Steve Sanghi joined Microchip Technology Inc. as president and CEO in 1990--when the company was near liquidation--and engineered a remarkable turnaround. Today Microchip is the world's leading supplier of 8-bit microcontrollers, has branched out into 16-bit MCUs and analog, and surpassed the $1 billion sales mark in fiscal 2007, ended March 31. In an interview with EE Times' Mark LaPedus, Sanghi pulled no punches as he weighed in on the chip industry at large, the private-equity steamroller, the "fab lite" model and Microchip's place in it all.
EE Times: What is the general state of the semiconductor industry?
Steve Sanghi: The industry, in my view, has a historical challenge that its leadership has yet to recognize. In some cases, where it has recognized the challenge, the industry doesn't know how to deal with it.
Semiconductors are a $350 billion industry. Historically, the semiconductor industry has been built on growing 17 to 18 percent a year. This cannot happen forever. Semiconductor companies are also largely built on these ideas: "Build it, and they will come. Price the parts for tomorrow. Moore's Law. Move to the next geometry."
But the industry is slowing. Most pundits say the semiconductor industry will grow 7 to 8 percent a year. The semiconductor industry needs to adjust to these mature growth rates, but I'm not sure it knows how to do so. And the "build it and they will come" idea just won't happen anymore.
EE Times: What is your view of private equity?
Sanghi: It's not a new concept; it's happened in many industries before. Basically, money is too cheap for the private equities to get. That's the bottom line. They can raise billions of dollars in no time.
Meanwhile, the semiconductor industry has taken a lot of hits in the stock market. The stocks have gone nowhere. So private equity comes in, and they believe that semiconductor companies are sitting on all-time historic-high cash flows. However, I believe some of these private-equity deals will go bust. I believe private equity is going into the excessive, bubble phase.
EE Times: Is Microchip pondering the private-equity route?
Sanghi: Private equity doesn't talk about us. They talk about buying Atmel.
EE Times: The industry is worried about soaring fab, mask and design costs. Are you?
Sanghi: I don't see a step function change here. It's been going on for 30 years. Two-inch masks were more expensive than one-inch masks; four-inch masks were more than two-inch, and so on.
In terms of design, that's where programmability comes in. Microchip serves 55,000 customers. If we tried to make an 8-bit product for each customer and each customer had to make a mask, the cost would be prohibitive. So, we develop one programmable device, and every design team can write its firmware and customize its end product for an application.
EE Times: The company was at a low point when you joined in 1990. Microchip is a turnaround story, right?
Sanghi: When I joined the company, Microchip was up for sale in liquidation. Our sales were flat to dropping. The majority of our sales were commodity EPROM. We were losing $2.5 million a quarter. And manufacturing was pathetic.
The problems were so widespread that I needed a new formula. I didn't take any of the proposals from the consultants. We really needed to design our own system. Years later, we ended up giving it a name: the Aggregate System. It essentially means to redesign the enterprise, [with] all parts of the company working together.
EE Times: What was the strategy
Sanghi: In the product area, the company was involved in too many point products, and the majority of R&D was going into EPROM, which was highly commoditized and making negative gross margins. So we de-emphasized commodity EPROM and focused on 8-bit microcontrollers. Microchip also had an effort in building DSPs in competition with TI. I got out of that business.
In the process development area, Microchip was several generations behind. In manufacturing, our yields were pathetic. So I decided to leapfrog technology development.
EE Times: What's the outlook for calendar 2007? And what can you say about 2006?
Sanghi: We expect another record year in 2007. Last year was a record for us; it was the first year we crossed over $1 billion in sales in both the calendar  and fiscal  years. The surprise last year was, in fact, that calendar Q4 was down for most companies, which was the start of the current inventory correction. All of the customers and everybody were saying they had no inventory. And all of the sudden, there was inventory.
EE Times: Have we seen the last of the inventory glut?
Sanghi: Cell phones had a lot of inventory. Any company that is deeply focused on cell phones had a significant reduction in sales. Second, within the cell phone, you did better if you had share in Nokia, and you did worse if you had share in Motorola. Motorola lost share. The companies exposed to Motorola aren't doing well.
Microchip's commentary at the end of last quarter was that we were seeing the worst of the inventory correction in the rearview mirror. We guided our March quarter to be flat sequentially. Later on in March, we gave an update to that. The March quarter will be flat to slightly up sequentially.
