REGISTER | LOGIN
Breaking News
News & Analysis

Indian Fab Project Hires Former GloFo Exec

8/3/2015 11:05 AM EDT
7 comments
NO RATINGS
More Related Links
View Comments: Newest First | Oldest First | Threaded View
deosan
User Rank
Author
Re: A wise man always learns from others' mistakes!
deosan   8/5/2015 2:21:13 PM
NO RATINGS
GSMD: Thanks. This is great. God speed!

 

GSMD
User Rank
Author
Re: A wise man always learns from others' mistakes!
GSMD   8/5/2015 12:59:16 PM
NO RATINGS
And what makes you think that the Indian govt. has not done the same analysis that you have done !

The govt. is actually investing most of its money in creating design capabilities in India and letting the private sector handle the manufacturing. It concluded a while ago that the IP accounts for 60-70% of the product cost and manufacturing is actually lower value. But  India is big enough to handle both unlike Taiwan or any of the myriad examples you have quoted. So please look at the govt. R&D expenditure before concluding that it does not know what it is doing. The past 10 years have been used to seed a large number of universities with IC design flow and IP design capabilities.

So the design eco-system is pretty strong, especially the backend (there is a reason why Qualcomm does a lot of its backend design in India). IP portfolios in compnaies are slwoly increasing but is lagging in PHY IP. Lack of local foundries is part of the problem. RF IP is also an issue. But there is now more than enough local design capability to compete with any Mobile AP SoC vendor. Low power IoT processors are also seeing fabless OEMs like Ineda.

This is also why the Govt. is switching over to the RISC-V ISA (using a locally developed family of cores) for slowly displacing ARM and x86 for locally designed products. Don't take my word for it, ask ARM ! The first OO core was released last week. So this is not vaporware.

If you need acess to the source send me a request and you can validate it for yourself. Between our effort and UCB's efforts I think the dominance of ARM will slowly be reduced. 

All plans for these products assume TSMC, UMC or GF. Nobody in the govt. is even talking about local foundries for any of these programs. If they come fine, we will use them. If not, not a big deal.

Please realize that apart from some tax breaks and subdizied resources, teh govt. is not pushing these programs,  all these foundry programs are private sector programs. If a private investor wants to risk his money, the govt. is going to encourage and incentivize such efforts. But that does not imply that the govt. thinks that design does not have value. The contrary is true.

The same goes for manufacturing. By finally having a sensible tax structure, the cost differential between imports and local assembly is 11.5%. So from close to 100% imports, mobile manufacturing for phone below USD300 is slowly becoming 100% local. Foxconn plan to open 10 facilities with about 1mill employees. Sony and Xiaomi now manufacture locally. I am not implying this is something great. But the only role played by the govt. is duty structure rationalization, not pouring money into manufacturing. The same strategy is being adopted for the fabs.

All comments also forget that there a lot of research programs in VLSI in India nd sustaining these programs requires at least one small foundy , preferably 22nm or better. Try getting detailed process libraries from TSMC and you will know what I am talking about. So for R&D and strategic reasons, a couple of small foundries are badly needed.

To conclude, there is absolutely no evidence to support the premise that manufacturing is being prioritized over design.

 

Quix0
User Rank
Author
Re: A wise man always learns from others' mistakes!
Quix0   8/5/2015 4:45:16 AM
NO RATINGS
You hit the nail right on its head! Unfortunately Indian Government is hell-bent on hitting their head on the nail :-)

Amir11
User Rank
Author
Re: A wise man always learns from others' mistakes!
Amir11   8/5/2015 1:28:15 AM
NO RATINGS
I fully agree.  So well explained.

deosan
User Rank
Author
A wise man always learns from others' mistakes!
deosan   8/4/2015 12:23:18 PM
Folks:

First it was Japan, then Europe, then the U.S. - there are so many examples in front of India that show that it is more important to facilitate innovation in design rather than pouring billions of dollars in building something that you can collaborate with someone else. 

This attempt to build a fundamental industry like this is like making your own ocean because you like seafood! Instead, collaborate with the already existing ocean and figure out a way to get more from that 'raw material'.

We have heard this same statement - "we are getting a lot of positive response from the government" - for the past 2 decades. And we will keep hearing it for the next 2 decades. 

Instead, put that money as an investment in new design companies. 

Here's simple math. If India is consuming $100M worth of network controllers and about 1/3rd of it is going to the fabs and 2/3rd of it is going to the designers of that chip. Then, invest $10M in a company that can design network controllers for Indian market. Then you will keep 2/3rd of that $100M within the Indian economy. 

Another thing is that it takes time, money, resources (water, land, electricity) and skilled workers. These things are grossly under-estimated.

So, leave the manufacturing to Chinese and Taiwanese - don't get too hung up on country pride - its business. And invest in innovation.

Or you can spend two more decades, make a few more individuals millionnaires and still not have any fab to show for! If that's what you want India - go ahead. 

GSMD
User Rank
Author
Re: 200mm or 300mm wafer facility in India
GSMD   8/3/2015 11:56:33 PM
NO RATINGS
The other fabs are pretty much history. At least that is what I hear from the folks who should know.

The problem is  financing. As this arcticle quotes the founder of cricket, the project should have something to attract investor attention. I suspect the other two were a little too fuzzy in their projections.

This one seems more realistic and the addition of analog helps. There is a market in India, there is no doubt on that but a clear strategy to adress that market is key. Power electronics is a huge market, smartcards/payment cards are equally big and  so is auto/industrial. And to an extent milgrade parts.

These sectors can be tapped without govt. help. General compute/mobile will require some mandate for local parts but in my talks with Mobile OEMs, local tablet SoCs may gain traction (AP parts not with radio).

There is no crying need for a cutting edge fab in India, 90 - 45 nm is more than sufficient for now.More important is supporting embedded flash, high power analog, RF parts (GaN is key). On a related note an LCD fab will not hurt. Demand is there, even Sony has started local manufacturing of 40+ in LED backlit TVs.

geekmaster
User Rank
Author
200mm or 300mm wafer facility in India
geekmaster   8/3/2015 2:10:29 PM
NO RATINGS
Interesting, indeed! Will this be a 200mm or 300mm wafer facility? How does this compare with the other wafer fab announcements we heard in the recent past?

Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed