Does India really need a $5bn semiconductor unit?
By Vinod Dham
There has been for a long time, an on-again and off-again debate about whether India needs a fab. A fab is a plant for
manufacturing semiconductor chips, also popularly called
silicon chips or simply
chips. Silicon chips are ubiquitous today. They are in side all types of gadgets that we use in our daily lives: TVs, computers, cell phones, radios, cars, etc.
So clearly there is a big market for making these chips and, therefore, a case can be made that India should build its own fab to meet India's domestic needs. Chip building business is highly competitive and cyclical. It has been around for over 50 years and there are very well established and well entrenched players. The most recent chip fab, China's state funded SMIC (Semiconductor Manufacturing International Corporation founded in 2000), has failed to make a dent in the chip making industry despite 15 years of state-sponsored effort.
Building a fab is a very expensive undertaking. While Moore's law (named after Gordon Moore, Intel's co-founder and observer of this trend) has commoditized semiconductor chips resulting in a staggering reduction of roughly a factor of a billion in price over the last 50 years, the cost of manufacturing equipment for these chips has also increased exponentially over this period. To build a fab using current state-of-the-art technology costs about $5bn (Rs 30,000 crore) initially.
In addition, the fab will required to be retooled every two to three years, at an additional enormous cost, to stay competitive. The cost of building and retooling a chip plant has become so prohibitive that in USA only one company, Intel, has the wherewithal to build its own chip plant. In fact, the astronomical cost of building a fab has forced chip companies to pool together their financial resources to set up plants like "Global Foundries" for manufacturing their chips.
Even if we were to build a fab in India despite such prohibitive reoccurring financial undertaking, one of the biggest challenges will be what chips to make in them and how to keep the fab fully loaded. Let us say we decide to build chips for cell phones. The top three popular selling cell phone brands in India are Samsung, Micromax and Nokia. Since a new Indian fab will offer no competitive advantage in terms of price or performance, it will be hard to convince Samsung, Micromax or Nokia to build their chips in this fab.
This same logic prevails for building chips for the TV, personal computers, automotive and other such markets. And even if all three cell phone suppliers were to agree to build their chips in India, the volume of these chips may not be enough to keep the fab fully utilized. Running a partially filled fab is like flying a partially filled Dreamliner — a losing value proposition.
One way to, however, get around the issue of how to keep the fab fully utilized is to invite someone like Samsung or Intel set up a plant in India to meet their own global market needs, including those for their sales in India. Clearly, the Indian government will still need to guarantee uninterrupted electric supply and millions of gallons of water to run these plants besides importing very expensive highly skilled semiconductor specialists to manage the fab's highly automated and expensive equipment.
Thus, given enough subsidies and incentives, if a foreign MNC could still justify building their chips in India, it will be a good way for India to get into the business of building chips. Over a very long period of time, such a move could end up creating a critical mass of an ecosystem required to sustain such a business in the long run.
Recently, the Indian government has expressed concerns about security risks posed by the use of some foreign-made chips. India is reportedly considering a proposal to make it mandatory for the core strategic sectors of defence, space, atomic energy and cyber security to use 'made in India' chips to protect its national security needs while kick-starting the domestic semiconductor manufacturing business. Unfortunately, these sectors alone can't generate enough volume to justify a state-of-the-art fab, besides the concerns I have discussed earlier.
However, once the government has defined and designed such special chips, it may be better to build them in an older generation fab. The older fab could cost one fifth to one tenth of the state-of-the-art fab and be optimal for producing lower volume, larger variety, mixed signal chips with higher reliability resulting from use of more mature technology.
Cyber-security has become a major concern for countries and companies worldwide. Cyber-security is very essential for national security. The entire cyber-warfare infrastructure in the USA is reportedly based on information technology consisting of standard networks of low cost computing devices. The secret to addressing this concern in a practical manner for India is to strategically use software (India's strength) and firmware as its secret sauce. Moreover, India should implement a "Trusted Manufacturer's Program" for its core strategic needs. This program should be designed to certify chips, software and systems produced by foreign manufacturers and avoid buying from adversary countries.
The writer is an inventor, entrepreneur and venture capitalist
Source:-
Does India really need a $5bn semiconductor unit? - The Times of India
