The royalty walk quietly among the common folk here. The thick-carpeted mahogany-veneered opulence typical of high-tech presidential palaces is nowhere to be seen. Instead, on an ordinary floor in an ordinary building overlooking a nondescript cube farm are two adjacent rooms that would probably be conference rooms at your company. Not the special, glossy, wood-wrapped, electric-view-screen-and-sound-system, “we entertain important clients in here” style of conference rooms, but rather the “grab your coffee and duck in here so we can scribble some stuff on the whiteboard” type. In your company, each of these would probably be named after a river or stream in the nearby county, and there would be a sign-up list somewhere online that nobody uses, except that annoying rule-follower dude that always comes in and bumps you out because he has a “reservation” even though the identical room next door has been unoccupied for weeks.
Today, at Xilinx, these two rooms are repurposed as offices. Looking in the door of one, we see potted plants wrapped in florists foil, well-wishing memorabilia, gift-wrapped packages, a stack of greeting cards and photos, and several boxes in the process of being packed. Peering into the other, we see the signs of recent arrival. On one end sits a desk hastily arranged for work, but not yet bearing the usual “lived in” signs. On the other end, a conference table is placed adjacent to a whiteboard-coated wall with yet-to-be-unpacked boxes strewn about. Sitting proudly atop the pile, yet to be hung but already dominating the space, is a framed, autographed version of Pele’s famous #10 jersey with a photo of the legendary athlete signing his name.
In the first room, Wim Roelandts, the departing President and CEO who has led the world’s largest and most successful programmable logic company for the past dozen years, is working his last day. In the second, Moshe Gavrielov is moving in to take the reins of a company continuing to challenge the status-quo in programmable logic – forging new growth in a semiconductor segment the company has dominated for two decades.
This austere and unpretentious leader lair is part of Wim Roelandts’s legacy. Wim has made a habit of moving his office from building to building on an annual basis, mingling with the masses in an effort to break down organizational barriers and inspire Innovation – something Wim believes is the essential fuel of a successful high-tech company. When we interviewed Wim for a feature article back in 2004, he explained his philosophy for encouraging risk-taking to foster a corporate culture of Innovation.
Chatting with Wim today, on his last official day as CEO, he is more introspective. “Xilinx is renowned for the way we treat people and the way that people treat themselves and work together,” Roelandts explains. “That is something I’m very proud of, not just because it’s created a pretty unique environment, but I believe that the whole management culture of companies has to change. Most of our management principles started at the beginning of the 1900s with the industrial revolution. Peter Drucker is still quoted today even though most of his work was done in the 1930s. Frederic Herzberg wrote about motivation in the 1960s– and these are our current references. The way people are managed and motivated has to change for a lot of reasons. First, people are much better educated today – the work is not just labor like it used to be – it’s mental labor, and you cannot order somebody to think, you cannot order somebody to be creative. You have to create an environment where people can be creative. That’s one element of the culture that allows creativity at the management level and not just at the technology level. Here at Xilinx, I’ve tried to experiment with what this type of culture would be – and the result is something that I’m very proud of. The ideas that we pioneered here have really created a different environment where people are happier, work harder, and are more creative. This kind of culture has to come from the top down, however. You can’t create a culture if the CEO doesn’t practice it.”
In the office next door, Moshe Gavrielov has already observed these unique qualities of the Xilinx culture. “I’ve met a lot of people already at Xilinx, and I’ll be traveling extensively for the next few months meeting many more,” Gavrielov explains. “What I’ve noticed is that employees align themselves strongly with the company, and everybody acts like the CEO or chairman of the board in terms of having an opinion. It’s a good situation to inherit – to have all of these people who really have a very educated view on where the company should go and what we should do. The company is going through a verticalization – a transition from being solely a horizontal technology supplier to being a vertical solutions supplier as well. The openness the employees have expressed in going through this transition is a very positive reflection of the culture.”
Wim explains his emphasis on developing a culture of innovation in terms even a bean-counter could understand. “To be a successful technology company, you have to maximize your return on human capital. One of our key strategies for this is to train management. Most of our training is given by managers in the company because consultants don’t know the company, the culture. One of the courses I give myself here is on management philosophy and how this whole culture works. It starts with the belief that if you want to be successful, you have to be an innovator. If you don’t innovate, you end up in a market where the business goes to the lowest cost provider, and you’ve become a commodity.”
Wim continues, “My definition of innovation is to use creativity to solve customer needs. One of the critical elements in there is return on human capital. If people are excited and motivated and passionate, your return is two- three-hundred percent better. People even at home continue to think and wake up at night at three o’clock with an idea and have to write it down. Most companies are really operating on the lower level where people punch in and work eight hours and go home. It’s the tragedy of the fact that high-tech can be dominated by financial people who think only about return on money and not about return on human capital – which is much more important. When I first came here, the only IP we had was programmable logic. Today, we have knowledge in DSP, embedded processing, gigabit transceivers, memories – very few companies have that breadth of know-how. We just passed 1500 patents for the company – we’re now something like number six in semiconductors and number 76 or 78 of all US companies in the number of patents created.”
