Friday, August 31, 2007

Does India need fabs? Worth a try!

A friend asked me whether India needs a fab. My answer quick and short was no! While it would be enchanting to see India join the global "fab club" or even have Indians comment "real men own fabs" for a change, I just don't see the ecosystem -- as people like to call it -- there. Maybe, once the odd fabs come up, that would develop as well. However, it can be quite some time away.

India, as most of us know, are strong in embedded and SoC related work. We are strong in design services. We are good at playing to our strengths. We should continue to do so. Note that we are not yet a one-stop design shop, though many people seem to see it that way. This is not exactly software and services!

However, to move up the so-called semicon value chain, India needs to do high-end designs and product development. The last one is currently the problem area.

How many Indian firms are involved in product development? Can you name them? Do you have the names on your fingertips? Most importantly, are those aimed for captive consumption (within the country) or are those serving the global markets? What are the product differentiators?

Right! Let's get back to the fab business. I asked in a earlier piece that whether everyone are aware of the kind of investment that is required for a fab. Do people even have an idea how long would it take for a fab to break even?

First, the investment. The fab is not going to be a small building built on some piece of land. If it's going to be a 300mm fab, the expenses are going to be huge. Let's keep this easy. For starters, there is going to be a fixed cost for maintaining the day-to-day running of a fab. That itself is going to be huge.

Two, most of the fab work would be automated. A fab won't exactly be hiring numbers running well over thousands. Even if huge numbers were hired, do we have people in the country with experience of working in green rooms? Let's assume there are!

Next, there are several other processes involved in developing wafer out of silicon. Do we have people with that kind of experience? Let's again assume that there are. Again, the operating costs for maintaining such personnel would be quite high.

Three, let's get down to the equipment required for a fab. That's going to be really expensive. Most importantly, all of it has to be in place, running, before the fab actually goes live. Next, a fab can't survive for long if it rests on using certain technologies. It has to use all possible latest technologies. Again, getting those would be expensive.

Finally, the wafers coming out have to be world-class and the yield, high, rather, very high. Those should be able to serve multiple product needs as well -- niche and vogue. Oh yes, the fab has to serve the global market. So, do add the marketing costs as well.

How long will it take for such a fab to break-even? Maybe, three to five years. Add the fact that technologies and process geometries would have also progressed a lot by that time. Which means, all of those need to be added on to the fab.

Well, it's worth a challenge. India is betting big on semicon. Let's have those fabs along with the fabless folks. We'll know who stands where, and whether India really has the capability to move up the semiconductor value chain.