原创 Revisiting the discussion on fabs in India

The India Semiconductor Association's Vision Event in Hyderabad last week drew attention again to the hot topic of setting up a semiconductor fab in India.

A fab costs a few billion dollars to set up, give or take a few hundred million, depending upon various factors including parameters such as process nodes, equipment purchases and so on. There are several arguments for and against having a fab. Many of the arguments center on the cost and benefit analysis of setting up a fab, such as investment in necessary infrastructure to support a fab and economies of scale that apply to manufacturing. In some instances it may really have more to do with geo- political risk concerns and the strategic advantage of having a home grown fab rather than any particular cost benefit factor. It is an acknowledged fact that India's semiconductor consumption is growing by leaps and bounds. Analyst firms publish differing estimates for semiconductor consumption in India, ranging from In-stat's $1.8billion in 2005 to $2.8 billion in 2005 from Frost & Sullivan. The numbers may vary, but the trend is definitely upward.

India has established itself as a center for software excellence in application and infrastructure software as well as embedded software. Chip design, which requires a different set of skills, has been a relative newcomer with much of the growth happening in the past 2-3 years. Three years ago most of the design work being done was more mainstream and trailing edge than leading edge. That has changed substantially in the past year as more and more semiconductor companies are opening design centers in India that focus on leading edge designs. This trend is also reflected in the contracts that Services companies in India- Wipro, TCS and others are undertaking, with a rising percentage of designs coming from the 90 nm and below technology node. Given the devolution of the electronics supply chain and the semiconductor design and manufacturing chain in particular, the fabless model has worked well for the semiconductor industry, allowing for leading edge designs to be done in India while the manufacturing happens elsewhere.

Foundry decisions are usually made by the company that is outsourcing the design. Larger volume semiconductor manufacturers can obviously negotiate better rates and have some capacity guarantees. Will the local fab have to set up an attractive pricing strategy and guarantee capacity to pull in designs that would have otherwise gone to other foundries in the region?