ST would of course like to find new markets for their products that have become less competitive even in Europe. They have APs for Mobiles and other leading node SoCs, micro-controllers and then process for FD-SOI that could still turn out to be an alternative to FinFET.
But what does a pre-HKMG Foundry like Tower Jazz bring into the Party ( the 2 nd Fab ) ? To what extent would IBM keep supplying them with leading edge process technologies ( HKMG ) ? Would they ever do process R&D in India ?
On the other side of the Eq. the $ 8 billion per year of semiconductor import by India could be a bit of a red herring. One would suspect that much of that import bill is pre - decided by the OEM System importers ( Samsung ) and assemblers ( Nokia, local MicroMax - in bed w/ China's SpreadTurm ) in India.
But older nodes are quite adequate for new applications like ID and Smart Cards that are about to see volume application in India. So these Fabs could start even at 130 nm to keep yields high with a green crew.
Saw somewhere that both these CMOS Fabs are aiming for only 40k wafer starts per mo. To use up even that capacity India would need to develop balancing facilities ( Marketing, System Design, Assembly & Test,.. ) domestically. Otherwise these Fabs will have to turn into late node Foundries for export and we know how that goes. Won't do much to offset India's import bill for semis - the original purpose behind the generous Govt. subsidies.
Folks in India seem to be fixated on Digital Logic and are totally oblivious of Fabs for Power Electronics ( IGBTs ) & Compound semiconductors ( RF, LED, .. ) that are more relevant to an agrarian & rural economy ( not to mention that those Fabs are an order of magnitude cheaper ). Or to jump well ahead of the pack, they could focus on MEMS - especially for Medical sensors.
As usual the governement thinks it can just create magic by just starting something 20 years late and with crumbling infrastructure and high levels of beuracratic inefficiency this is surely a disaster in the making. Just goes to show how foolish the government can be when it's people are mostly illetrates.
I don't know on what basis this counts as the best way to boost economy by throwing billions in an untested and uncompetitive product called siclicon. I agree it would need one small unit just to show that it can but commercially viable, that's a pipe dream. When most companies are going fabless what makes the govt think the ones on Indian soil can churn out profits?
There are much better ways to utilize these billions to shore up the suffering manufacturing sector.
Making the infrastructure fab-friendly is a very big deal. That is why it is considered in fab locations. The cost of making the fab-friendly infrastructure "merely" needs to be deducted from the $4 billion. It could leave about $2-3 billion, which is just right for 65-90 nm maybe. So to get to the advanced nodes would need another round of funding.
There is in fact a huge local demand for lower end products. Think energy meters, zig bee controllers, smart cards, processors for appliances, automotive/industrial parts. The list goes on. These are what really create a digital economy not 16 core processors fabbed at 14 Nm. Look at the server market growth vs appliances growth and you will see where the real money is. Similarly mass market setup boxes, tablets and phones in India are lower end products and will continue to be so. 28 Nm nodes are more than sufficient.
Look the primary goal is strategic electronics and import substitution. These nodes will help achieve that goal. I agree that the oems who will use these fabs and churn out products are yet to be identifies. Non Indian Product vendors will be in the forefront initially but you will need India companies ultimately. This is the real challenge. But considering there are a lot if tier 2 and tier 3 vendors available for sale, buying out a few of these may not be a bad idea.
Yes, Prabhakar you are right - it is never too late to start something. Sure, India missed this bus almost 20 years back - there was a time when Seagate was exploring the viability of setting up a manufacturing unit ( time frame - something about 20-25 years ago) in the southern state of Kerala but it came to naught.
Sure, the govt officials have their own personal agenda and things do not move the way they should but at least there is a beginning... lets hope it takes off and if the multinational companies involved in the consortia put their foot down ( and accelerate) the process, we can have working fabs in a couple of years and two years will just fly... maybe processes would change but India being a huge market, there would still be demand for the older ones
3-4bil (US$ ?) buys logic tool capacity at 45nm that would fit in a small fab about 18, 000m2 cleanroom. That assumes zero cost of construction or infrastructure. Probably 2bil alone in older generation litho tools. although landmark, this doesn't seem to be even comparable commitment in terms of current generation fabs.
I will say it is still worth to have the fabs even though it may be late . It is never too late to start something which have a good impact in the future . India's first nuclear submarine has come into existance almost 40 years after the project was conceived but still it is relevant in the strategic terms and benefits to the nation.
Not sure if it is too late for this to happen, if we are talking about a timeline in terms of "a couple of decades" to be successful. This should have happenned a long time ago. I read somewhere about the semiconductor import bill was $8.2 Billion in 2012 and it was predicted by the Department of Electronics and Information Technology to grow by 20% every year. What those guys had been doing since last five years...did they not foresee this issue then? I had been hearing the Fab story for more than last three years. The process is frustratingly slow and unfortunately the foresight is missing (or the willingness to foresee). I don't know where the rest of the world will be in the next two decades when the Fabs in India would hopefully reap the benefits. Atleast I will retire. :)
Genuine question! Who will buy chips? But system level company are ready to buy chips and complete ecosystem will be established. Design house are functioning with profit, assembly unit are also ready to establish. Most important is market which all know about import capacity of India. Export of chips are still opportunity with lesser cost due to relatively cheaper manufacturing cost.
Blog Doing Math in FPGAs Tom Burke 7 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...