgers: In 1982 our original business plan had static RAM memories, programmable memories, programmable logic, and fixed logic in it. Four divisions. We proceeded with that plan for 20 years, and the SRAM part of it caused us to be a Moore's Law company. And we were creating our own technologies, so we had our own fabs and made every notable technology from 1.8 microns on down through 90 nm. The SRAM business at times was glorious for us, as we made $100 million in profit a couple of quarters in 2000, but it was miserable [at other times] and the programmable-logic business and our other programmable products always made money.
So in 2000, we decided to expand our programmables products business. We started an internal startup called Cypress Microsystems and we invented the Programmable System on Chip, or PSoC, our flagship programmable product. We've effected a change in the corporation, which has not been simple. If you consider proprietary products, they now account for 75% of our business, and programmable products account for 50% of our business, PSoC being a programmable product.
We've kind of turned the oil tanker around here and gone from being a Moore's Law memory company to being a proprietary and programmable products company with the majority of our portfolio--and that's led to profitability. We're in dicey economic times in the semiconductor industry and we are doing fine on profit. Typically what happened to us as a memory company is when there was a downturn, we'd lose money for a couple of quarters at the bottom, and in this cycle it's not happening. The upsides will be brighter, as well.
Electronic Business: "Fine" is a bit of an understatement. Your revenue grew 34% last quarter.
Rodgers: I have to remind you that that includes SunPower, but that counts. As another diversification effort, we decided to get into a semiconductor business that wasn't integrated circuits, namely solar energy, which nonetheless needs wafer fabrication and all of the things we know how to do. That business has been doing great for us. We've transformed the corporation by having a solar energy business as big as the rest of the company and we've transformed the semiconductor business itself by going programmable. That's all happened since 2000.
Electronic Business: What prompted the change?
Rodgers: In Q4 2000, Cypress' revenues were $370 million and our bookings were about $500 million. Six months later in Q2 2001, our total corporate bookings were $13 million. We went from $500 million to $13 million in two quarters in the dot-com crash. That's when we pulled our Scarlett O'Hara number and said we're not going to let this happen anymore.
Electronic Business: A lot of your programmable offerings go into consumer products. How are you weathering that with the economic downturn? Do you think consumers will continue to gobble up these products?
Rodgers: You can see the slowdown in the economy in all segments, but it takes a very different form on the consumer side. On the consumer side, you still get design wins, about the same number, people still take products to market, but what happens is your customer ships fewer consumer products than you thought they would ship. Your business is still broad-based, you still have thousands of customers who are active and buying, but their end volumes are down. It's a We've kind of turned the oil much more endurable downturn, where basically, the volumes aren't as high as you'd like them to be but you're still making money and the pricing is still OK. The contrast to that is in the memory market. The way it would work was, just before you'd have a memory crash you'd have a boom, and everybody would be allocating memories. And the crash would happen, and your customers would inform you they had a year's supply of memories and they weren't going to be buying. You'd go into this freefall. You'd be suffering for something like three quarters until the excess memory supply got burned off, and even then, at the end of it, the pricing would be trashed and the volumes mediocre. That sort of boom-bust dynamic you don't see in programmable products.
Electronic Business: What are your estimates for Cypress' revenue from programmables in 2008?
Rodgers: From 2003, which is when we turned on programmable products, our proprietary and programmable products were 48%. From 2003 to 2008, we have moved proprietary and programmable products from 45% to 85%. That's our estimate this year, 85%.
Electronic Business: Can you give our readers the 60-second rundown of PSoC?
Rodgers: PSoC is, I believe, a new category of product. It is a hybrid between a programmable-logic device, a microcontroller, and programmable linear integrated circuits all on one chip. If you think about programmable logic, you've got PLDs and gate arrays, and we've got that on PSoC and we can program it. … We have programmable analog such that we can create different analog functions. … Then we have a programmable memory that is in effect the program memory for the microcontroller. And the microcontroller on there is the traffic cop or the central processor of the whole chip. Electronic Business: Let's talk about growth areas for programmables. There's a "touch" revolution going on right now in electronics. How does PSoC fit in?
Rodgers: When we invented PSoC, we really didn't understand what we had. We lucked out. What I just described is what PSoC is, but it wasn't created to be that. As a diversification effort, we wanted to create a microcontroller product line and when we studied microcontroller competitors that mattered, like Freescale, for example, the barrier to entry was not the technology. The technology for a RAM company was pretty trivial. The problem was Freescale has 5000 parts, and to enter the market and be credible you have to make 5000 different parts. … Our original constraint was design. I wanted to enter by making a family of fewer than 10 chips but I wanted those 10 chips to do the same functions as all 5000 Freescale parts. That was the mindset that we took into the market. What we found out later was that it [PSoC] wasn't a microcontroller, it was a system on a chip, and that the value in it was more the analog-programmable, digital-programmable I/O—more of that than the microcontroller itself. When we made our first PSoC chips, the family was six members. What we call cap sense [capacitive sense], and so-called capacitive touch, the electronic button, if you will, didn't exist. Then all of a sudden we saw some of the leaders like Apple and the iPods, for example, wanting cap sense. That programmable system, the very same one that runs a magnetic door latch, can be programmed with a different program to be a button-sensing machine. … All of a sudden you realize that capacitive sensing is an input, and it's a valuable one. We never made another chip for it. It just worked.
