Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/india-signs-deal-with-belgium-based-imec-to-get-chip-making-technology.16933/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

India signs deal with Belgium-based IMEC to get chip-making technology

Daniel Nenni

Admin
Staff member

Cutting-edge tech from Belgium-based IMEC will be transferred to the company that sets up a semiconductor plant in India​


India’s ambition to set up a semiconductor plant in the country is slowly coming to fruition. The government has signed an agreement with Belgium-based research and innovation hub, Interuniversity Microelectronics Centre (IMEC), which will provide the technology to manufacture chips of 28 nanometres and above, for which the user has to pay a royalty.

 

Cutting-edge tech from Belgium-based IMEC will be transferred to the company that sets up a semiconductor plant in India​


India’s ambition to set up a semiconductor plant in the country is slowly coming to fruition. The government has signed an agreement with Belgium-based research and innovation hub, Interuniversity Microelectronics Centre (IMEC), which will provide the technology to manufacture chips of 28 nanometres and above, for which the user has to pay a royalty.


TO READ THE FULL STORY, SUBSCRIBE NOW AT JUST 249 A MONTH.

Uhh... 249. Does anybody have a full text?
 
India’s ambition to set up a semiconductor plant in the country is slowly coming to fruition. The government has signed an agreement with Belgium-based research and innovation hub, Interuniversity Microelectronics Centre (IMEC), which will provide the technology to manufacture chips of 28 nanometres and above, for which the user has to pay a royalty.

A top government official in the ministry of IT and communications told Business Standard: “We have signed an agreement with IMEC to provide us the technology for chips of 28 nanometers and above. It is the best way to begin, as there is a large demand for chips above 28 nanometres in the country in automotive, consumer electronics and power electronics sectors.”

According to the official, one company out of the three that have applied to set up a chip plant in India will be selected. It will be given the technology and will have to pay the royalty to IMEC.

With most big chip companies like TSMC, Samsung, and Intel concentrating on markets in the US and Europe, where they are setting up foundry plants or expanding their capacity in their own countries, the government has proactively scouted for the technology as that was a key requirement for setting up a chip plant. In fact, IT minister Ashwini Vaishnaw visited IMEC’s Belgium headquarters and played an important role in inking the deal.

The official added: “We do not see any of the big global chip makers coming to India in the initial stages. They will come once an ecosystem is established. We needed to be cautious as we have had two failures earlier in trying to build a chip plant,” said the official.

He also pointed out that the incentives were also tweaked so that chips, irrespective of their nanometre size, would disburse 50 per cent of the investment cost of the plant. Earlier, the disbursal was graded between 35- 50 per cent, based on more incentives for lower nodes.

IMEC, set up in 1984 with the support of the Flemish government to focus on semiconductor research and their applications, has collaborated with the world’s top names in semiconductors, including Intel, TSMC, Samsung, Texas Instruments, and so on. Currently, 75 per cent of the organisation’s budget is funded by industry and the rest from the European Commission and the Flemish government.

The company does not design chips or manufacture wafers for commercial production, but it is the neutral ground for companies working in the hyper-competitive semiconductor industry to get the latest cutting-edge technology in chip-making.

The government is also closely looking at the ability of the applicants to find buyers for their chips. Says the official mentioned above: “In the Vedanta-Foxconn application, the Taiwanese EMS player told us that they would require about $30 billion worth of chips annually for their own electric vehicles and consumer electronics manufacturing globally. They have committed to buy from the Indian plant.”

The official added that the government is aware of the fact that it will take the new fab players some time to woo global chip design makers like Qualcomm and Mediatek to shift some part of their outsourcing to Indian fab plants.

The IMEC deal suits Vedanta-Foxconn, which has announced that it will make chips of 28 nanometres and above. The company has also signed up with the Gujarat government to set up the plant there. It has also committed to investing $7.5 billion to set up the chip plant.

The buyback clause is essential because, while the total market for chips in the country is expected to hit $64 billion by 2026, the challenge is that foundries make chips for intermediaries like chip design companies (such as Qualcomm and Mediatek), which in turn sell them to end users like mobile device players or auto companies.