Roelandts is excited about the company’s prospects with Gavrielov as his successor. “It’s always a risk when there’s new leadership. Someone could come in and change the culture or change the strategy. Moshe has a very similar philosophy to mine. His background in both hardware and software is a very good fit for us. I am very confident he can steer the company to the next level.”
Gavrielov has his own observations on how his experience has prepared him for the top job at Xilinx.
“I have thirty years experience,” Gavirelov continues. “The first ten years I was doing high-performance microprocessor design. Then I moved to LSI Logic where I had numerous roles, including bringing the MIPS microprocessor core into LSI, managing ASIC engineering, and ultimately all of product engineering and marketing. When I decided to leave, it was becoming more and more apparent to me that the bottleneck in getting designs out was verification. I joined Verisity as CEO. We defined a totally new market for automation of functional verification, and the company grew from something like $4M yr when I started to over $80M when Cadence acquired us in 2005. We also went public in 2001 in the middle of the meltdown. We were growing pretty rapidly, even though most of our customers were suffering hard times. It highlights that if you can identify a key problem, you can expand even in the darkest days.”
“This role at Xilinx is a dream position for me because my background has 3 elements. First, I have my experience in the design of high-performance microprocessors, which is a stepping stone in the direction of system design. Second comes my LSI experience in making the transition from the horizontal, vanilla gate array business to the IP-oriented vertically-driven standard cell market. There are a lot of parallels between what programmable logic is going through now and what ASIC was going through in the mid-90s. It’s never identical of course, but the similarity is striking. Finally, I have ten years in the EDA industry, and EDA is perhaps the largest enabler for what we do. I have a customer perspective, having done design. I have a business perspective in moving a company from a horizontal chip business to a vertical solutions business, and most recently I have a tools perspective. When this came up I couldn’t architect or dream up a better job – it’s perfect.”
Roelandts agrees on the impending verticalization of programmable logic. “We now have to reach into the systems business. You have to take a bunch of engineers that know how to build chips and get them to start thinking in terms of solutions. Like any change, you have an S curve, and we are somewhere in the middle of that curve. With a system-level engagement, every system is unique and the challenge is support. You have to build a customization arm like we have with Xilinx Design Services that helps the customer meet their needs.”
“We are also engaging with customers much earlier in the process,” Roelandts continues. “You sell to the VP of engineering now. Purchasing doesn’t play as big a role, because your solution is unique. You have what we call “sticky sockets” so that when customers have used your solution once, they bring it back for the next one and the next one. Now, however, the design to volume cycle is stretched out. You’re designed in earlier so it can stretch out to three or four years from prototype to production.”
Laura Domela, our portrait photographer, works her magic on both executives. Wim is relaxed and comfortable in front of the camera. He’s done this for decades. His direct gaze and slightly crooked smile project a warm confidence born of years at the helm, steering a successful ship. Gavrielov is more energetic during his photo shoot. His face reflects an impatience to get on with business.
Gavrielov’s wrist is in a brace and I ask about the connection between that and the Pele jersey awaiting its place of honor on his new office wall. “Soccer is a passion,” Moshe beams. “I play, and I have the injuries to show for it. I just fractured my wrist and dislocated my finger two weeks ago. Oh, and bruised my ribs and got a black eye. The black eye is gone now.”
“…so you play pretty seriously, then?”
“No, I think that may have been the problem. I was having a lot of fun and it’s a 25-and-over league. Someone who’s a lot closer to 25 than I am upended me and I landed like a pile of bricks. Perhaps it fed into my wife’s desire for me to behave more like an adult, but this is my passion, I just can’t help it.”
This pursuit of passion is what has driven both these executives to succeed in their respective roles. “I’ve been interested in electronics since I was thirteen,” Roelandts observes. “I still build my own PCs, still, just for the fun of doing it. It really is a labor of love because it takes several days of my time, and that makes for a very expensive PC.”
“One of my other passions, aside from technology, is history,” Roelandts continues. “Over my years in high-tech, I’ve done several million miles of flying. In all that travel, I’ve never seen much more than conference rooms and airports, so that’s one of the things I’ve decided to change. Instead of flying over, giving my speech and flying back, I’m going to vacation a bit – exploring the places where so much history has happened. I’d also like to go to places where there isn’t really a lot of technology business for me like Egypt. Egypt is very high on my list. Even as the CEO of a company, there are a lot of lessons we can learn about what happened in the past and particularly anything to do with the behavior of people. Technology is new, but people – their motivations and fears – the reasons why they do things – nothing has changed much about that in 10,000 years of civilized history.”
