Expert Speak Digital Frontiers
Published on Jun 14, 2020
An inflection point for India’s semiconductor fab ambitions

A 5th June commentary published in Economic Times talks about the three schemes recently announced by Government of India and how they play an important role in deciding "India’s ability to manufacture for, and service, the growing digital markets, as well as shape the norms, rules, standards and topography of global physical and digital supply chains"

Similar reports in media - mostly India based - have predominantly focused on covering the details of the Performance Linked Incentive (PLI), Promotion of Electronic Components and Semiconductors (SPECS) and Electronics Manufacturing Clusters 2.0 (EMC2) schemes at a high level, quoting mostly from the government press release. Some indicate that large mobile manufacturers are already starting to apply for the incentives under the PLI scheme. With an outlay of over INR 40000 crore, the PLI scheme has clearly come out as the creme la creme and governments push is understandable for multiple reasons. One, since the scheme incentivises production over and above the 2019-20 base year numbers, companies who already have Assembly, Testing, Marking and Packaging  (ATMP) units in India can leverage it immediately to make in india. Second, collateral results, such as job creation, will follow immediately, which is a need of the hour to boost confidence in the wake of Covid-19. The "flee China" sentiment is the third reason, although the government has repeatedly conveyed that India's efforts for self-reliance do not single out any nation. In fact, some of the players seeking to benefit from the PLI scheme are China based. In short, the PLI and related schemes will certainly do a lot of good for India in the next two to three years. However, the manufacturing of chips and components going into these electronics products is far from taking off in India, a fact that takes the shine off from India's efforts at being “Atma Nirbhar” in electronics.

Before studying India's chances of moving up the supply chain — that is to be able to manufacture the electronic components and chips, and eventually even the highly sophisticated machinery needed for such a semiconductor fab— it will be worthwhile to take a quick look how this sector is evolving globally in the post Covid-19 scenario. Taiwan based TSMC, the largest pure-play foundry with over 50% market share in the overall foundry market, is now planning $12 billion initiatives in the US. There is news that the second largest foundry player, GlobalFoundries, pulled out of the fab initiative in Chengdu China, construction for which had already started. Samsung, which is an Integrated Device Manufacturer, that is to say it runs foundries as well as makes its own products, plans to invest more heavily in its foundry business in South Korea.  Though Covid-19 disrupted the overall supply chain in the semiconductor industry, initial indications are that apart from some relatively smaller players, who were perhaps stressed even pre-Covid, most large firms are showing strong signs of recovery and expansion. This trend has lessons for India. For instance, many small startups who are in a incubation (per-launch) stage, with their semiconductor devices being "fabbed" in IISc's R & D labs, are now likely set behind by three to four months due to COVID-19. If there were  highly automated commercial fabs in India, which also support prototyping efforts of India's smaller yet futuristic startups and the academic efforts, it would have been a different story.

This is the larger context in which two schemes launched by the government, namely SPECS and EMC2 assume significance. This Swarajya magazine article looks at the big picture in terms of how those schemes could potentially help trigger a $1-2 billion fab in India to begin with, and this one gives a high level idea of some possible players. One week after the announcement of these schemes, reactions have been mixed. While many have welcomed the move after decades of inaction, others argue that further reforms are an imperative. Some examples of potential special incentives include low tax rates for fabs, long term sovereign guarantees, introducing duty on the parts that are currently being imported by the ATMP units, and broadening conditions for Capex reimbursement. Perhaps there are lessons learned from what seems to have happened to an 'irritant' clause in PLI - let a big player or a consortium of smaller players come with a solid ask of the government -  and there is a high likelihood that it will be considered.

While many have welcomed the move after decades of inaction, others argue that further reforms are an imperative. Some examples of potential special incentives include low tax rates for fabs, long term sovereign guarantees, introducing duty on the parts that are currently being imported by the ATMP units, and broadening conditions for Capex reimbursement

Meanwhile, what the scheme and articles have not covered is the market ecosystem if a fab were to indeed take off in India, which the rest of this article will attempt to highlight. This is not an exhaustive attempt, but merely a means to trigger debates into how evidence based policy can bolster this sector. This attempt at building a market story will first explore a few start-ups and then explore the possibility of larger manufacturers.  Let us take three types of electronics related Indian startups as examples. The ones mentioned in the IISc example earlier are typically what one may call "deep technology" related ones. Ideas which are in the research and prototyping stage applied in an effort to make cost effective products to replace existing solutions. One of the challenges here is that the wafer or key materials used may not be Silicon. More often than not, they are advanced materials like Carbon Nano Tube or 'III-V semiconductors' like Gallium Nitride or similar.