Most chip design companies have already signed up long-term deals with foundries in Taiwan and other countries, aggregating global demand and hence getting a better price. So Indian chip makers need a substantial initial buyback agreement to sell their wafers.

India’s second applicant, Singapore-based IGSS, has also said it has tied up with IMEC non-exclusively for certain areas of chip technology. However, it has roped in a medium-sized integrated device manufacturer in the US, which will provide them with chip technology and also buy back 20 per cent of the production.

The third applicant, ISMC, also a consortium, has signed up with Tower Corporation of Israel for the technology, in return for which they will take a minority stake. However, Tower has been acquired by Intel. ISMC wants to manufacture analog chips from 65nm, going down to 45nm, to begin with.

Chipping in
The company that has applied to set up a chip plant will be given the technology and will have to pay the royalty to IMEC
An official pointed out that the incentives were also tweaked so that chips, irrespective of their nanometre size, would disburse 50 per cent of the investment cost of the plant
The government is also closely looking at the ability of the applicants to find buyers for their chips
An official added that the government is aware of the fact that it will take the new fab players some time to woo global chip design makers
 
He also pointed out that the incentives were also tweaked so that chips, irrespective of their nanometre size, would disburse 50 per cent of the investment cost of the plant. Earlier, the disbursal was graded between 35- 50 per cent, based on more incentives for lower nodes.

Holy Molly.

It's also curious why Foxconn, out of all players, can't buy a process by itself.

Also, Gujarat is an extremely strange place to setup a fab.
 

Cutting-edge tech from Belgium-based IMEC will be transferred to the company that sets up a semiconductor plant in India​


India’s ambition to set up a semiconductor plant in the country is slowly coming to fruition. The government has signed an agreement with Belgium-based research and innovation hub, Interuniversity Microelectronics Centre (IMEC), which will provide the technology to manufacture chips of 28 nanometres and above, for which the user has to pay a royalty.

While cool, and undoubtedly interesting, how big of a deal is this really? It is one thing to have a patent for advanced planar and potentially getting first generation finfets, but it is another issue to have an actually manufacturable process *COUGH IBM-Samsung partnership COUGH*.
 
While cool, and undoubtedly interesting, how big of a deal is this really? It is one thing to have a patent for advanced planar and potentially getting first generation finfets, but it is another issue to have an actually manufacturable process *COUGH IBM-Samsung partnership COUGH*.

I agree. If they worked with TSMC or Intel it would actually get done and would have a longer lifespan. No offense to imec but India needs a turnkey high volume manufacturing partner.
 
I agree. If they worked with TSMC or Intel it would actually get done and would have a longer lifespan. No offense to imec but India needs a turnkey high volume manufacturing partner.

Partner with a reputable and very capable research institution like IMEC is excellent. But 28nm is a decade old already. The competition from multiple existing manufacturers and the time to market is not favorable to a new player starts from scratch.
 
Partner with a reputable and very capable research institution like IMEC is excellent. But 28nm is a decade old already. The competition from multiple existing manufacturers and the time to market is not favorable to a new player starts from scratch.
I wouldn't get to worked up about 28nm. After all Samsung and TSMC got into the race in the same way. But yeah individually this would not be profitable, and like the ROK and ROC India would need to be in this for a long haul if they want to oneday be competitive in the space.
 
I agree. If they worked with TSMC or Intel it would actually get done and would have a longer lifespan. No offense to imec but India needs a turnkey high volume manufacturing partner.
Heck I wouldn't hate a joint venture with GloFo or UMC. GloFo could use the extra scale, and GloFo/UMC's technological/business expertise would meld well with the market segment that india would be entering. Those two companies might also be cheaper and easier to woo than intel or TSMC. But who knows with the money India seems like they are willing to throw around, maybe this is not a concern.

As for IMEC, I don't think they would argue with you. Their strength lies in solving issues that their members are having with next generation transistor architectures and equipment, not lean high volume manufacturing.
 
Holy Molly.

It's also curious why Foxconn, out of all players, can't buy a process by itself.