Both the past and future leaders of Xilinx agree that innovation in new and emerging markets is key to the success of programmable logic companies like Xilinx. “Programmable Logic seven years ago was overly dependent on communications,” Roelandts explains. “My feel is that we’re slowly getting out of that. If you look at the bubble of 1929, it took over 10 years and a world war to get over it. The .com bubble was much smaller because it was isolated mainly to one sector. Programmable logic really got hurt because we were so dependent on communications. Communications is going to accelerate again, though. With the increase in video and VOIP coming over the internet, I read that by 2011 the current infrastructure will be saturated. That means there will have to be a new wave of investment in infrastructure. We introduced Virtex and Spartan back in 1998, so we’ve been diversifying for ten years now. When the .com bubble collapsed, we were already started into these new industries and markets. In 2000, 80% of our business was communications related. Today it’s about 44%, so the 8% that was outside of communications and computing is now almost 50% of our business. In other segments like military, automotive, medical, and broadcast – the PLD market in those has been growing at 20% or so, and the shrinkage in the communications market is the only thing that’s held our growth back to the 10% range. With communications growing once again, and with all the new markets programmable logic is now playing in, it should be a good time for programmable logic. ”
Moshe agrees with the prognostications of his predecessor. “Even though we’ve had a great quarter, there’s the question of ‘When is the big move?’ I call it the programmable imperative. Programmable is the way to go for system design as ASIC design becomes more expensive and risky. I’ve been on the outside for a long time and have been so jealous of what the programmable companies have to offer. With Xilinx being the largest of these, we have the biggest footprint and the most opportunity to effect the move of programmable logic into new vertical markets. From my perspective, what we need to do is to provide system-on-chip solutions that target specific markets, offering a broader and more comprehensive solution for the areas we target. We also have to be more deliberate in which markets we pursue. If you look, there are more and more that need system-on-a-chip solutions. Industrial, military, and even a lot of more consumer-oriented applications need system-on-chip integration but can’t afford ASIC risk, complexity, and time-to-market. By the time you spend two years getting your ASIC done, the standard has changed.”
Tools, power, and the next process node are among the biggest technical challenges faced by programmable logic companies. Gavrielov elaborates on tools and methodology – “At the basis, tools are an enabler, and methodology is the most important thing. Xilinx has the opportunity for more control of the methodology than the ASIC vendors had, and that puts us in a great position to help drive solutions for our customers because there are more things I believe we can control. I can’t overstate the importance of tools. They’re a key part of a methodology that lets us help our customers, and it’s a much better situation than the “best of breed” model that EDA for ASIC design has followed for the past several years with the considerable integration costs that the customer has to pay. Even if ASIC customers go with one EDA vendor, they don’t necessarily get a comprehensive solution because many of those tools were acquired from different startups along the way. It’s very difficult in the ASIC customer base with tool costs and mask costs. I truly believe we in the programmable world have a tremendous wind in our sails here – I’ve been in the ASIC world and the EDA world, and the basic hand of cards we have is much better.”
Roelandts adds his explanation of the challenges of complexity over the next couple of semiconductor process nodes. “From a technology point of view, another key issue is the increasing complexity and expense of chip design. Very few companies will be able to do that alone. That’s why you see these industry consortia happening and so many fewer companies able to embark on 45nm and 32nm designs. Our challenge is to work with our partners to stay ahead of that curve. 45nm we have under control, but 32 will be a very expensive node for anyone. We are very well prepared, but gut-wrenching decisions and tradeoffs will undoubtedly eventually have to be made.”
What will Roelandts do after this auspicious retirement?
“I’m certainly not fading away or leaving technology. I still will be on the Xilinx board, advising the new CEO. I plan to be involved for many years to come. I’ve had 12 years of history and experiences here including both good things and bad things, and it takes a long time to transfer that. With 41 years in the high tech industry, technology is in my blood.”
“So, you’re not going fishing then?”
“No, I’m not going fishing, I don’t like fishing – or sitting on the beach. To me, my idea of hell is sitting on the beach with nothing to do. I always have a book with me that I’m reading. My wife and I are also passionate about charity. I’m encouraging her to move into the limelight and to do more of that, and I can be in the background helping her for a change. She’s interested in the AIDS Foundation of Silicon Valley. Xilinx has a legacy of supporting AIDS research. She’s also co-hosting a TV program where they feature smaller, unknown charities to give them exposure to raise funds. She interviews the founder or the manager and tries to get more visibility for the charity.”
Moshe Gavrielov summarizes his less-than-modest goals at the helm of Xilinx – “Ultimately, there will be a ten-billion dollar systems solution company that is driven by a programmable infrastructure. My goal is for that company to be Xilinx.”
