While much of the semiconductor world is now busy catering to PCs and mobile phones, there are a plethora of emerging industrial, medical, aerospace and defence applications. They all need silicon, software and services. A small segment of the chip industry ?the programmable logic device companies?say they are ideally placed to seize the opportunity in smaller, fragmented markets. They sell integrated circuits known as field-programmable gate arrays (FPGA), which unlike application-specific integrated circuits (ASIC) or application specific standard products (ASSP), are designed to be configured by the customer, giving him more flexibility and lower costs. Xilinx, with revenues of $1.8 billion in fiscal year 2010, is one of the prominent programmable logic device companies in the world. The firm?s president and CEO Moshe Gavrielov told Goutam Das why India should not attempt semiconductor manufacturing and why FPGA may be reaching an inflection point. Excerpts:
The Indian government wants to promote chip manufacturing. Is the world searching for alternatives to China and Taiwan when semiconductor manufacturing is concerned?
India has incredible strengths; manufacturing semiconductors is not one of them. It is not one that is necessary to invest in. There are other alternative destinations?big investments are being made by global foundries. India has so many great assets; I would invest more in system design than on the manufacturing of semiconductors. This is a business opinion. But these tend to be political decisions and I don?t know how it will pan out. If it was my money, I would look at where the highest return is, including where the benefit is to the people of India. System design would raise more jobs. Fabs today don?t employ a lot of people?it is not what it used to be. They are largely automated. India has phenomenal software base, a great hardware base, tremendous visibility into end applications, a large market. Semiconductor companies used to say ?real men have fabs?. I think real countries don?t need to have fabs today.
Some of the bigger semiconductor players are currently selling to what some say are legacy applications?PCs and mobile phones. Going ahead, they would need FPGAs for growth. However, many have failed to diversify. Why are the barriers to entry so high in FPGA?
The semiconductor world is maturing now; it is becoming a world where investments need to be large and there is a merging of applications. My expectation is that the FPGA world will play a larger and larger part in more and more applications. Customers will have, overtime, an alternative either with an ASSP or with an FPGA to address these markets. If it is very ultra high volume consumer like cellphones and PCs, very likely ASSPs will continue to dominate that market. If it is small and mid-market, FPGAs will play a larger part. Will there be combination products? Yes, there will be. We are coming out with our own combination product. Could other companies?the Intels and the TIs? try to do it based on their strengths? Yes, they possibly can and they likely will. The challenge at that point is their applications expertise and knowledge. Intel understands a lot about the PC world but lacks knowledge of other applications. They also lack the service element of the business. A large part of what we do is provide the customer with a broad set of technologies?silicon, software, IP and a tremendously strong service mentality that enables them to customise their products. The barriers some of these larger companies have is that they don?t have all the technology elements; they just have the silicon. Two, they don?t have the service mentality and the infrastructure to support customisation.
The easier option for an Intel would be to acquire an FPGA player even though it could be very expensive…Regardless of what the cost is, the company culture cannot be addressed only through money. Acquisition is always an option but different cultures is a huge barrier.
How do you fancy your chances in the Indian market?
The domestic India market is still relatively small. My expectation is that there will be more markets pursued. Once those start emerging, a lot of them would have FPGAs. One of the trends is that the percentage of the middle class is growing. Local market demand will be increasing; this would generate a lot more opportunities for local companies. FPGA is a perfect vehicle to support their products. My guess is aero, surveillance, medical are particularly promising. Some industrial applications?smart grids, smart motors?are great opportunities as well. Green IT is another one that can fall into the sweet spot.