Whatever be the current odds, this initiative by Indian Govt will definitely benefit the Indian economy in the long run.
Like the nuclear reactor facilities which are successfully operating now for the last 40 years or so, the fabs project will also be successful and the government will be prudent enough to loacte these facilities away from the traffic conjestion zones .
Who buys these chips?
If the idea is to reduce chip imports who are the systems OEMs rounding out the picture? Are they serving a significant market now?
Didn't India try this at the systems level long ago unsuccessfully?
CPUs being designed in Undia are being taped out at 45 Nm for 1ghz class designs. Probably shift to 32/28 nm next year. 32 nm shuttles are approaching the affordable price range. For our research, we are setting up a flow at these nodes, so hopefully we will have my grad. students doing test tape outs at 28nm in a couple of years. But these fabs come on line in 2 years, 45 Nm in 2016 seems a bit undewhelming. Compared to 14 nm intel processes, 45 nm seems like hand drawn lithography ! Still I will take what we get, getting tired of running to Taiwan for tape outs.
Its a big risk to set up fab companies in bangalore. The city is already struggling with infrastructure issues, almost all the roads in Bangalore have so big potholes that your cars one tire is on road and other can fall inside hole. The city already has hut the headlines in terms of mishandlingof garbage. Its called as garbage city. The city municipal authorities are still to make concrete plan and implementationof solid waste management. How the silicon waste be handled?
Problem with Indian political system or societyas whole is they hate to plan and take too much in the plate.
India is not China, where right things would be done and can be done. Just having huge population and being a big market doesn't mean you can have the semicnductor fab. It requires strict discipline and proper planning. In India everything changes for political mileage and political parties can go to any extreme fir their own benefits.
Cities like Bangalore, delhi and Mumbai are already struggling to breathe, this will be too much on plate.
I guess everyone was under the impression that only one fab would be set up in India but it sure came as a surprise that two were getting the in-principle approval. Not too sure of all the fine print -will keep you guys posted oncce the weekend is over - but with IBM and STMicro involved, things cannot go too wrong.
Sometimes, incredible things happen in India - like say the mobile revolution - the way people who cannot read a word of English land up downloading regional movies ( the language is sometimes the regional one but still to get a hang of the technology is still something I am amazed at.
And, like someone said, Nokia did a great job of setting up a greenfield project but then you got to keep in mind that it was labour intensive one at that. They were able to get set it up in the then-rural area and were able to get cheap labour ( young girls who were paid quite well for an 8-hour shift) but fab is a different story.. and yes, red tape is also an issue as well as power and water... but then, many things in India are an issue. So, its nothing new.
I guess, this time round, the consortia should have learnt not to make the mistakes that were made earlier and d\ work on this to get it operational on a war footing.
Will keep you posted about operations and capacity and the size as soon as I get it.
Emerging markets are also unpredictable when it comes to tech companies succeeding. There is a lesson to be learnt from the mobile market. For the longest time, I held the strong opinion that India mobile vendors had no chance of success and the likes of samsung, LG, Nokia were simply too far ahead to let in new entrants.
I am still amazed at the success of Micromax and cannot quite explain how they succeeded over LG for example. LG had the tech, the product range, distribution, support and brand awareness. But still lost, partly due to pricing.
Maybe the fab market is also heading towards commoditisation in the mature nodes and the dynamics may be different in that arena. I frankly have no clue as to how this will all turn out but just adding a note of caution against any forecasts ! Bundle free IP with the fab services and you could get value sensitive customers.. Add in free design services and suddenly the fab model looks a lot different. Of course commodity nodes are assumed.
Nobody is talking about the process nodes. Some of my designs will be the first ones to get fabbed in these facilities and I have no idea about the
node. I think they will start with 40 or 32 and go to the 20s.
The most volumes per part is probably microcontrollers, which are 65 Nm max, next is cortex a8 class parts for 200-800mhz grade controllers.which are ok at 40 Nm, PC , tablet , media and server grade parts need 28 Nm though.
I will be happy with 40nm to begin with though.
Blog Doing Math in FPGAs Tom Burke 14 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 ...