“The only way to be really competitive is to be innovative.”

Xilinx Chairman of the Board, Willem “Wim” Roelandts, recently spoke with EDN Asia on the challenges the company faces as it moves toward the 45nm node. He also offered insights into applications which were earlier using ASICs but have now moved to FPGAs. Excerpts:


EDNA: In the last one year, what are the major technological changes that have come over the FPGA industry and how has Xilinx addressed these challenges?

Roelandts: The biggest change was the introduction of the 65nm products. We were the first to introduce these products, and we are 15 months ahead of the competition, so that was significant. The product has been very well received in the market and is ramping up very rapidly. There are other things we introduced in the Spartan range family such as the Spartan-3ADSP.


EDNA: Please specify some applications that were using ASICs but are now using FPGAs?

Roelandts: There are two ways of looking at this. If you look at low volume production such as network infrastructures, that is now probably 100 percent FPGAs. As volumes are too low, the cost of doing that with ASICS is too high, so these customers have moved to FPGAs as their major area. Where we have made most progress is in the consumer space. Not so long ago, Xilinx had no sales in consumer electronics. We arenow seeing really big orders from consumer products. In 2006, our biggest order was eight million units for FPGAs in HDTV sets. This was our highest order in terms of number of units. So we really are starting to displace ASICs even in the consumer space.


EDNA: What problems are your designers facing as you move toward 45nm?

Roelandts: There are a lot of challenges. From a process point of view, at 45nm, we will have to use the immersion scanners. That means the wafer, the lens; everything will be on the roller. To reduce the diffraction, that’s probably the biggest innovation in the first generation of 45nm.
The other thing that is coming, although at the end of the 45 or 32nm, will be the high-k metal gate technology that Intel has talked about. There are a lot of process changes that are happening in the 45nm, so it is going to be a difficult node, I believe.


EDNA: When you were CEO and President of Xilinx, what were the biggest problems you faced, and what were the most important decisions you made?

Roelandts: The biggest problem that I faced was the recession of 2000-2001. It was probably the worst recession in the history of the semiconductor industry. We had to be innovative from the management point of view. While more companies were doing lay offs, we decided instead to have salary reductions and take other cost-cutting measures. This is a cyclic industry, so what goes down will eventually come up. So there’s no sense to lay off people today, and a year after, rehire them. The move to reduce salaries was very well received, and
really helped everybody. We were able to, after eight months, restore salaries to previous levels.
Probably the most important decision I took very early when I joined Xilinx was really to focus the company on innovation. My rationale was that if you are not an innovator, you are going to have to compete on price, and that is always going to be very difficult. The only way to be really competitive is to be innovative.

You can reach Stephen Las Marias at stephen.lasmarias@rbi-asia.com