Do we need a Vitamin or Painkiller?
Date: Friday , February 02, 2007
There seems to be some heat wave making its way through the Indian semiconductor space on whether it is a wise decision to start a fab in India. On the upfront, it may sound like a very decent idea but one should look beyond the surface level and dwell deep into its actual value proposition. Personally, I feel that setting up a fab would be a good thing for India, but it should be done very thoughtfully. Fab is not equivalent to a start up where one can invest five-ten million dollars and then let go when it doesn’t work out. We are talking about a billion-dollar state-of-the-art enterprise, something that would need more than a second thought.
Venturing from the chip design world into fab will not be an easy journey. As a country we have been excelling in designing chips, slowly scaling up the ladder to execute complex design architectures. The sudden hue and cry about fab has defocused our vision from the design arena, which has witnessed a good amount of achievement. There is enormous talent available for chip designing. We have to focus on leveraging this talent pool to get to the next level of chip designing—conceptualizing the design architecture, the most difficult part in chip making. In my sense the industry leaders are tackling this with ease.
So would it be wise to deviate our focus from what we already are good at on to something that may not be our cup of tea? In other words, before we start to envision a fab in the country, we should ask ourselves: Are we well equipped to create one?
One option of creating a fab in the country would be to allow companies like AMD and Intel come in and set up their fabs here. This would in turn create an eco system of fabs, something not present in India. It is not only about money, land and water. Even if we solve all of those problems, we don’t have enough people and enough semiconductor experts here. Building a fab is like dealing with a very unique technology, which calls for people with five to ten years of experience in this field. They would possess the knowhow of the concerned area.
The reason why we are doing well in chip design today is because of companies like Intel, TI, et al. They came into the country and started designing and thus created an eco system. If those companies are encouraged to start a fab in India, then ten years from now we would have thriving eco system. At least you would have the talent. Right now, we don’t have even that.
Also, it would also be a smaller risk as people who come in here would buy their own products and we wont have to worry about where the markets are and what the pricing and profitability would be.
The second option to build a fab would be to get the current yield capacity sold to three or four vendors. Take Qualcomm for instance, if they feel they want to move their operations from China to India then they should partner the process. They should help build the fab, make investments and also use the current yield or x amount of product from the fab. They should also partner with similar companies and have an understanding on purchasing the product. This way, at least when we start building it, we don’t have to worry about who we are going to sell them to.
Interestingly, we have made no head start in both these directions. So why are we bent upon creating a fab in the country? So to understand the value propositions we can offer, we should answer two basic questions.
- How are we going to be different?
- Who are going to be our customers?
Entering the field 30 years later, it is highly necessary that we can surpass the available competition. There are dozens of fabs in Taiwan and China is opening up new ones. The size of the wafer has gone from eight to 12 inch, which is about four times increment in the number of die locations. Somebody has to do some real analysis and say if there is excess capacity at the same time Moore’s Law is not proceeding at the same rate as it was, then what are we trying to build?
India needs to realize that even though stagnation could be a possibility, we still have about three generations to cover. Process technology is changing rapidly. We have moved from 130 to 95, 64, 45 30 nm technology. Each time the process technology changes, we have to put in few billions and stay up with the processes. In this din of fabs, we have to find our unique selling point and what is going to differentiate us from the rest. Do we have the money? Are we ready to innovate or is this going to be another white elephant?
It is not so much about building the fab but selling the output. Once you build the fab you need to worry about who is going to buy the products from you. A Frost & Sullivan report says that by 2010, the consumption in the local market for semiconductors would amount to $200 billion, but the report however does not say that this local consumption would be through local manufacturing markets or otherwise. We have to ask why will OEMs buy from Indian fabs? OEMs will go to geographies that offer them least expensive chips. They are not going to buy the chip from India just because they sell the product in India.
To elucidate, it would be like creating a vitamin tablet and a painkiller. A vitamin tablet has value addition but that can be avoided despite the need, whereas a painkiller has a value addition that cannot be avoided once the need is there. The solution you produce from having the fab in India should be a painkiller. We are not a country that can afford five billion dollars just to experiment. So we have to be extra careful and ensure that we truly have understood the cause and the failures.
If your solution is bound to be the painkiller, you should also be able to answer whose pain are you solving? India’s pain? U.S’s pain? China’s pain? We have to answer these set of questions before we take the next step. Answers would clear everything up. It is not that complex, it is in fact very simple.
The author is Executive Managing Director of NEA-IndoUS Ventures.