Not a 'fab' policy

The country finally has a semiconductor policy that could eventually put it on the world map of semiconductor (wafer fab) manufacturing, keeping company with countries like Taiwan, China, South Korea, Japan, Singapore and, of course, the US. There have been two views on this issue. One point of view has been that there is no need to design a special policy for this sector, incorporating large government subsidies, since chip manufacture has become a competitive and commoditised business involving huge volumes and low margins. Far better for the country to focus on the design end (where more than 125 players already exist in the country), and to leave fabrication to the people who are already in the business. But other players had argued that India will need a fabrication industry of its own, in order to create a full semiconductor ecosystem. Given the rise in the local consumption of electronics goods (the domestic market is expected to reach $363 billion by 2015, with the demand for semiconductors alone predicted to touch $43 billion, according to a Frost & Sullivan prediction), the argument has been that there would be cost advantages in manufacturing wafers here. That argument militates against any government subsidy for the industry, but the government has clearly been swayed by these arguments and by the expectation that new fab units will attract $10 billion worth of investments.

A full-fledged fab requires an investment of $3-4 billion (Rs 13,500-18,000 crore). It is said that the delay in announcing the policy caused Intel to choose Israel over India. One question now is whether the policy announcement will induce Intel to set a base in India too: The chip giant's response is it "will evaluate and respond". But the niggling question remains: Would Intel set up a new fab in India when it can produce the same chips at its other fabs, some of which are said to have excess capacity?

There is of course the issue of competitiveness. Why should any company come to India for manufacturing chips when (say) Taiwan can execute the same work at a fraction of the price? Indeed, foundry revenue has slowed the world over because of price competition from recent market entrants like China. And the gestation period for a chip-manufacturing plant is around 18 months, by which time the technology involved could change and leave a new entrant stranded. Also, Gartner Dataquest has argued there will be overcapacity and saturation in the market by 2009 (by then SemIndia plans to set up its $3 billion chip fab). SemIndia, of course, argues to the contrary, and also claims to have commitments from the likes of AMD (which may take a stake, now that the policy has been anncounced), Sandisk, Flextronics and Broadcom.