EE Times: What is driving sales--what's your next killer app?
Sanghi: Microchip's strategy is not to look for the next killer application or killer market. We're not a killer-application-dependent company. We're essentially in all of the hot products. We are in iPods and accessories. We are in gaming platforms. We are in set-top boxes and TiVos.
We're balanced by industry segment. The largest portion of our business is consumer. The second largest is automotive, then industrial and computing. In the past, medical had been part of industrial. We started the medical-products group because we saw an increasing need and an increased use of devices in medical applications.
EE Times: What are the hot geographic markets for the company?
Sanghi: Europe has been between 26 to 28 percent of our business for the last seven or eight years. The Americas region had been as much 35 percent of our business; now it's smaller. That's not because the Americas region is growing slower. A lot of American companies design products here and give them to subcontractors that buy products in Asia.
Asia has grown from 33 percent of our business to 43 percent of our business in the last seven or eight years.
EE Times: There is some debate about who is really leading the 8-bit -controller market. Is it Freescale, Microchip or Renesas? And what's driving the business?
Sanghi: Is there a doubt who's leading the 8-bit market? We were at the bottom in 1991 and at the top in 2006. In 8-bit, there are new applications. There are microcontrollers today in hedge trimmers, lawn mowers--things you never thought about.
EE Times: What is your share in 16-bit MCUs?
Sanghi: It's small. We're just entering into the fourth year of volume production in 16-bit, but we're doing well. Our 16-bit business was up 170 to 180 percent last year. It's our fastest growth market.
The second highest growth market for us is analog. Analog was up 40 percent.
EE Times: What is Microchip's fab strategy?
Sanghi: TI is getting a lot of air time with the hybrid, or "fab lite," strategy. That's the same strategy we've been telling investors about for several years now. We have two fabs in production: Fab 2 and Fab 4. [The 200-mm fabs are in Tempe, Ariz., and Gresham, Ore.] Fab 1 no longer exists. Fab 3 [in Puyallup, Wash., purchased from Matsushita in 2000] is available if anyone wants to buy it. It's been up for sale for years.
I think Fab 3 will get sold. If it doesn't get sold, I could equip it. But if I need another fab, I think I can buy a fully equipped fab cheaper than the cost of equipping Fab 3.
EE Times: Has Microchip thought of going fabless? Do you use foundries?
Sanghi: We don't want to go fabless. TSMC builds general-purpose products. Our 0.35-micron technology [in-house], in many cases, can give smaller die sizes than competitive 0.25-micron technologies that are built at foundries. We can also customize our devices. We don't want to lose that.
But we are also using a number of professional foundries in Taiwan and other places. We use foundries if we don't have a given technology. It could be high-voltage, BiCMOS, low-voltage or RF.
Right now, our outsourced production is in the low single digits. It will be going up from low single digits to 10 percent.
EE Times: Will Microchip build another fab?
Sanghi: We don't ever build fabs. We buy other people's mistakes.
The world is going to 300 mm, where I don't need to go. Our products don't require the bleeding edge of lithography at 45 nm. There are so many 200-mm fabs that are up for sale and going on the market. We also have additional clean-room capacity in our Oregon fab. n
July 20, 1955, in Punjab, India
MS, electrical and computer engineering, University of Massachusetts; BS, electronics and communications, Punjab University
President and CEO, Microchip Technology Inc., 1990-present
Vice president of operations, Waferscale Integration, 1988-1990
General manager of programmable-memory operations, Intel Corp., 1978-1988
Author of the book Driving Excellence: How the Aggregate System turned Microchip Technology from a Failing Company to a Market Leader (Wiley; 2006)
Associations and awards: Elected this month to the board of directors of FIRST (For Inspiration and Recognition of Science and Technology)
Board member, Xyratex Ltd.
Member of the State of Arizona Governor's Council on Innovation and Technology
Member of the board of trustees, Kettering University
Board member, Enterprise Network, a Phoenix-based nonprofit organization
Named among Phoenix Business Journal's "2006 Power Players," listing the most influential business leaders in Arizona
IndUS Business Journal's "2006 Person of the Year"
bizAZ magazine's "2005 Arizona's Best Business Leader"
Named by the Phoenix business community as the "Most Admired CEO" in tech in 2002