Electronic Business: Where did the touch opportunity start and what are its challenges?
Rodgers: Ever since the iPhone came out, there has been a lust in the market for cap sensing on touch screens. To a first order, those screens have electrodes on them like copper, except the electrode is typically made out of indium tin oxide, or ITO. In effect, it's a trace, but you can see through it. Basically you can light up an LCD underneath it and create a button for an invisible conductor, and when you touch it, you're touching on the LCD screen. That particular problem is a very close cousin to copper, but it turns out the ITO material is resistant and it's more difficult to deal with. You have to program the machine differently and it's specialized. It took us six months to figure out how to use ITO, but at the end of the day the solution for a touch screen is nothing but a software program on the very same chips we've been describing. tanker around here and gone from being a Moore's Law memory company to being a proprietary and programmable products company with the majority of our portfolio -- and that's led to profitability.
Electronic Business: What are your PSoC shipments like?
Rodgers: In 2003 when we had our first shipments we shipped about 4 million units. In 2007, we shipped 160 million units of PSoCs, and that's been in a market where microcontrollers are flat.
Electronic Business: You're obviously very excited about programmability.
Rodgers: One of the things I really like as a person is that it's hard to interest anybody in what a RAM is. You can tell them how many bits it stores or how many books can be fit into it, but after you're done with that, there's not a lot to say. And the other stuff we made, content addressable memory, Internet chips, stuff like that, when somebody says to you at a cocktail party "What do you do?" they are glazing over in about a minute. What I like personally about PSoC as an engineer is that we aren't always cool on consumer gizmos and people do intuitively latch on to the story and they can get it. Cap sense is the new thing and light emitting diodes is the next new craze. Again, it was totally unanticipated for our business line in PSoC. I've got the coolest, most sophisticated light-emitting-diode color-controlling device on a demo board where we, with one PSoC, control red, green, and blue light emitting diodes and allow you to mix any color you want … and we do it all with cap sense. … For us, getting in a consumer market and being able to grow along with things people like is a big attraction. It makes for better hiring. We are now in the position of taking the lucky of five candidates for a position as opposed to trying to convince some engineer that memories are the future.
Electronic Business: Cypress had an excellent year in 2007, growing revenue by 46% to $1.6 billion, an all-time record. Plans quoted in the company's 2007 annual report call for Cypress to break the $2 billion mark in 2008. PSoC should help you break your 2008 revenue goal. What else is helping you along the way?
Rodgers: We have a pretty large programmable-clock business. We also have West Bridge. We noticed when the new cell phones came out and people wanted to start loading songs and or movies into their cell phones, they went away from loading through the cell phone network to loading from the personal computer. We did some calculations and found that the loading time could be tens of minutes. We decided to use our USB technology, which is also programmable, to make a so-called West Bridge product. The job of West Bridge is basically you plug it into a cell phone, you plug a USB into it and it takes full 480-Mbit USB on and off the Web and it can stick programs into NAND flash and not burden the processor on the cell phone while this is happening. It's like a memory-storage subsystem that has a side port to USB. We're engaged with every major cell phone manufacturer and we're ramping up our business this year. We'll be in the range of $50 million and it's going to get bigger next year. … West Bridge is built on our programmable USB. In 1998, we were one of the first companies to that bring out . We have about 40% market share now. We solved the USB problem, not by creating a fixed-function USB chip for mouse and keyboard, but with a microcontroller-based product. We've got an entire family of products that allow you to hook up USB to anything and manipulate information and have general-purpose I/Os that are programmable such that the USB plugs into whatever you need. So when I say 85% proprietary and programmable products, I'm really referring to all of those products even though in a verbal discussion, PSoC is most amenable to "cool" concepts.
Electronic Business: How's the SunPower spinout going?
Rodgers: SunPower has grown to be bigger than the semiconductor side already. We've announced that we are going to spin out SunPower to our shareholders and we're trying to get it done this year. We've grown to about $580 million a quarter, up from our traditional $200 million to $250 million a quarter. We are going to reset and become a smaller semiconductor company in 2008. The growth will come from these programmable products. Electronic Business: How does your manufacturing strategy fit in? You announced you were closing your Texas facility in December. Cypress has three other fabs and does some work in Asia-Pacific.
