@peter.clarke: I got back from DAC two days ago and my recent learning there reinforces some of your observations on the European foundries. But the sheer volume (local market) alone is justification enough for opening multiple MEMS/older technology fabs in India.
It is interesting that you bring up Abu Dhabi model (hence Global Foundries!). The presence of a relatively better-, larger-educated workforce (& I am not looking to pick a fight with any one from Abu Dhabi!) gives India an edge that can be leveraged than just cash alone. Only time will tell what success Global Foundries will enjoy.
Since I posted my first comment, Pradeep has written more on the topic:
Chinese fabless market set to double! And India’s?
The More-than-Moore approach is ideal for the Indian market -it has the potential to effectively leverage the country's software expertise to cohabit the hardware potential.
I wondered if Pitroda's comments were to the Global Semiconductor Alliance, which has a connection to the Indian Semiconductor Association.
But obviously India will be trying all approaches. I also agree with Doc Divikar that something along the More-than-Moore direction is a more cost effective way to start.
Only a handful of chip companies have the appetite to build leading-edge digital wafer fabs and they do not seem inclined to do it in India...unless India can find a sweetener of more than $10 billion, as Abu Dhabi has done.
However, in areas such as MEMS the costs are considerably lower and groups such as IMEC have MEMS processes they are willing to export.
The bad news is that many legacy fabs around the world, particularly in Europe and the United States will ALSO be turning to More-than-Moore making for a crowded, reduced margin, landscape.
It was heartening to notice, Dr. Pitroda made an indirect reference to SEMI (he has been in touch with "global semiconductor associations") during his interaction with the Press. Meanwhile, we have identified a Member, manufactures power and signal management, logic, discrete and custom semiconductor devices, who is interested in learning more about the project.
India should first start with more basic assignments - chip packaging. Next step would be backward integration and go for fab house. But again, do not go for 22 nm and 400 mm. Start with some 45 nm and 300 mm. Gradually move to state of the art. Also, why not they involve Indian tycoons? If they are not interested, there is really no potential for fab house in India.
Long back during 2000, many design companies had initiated talks with SCL, NPL and other academic institutions to have a MEMS fab. The problem is, nothing goes beyond talks. We do not want seminars/workshops and conferences in the name of opening the fabs and at the end sit and procrastinate about advanced technologies and do nothing but publish papers. It is more than decades ever since the SCL accident, Indian government had been skeptic about the fabs. Think about courageous people in the past who built BARC and other pioneering institutes in the country just barely after our independence. They had a far better vision of getting things done than plain talk. Today, 5B is nothing to what the country spends on movies and glamor. It is unfortunate that there is very little done in facilitating hi-tech manufacturing in India. It might even solve the so called "brain-drain problem". India has too many smart engineers and can easily be on top of the world with design and fabrication. It just takes some very bold decisions to make this happen than just talking about it, cost model analysis etc etc.
Hope some miracle fall upon on the decision makers.
@peter.clarke: thanks for the report. The news release by the Gov of India (GoI) is a re-spin of an older release. A sad irony is that after these types of news releases, nothing happens!
We very well know $5B just about covers(?) expenses for setting up ONE 22nm fab, not two! Fabs for older technology nodes may be cheaper but I don't clearly what is GoI's goals are in setting up a fab!
I have argued in multiple forums that one of the best options for India is to setup MEMS fabs along with older technology node (65nm) logic/memory fabs for less than a tenth of the investment proposed. It not only offers the much-needed process experience, but it can spur other ecosystem plays like packaging & test. More over, it paves the way for 3D stacking-enabled designs where much of the 'system' is re-use of IP and the added value/innovation is in the application & software.
You can see some of my comments in the blog portal of my friend Mr. Praveen Chakraborty who has written extensively on this topic for several years now; links below.
1. ‘Long wait’ for Indian semiconductor industry?
2. Where is Indian semiconductor industry headed?
And on the news release on which your article is based on, my friend Pradeep also wrote yet another one!!
3. Yet another plan for semicon fabs in India?
Dr. MP Divakar
SCL still exists and is now part of the Indian government's Department of Space.
It as changed its name from Semiconductor Complex Ltd. to Semi-Conductor Laboratory thus keeping the same initials. It's website is www.scl.gov.in
SCL claims to have developed CMOS technologies down to 0.8-micron as well specialized technologies such as EEPROM and CCD. It supplies ASICs for high reliability and industrial applications.
What happened to SCL? How it got killed in the first place? Should have got funding to improve capabilities but was left to fend itself, ended up making energy meters!!!
CDOTs are not born everyday.
Somebody who can bring process technology with IP is needed in India. There is no dearth of manpower for fabless chip designs.
Blog Doing Math in FPGAs Tom Burke 23 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 ...