Also, Gujarat is an extremely strange place to setup a fab.
Its Modi home state.

So no surprise if its built there , the surprise would be if its in a state that is in opposition to Modi and his crew
 
28nm is a good starting point, widely adopted mature technology yet with good demand.
India will need decades, even longer than China, to try just to catch up. It's a long game, patience and readiness to burn lots of money are prerequisite to enter the race.
 
Heck I wouldn't hate a joint venture with GloFo or UMC. GloFo could use the extra scale, and GloFo/UMC's technological/business expertise would meld well with the market segment that india would be entering. Those two companies might also be cheaper and easier to woo than intel or TSMC. But who knows with the money India seems like they are willing to throw around, maybe this is not a concern.out tttt

As for IMEC, I don't think they would argue with you. Their strength lies in solving issues that their members are having with next generation transistor architectures and equipment, not lean high volume manufacturing.
If everything goes well as planned, this India 28nm fab will go into full production around 2026. It will be about 15 years after the world first 28nm volume production and 7 to 8 generations behind the leading-edge process nodes. At the same time we haven't heard any existing 28nm foundries who are retiring their 28nm capacity. And don't forget about a lot of those equipments have already fully depreciated in the accounting books. Their equipment expense advantage is hard to beat.

TSMC started when they were two to three generations behind the leading edge nodes in 1987 and it took them 20+ years to catch up. Does this India's 28nm fab have a fair chance to survive?
 
Last edited:
Its Modi home state.

So no surprise if its built there , the surprise would be if its in a state that is in opposition to Modi and his crew

BJP already have UP set as their tech hub, and spent gazillions to woo industry there from the south.

FYI GJ has 50% of labour force below standard V.
 
Last edited:
If everything goes well as planned, this India 28nm fab will go into full production around 2026. It will be about 15 years after the world first 28nm volume production and 7 to 8 generations behind the leading-edge process nodes. At the same time we haven't heard any existing 28nm foundries who are retiring their 28nm capacity. And don't forget about a lot of those equipments have already fully depreciated in the accounting books. Their equipment expense advantage is hard to beat.

TSMC started when they were two to three generations behind the leading edge nodes in 1987 and it took them 20+ years to catch up. Does this India's 28nm fab have a fair chance to survive?
Yes, but you must also consider that these nodes are orders of magnitude more complex than nodes in the 80s where you just preformed an optical shrink without any other changes. I think 28nm is a good entry point because of how basic it is, and from there they can work with more advanced structures like finfets/gaa and more advanced tools like euv and more extensive use of DUV double patterning. You must also remember that those early years at TSMC were not exactly profitable. The early days of UMC/TSMC were heavily subsided, and I don't think anyone should expect anything else from India. The bigger question is not whether this one fab is competitive, but rather does India have the stomach and wallet to be fast followers and grow after this 28nm fab is built. As for the depreciation thing, nothing is stopping this Indian fab from depreciating it's assets/building over a few years and being on a similar financial footing.
 
Yes, but you must also consider that these nodes are orders of magnitude more complex than nodes in the 80s where you just preformed an optical shrink without any other changes. I think 28nm is a good entry point because of how basic it is, and from there they can work with more advanced structures like finfets/gaa and more advanced tools like euv and more extensive use of DUV double patterning. You must also remember that those early years at TSMC were not exactly profitable. The early days of UMC/TSMC were heavily subsided, and I don't think anyone should expect anything else from India. The bigger question is not whether this one fab is competitive, but rather does India have the stomach and wallet to be fast followers and grow after this 28nm fab is built. As for the depreciation thing, nothing is stopping this Indian fab from depreciating it's assets/building over a few years and being on a similar financial footing.

The fattest pig on the market is the captive silicon on 200mm, but nobody can buy new equipment now. Second hand been emptied years ago.

A lot of legacy people will seek a service on 300mm with economics closer to 200mm, if they can port their design to newer processes.

But analog, power, high voltage, high temp, radhard, sensors will likely be stuck on 200mm for foreseeable future since nobody would develop such low volume speciality processes on 300mm,
 
Back
Top