Rodgers: Cypress' original business plan in 1982 was for those four divisions—SRAM memories, programmable memories, programmable logic, and fixed logic—and in order to make programmable logic and memory, we had to create programmable technology. We have made programmable technology at every node. We're shipping 130 nm right now and are working on 65 nm right now. We have a long history of understanding what's required to do programmable technology. In 2000, as I was looking at what was going to be next generation, I concluded that the floating-gate technology was going to run into scaling. So I did a search of technology and bought the assets of a company called NVM in Colorado Springs. That company specialized in the technology of silicon oxide nitride oxide silicon, SONOS. The way SONOS works is you make a normal transistor, but instead of a normal gate oxide between the gate and the silicon, you put in a SONOS dielectric. The dielectric has the property so that when you run current through it, tiny programming current, some of the electrons remain trapped in the nitride layer between the two oxide layers. The ONO in the middle forms a trap for charge and changes the threshold of the transistor, and that can be used to store data. Since 130 nm, our nonvolatile technology has been SONOS and state of the art. We have developed that in our Minnesota plant and made it in our Texas plant. The reason we are shutting down our Texas plant is not for capability. Our Texas plant is a 6-in. [wafer] plant, and its economics have simply reached end of life. Our Minnesota plant, a 90-nm fab on 8-in., is still our primary at about 80% of our production. We also are working with foundries, specifically with Grace Semiconductor in Shanghai and UMC in Taiwan.
Electronic Business: Do you plan to keep that 80% United States, 20% offshore mix?
Rodgers: It'll shift over time. We will do our expansion offshore to limit our investment in fabs. We will hit a 50:50 point sometime in the future. That will be a few years out.
Electronic Business: A general asset-lite strategy, then?
Electronic Business: If you were starting Cypress today, would you locate it in San Jose, or would you locate it somewhere with a lower cost of living, somewhere that offers a more business-friendly environment?
Rodgers: I just wrote a letter to the editor of the Wall Street Journal. What it said was California is a very hostile place to do business. From Cypress' start in 1982 through the mid-1990s, with the exception of our sales force, we were pretty much 100% California. Today, we have 8000 employees, 7000 of whom are not in California, and we are reducing our footprint there rapidly. It's not just uneconomical to do business here, it is an outright business-hostile environment. … If you go to 47 of the states or virtually any country in the world and say, "I will write a check for a billion dollars to build a plant, to employ people, to create economic benefit in jobs and taxes for your entity, what will you do for me?" You'll get answers like free training, road construction, tax holidays. In California—and I don't know what you want to print—the dumb bastards running the state actually charge you sales tax on your investment. So if I buy equipment for a fab in California, I actually have to pay 6.5% sales tax for the privilege of being here. They don't get it. They refuse to get it. They think they have a birthright somehow to Silicon Valley. What they don't understand is that they have already driven the silicon out of Silicon Valley. Most fabs are gone. Cypress doesn't have a fab in Silicon Valley anymore. LSI doesn't have a fab in Silicon Valley. Intel was supposed to shut its last fab in Silicon Valley and they just told them they have an 18-month reprieve—and they used the word "reprieve." There is no silicon left in Silicon Valley, and we've got politicians who spend their time talking about legislating trans fats out of our diets and other world-shaping things like that. Meanwhile their growth of government spending is preposterous. We have a Republican governor who is Republican in name only. Our budget's in trouble in California. And by the way, I'll say the same thing for George Bush. He's an economic disaster. He's about as Republican as Al Gore is when it comes to running the economy in a frugal way. You can say what you want, you can go on the 6 o'clock news and grandstand, but businesses have to—are morally obliged to—make decisions in the best interest of their shareholders and that is not California right now. I don't think the arrogant people in this state are going to get it until they wake up one day and the goose that laid the golden egg has waddled across the Nevada border.
Electronic Business: So what are you looking for from the next administration?
Rodgers: It used to be that every time I would get outraged, I would write an op ed for the Wall Street Journal or the New York Times, but I don't care anymore. I have concluded that, and I'll use a Reagan phrase here, government is a brake on the economy. It's a drag. It's a parasite. Therefore, I go where government leaves me alone and leaves my shareholders alone. And by leaving me alone, I mean two things: I want government out of my pocket and off of my back. And California is both in my pocket and on my back. So I look for places that want business, and there are many places in the world and in the United States that want business. We have sites all over the United States, not in California, and we have sites all over the world and we simply put our investments where we are wanted and that's doing what's required to encourage business to locate there. Electronic Business: So you'd be looking for a more general acceptance of tech and its importance to the United States?
Rodgers: Yes, and there are a lot of places where that happens. We have a large design center in Mississippi, and they were real happy to get IC designers there. We have a large design center in Bangalore, India, another one in Hyderabad, India, and they were real happy to have companies come in, train people, give jobs, and pay wages higher than the local scale, creating local wealth. And that's where we are investing our money right now, not in California.
Electronic Business: Do you see Cypress' near-term growth being more on a global scale than it is in the US?
Rodgers: Asia-Pac, in particular. I just shifted the first executive vice president, meaning a person who reports directly to me, to Shanghai. He's got 120 people working for him and that's just the kickoff. So yes, we are moving toward Asia-Pac, and that means design, manufacturing, product engineering, not just sales. That's where the business is going. We now export 65% of what we make. It used to be most of our stuff was used in the US and we exported about a third. Now it's the other way around.