Though some equipment and process steps needed in making chips out of wafers of Silicon vs those out of a III-V materials will be the same, many will be different. That will raise the question of how flexible should fabs in India be? It is not easy to diversify a fab to cater too many small and different needs, but some common minimum ground could be established. The second example of startups is the design houses, which have benefited from the governments push to “design in India”. Their 'end product' is typically a physical design (PD), which is then "taped-out" to a semiconductor fab outside India, where it further undergoes some modifications for 'printability' and ends up on photomasks used to fabricate the actual chip. A fab in India could be initially considered as an additional or backup source of getting the chip fabricated and if the quality looks good, may eventually become the primary source itself. The benefits could be many - cost saving, dedicated and faster test or production lines, better collaboration and so on. It should also be mentioned that apart from startups, many big players in the semiconductor industry, be it IDMs or fabless companies, have a strong presence of design engineers in India. So its not just startups, but many of the well established companies with a strong design team presence in India may also consider a fab in India as a second source for the same reasons.

A fab in India could be initially considered as an additional or backup source of getting the chip fabricated and if the quality looks good, may eventually become the primary source itself. The benefits could be many - cost saving, dedicated and faster test or production lines, better collaboration and so on

The third type of start ups are essentially the "assembling" types - buy chips and parts, put them together (many times with code to make one compatible with another) and eventually to make it work as a system and then sell the final assembled product. With a sound understanding of how the parts they bought work, or with help from the already established design house ecosystem in India, they could get into designing and eventually fabricating the parts in India . Over and above these startups, there are also academic level tape-outs that happen, whose volume is currently small. But if a system can be built, this could perhaps change the scenario at a fundamental level, including in the realm of semiconductor research, which in the long-run can help demonstrate break through ideas that could set standards too for the fab world . Without getting into details, what India may need is a MOSIS  type setting where small chips, of let us say 2mm X 2mm or similar sizes of multiple designers or companies could be fabricated simultaneously in an approximately 25 X 30 mm standard size of a die (limited typically by the exposure field of a Step-and-Scan tool), and multiples of such dies can be made on a 300mm diameter wafer, in this case know as Multi Project Wafer (MPW). Sharing space on a die will help reduce cost for the individual customer. Indeed, only chips that have similar "upward levels" and can undergo more or less the same processes can be put together in an MPW. In all the scenarios mentioned above - be it startups or academia or some established fabless company, the end user of some of these products is the defence sector of India, and with increasing security risk, it may be prudent on the government's part to eventually go for a Trusted Foundry model similar to US national security arrangements, assuming of course that a fab can take off in India. All this, however, may still add up to only a small percentage of what a $USD 1-2 billion fab will be capable of producing. Since the Capex for fabs are huge, equipment sitting idle means huge operational loss, and so making sure that there is always enough to produce is a key element in ensuring success, which means there needs clarity and strategies on the bigger and global pieces of the market puzzle.

Among the big or medium pure-play foundries or IDMs too, the reasons for investing in India will vary. The big billion dollar investments planned in US, South Korea and so on as mentioned in some of the news articles quoted earlier are focused on sub-10nm technology. While these companies focus on cutting edge technologies in geographies that have  well established ecosystems for such production, they could perhaps consider moving their relatively older tech-nodes or their variants in the digital chip space to India. Memory, hardware supporting display, display in itself, authentication and security, encryption standards and so on may open up large markets for a fab in India. There is an even more interesting possibility in the Analog or the Radio Frequency (RF)  world with the impending 5G revolution, and that wont be limited to mobiles. To quote an engineer working in the field about the impact of 5G just in one set of products, "Massive Multiple Input Multiple Output is one of the key underlying technologies in the global 5G standard, which in turn demands a significantly higher number of antennas and RF front-end modules. Proliferation of RF bands with 5G will also increase the demand for such RF components both in the handset and the base station. Many network carriers like Viettel in Vietnam and Rakuten in Japan are working towards their own in-house solutions for 5G base station and cutting their dependence on global Telecom majors like Nokia, Ericsson and Huawei. Owning a piece of both the base station solution and the semiconductor fab ecosystem can open up a lot of avenues for synergy and innovation" .  The message cannot be clearer to anyone interested in playing a part in the 5G revolution in India. Add to all this the security risks associated with 5G, explained in detail in this ORF study, it is increasingly clear that more the pieces of the supply chain in India, not just better but safer too for India. Big portions of the analog chips which are going to be needed in manifold numbers while 5G becomes the norm,  are still in the 180nm or 130nm process technology and apart from some of them, are not likely to move into the "below 45nm" regime in the near future. At those tech-nodes, the fab is not as costly as the sub-10nm ones, a $1-1.5bn sounds possible.

Big portions of the analog chips which are going to be needed in manifold numbers while 5G becomes the norm,  are still in the 180nm or 130nm process technology and apart from some of them, are not likely to move into the "below 45nm" regime in the near future. At those tech-nodes, the fab is not as costly as the sub-10nm ones, a $1-1.5bn sounds possible.

INR 3285 crores is the current outlay for SPECS; there are no upper limits or distinction on how much Capex reimbursement a "minimum 1 crore investment" category can apply for versus a "minimum 1000 crore investment" category can. Whether many small players, large manufacturers or consortiums claim the benefit of these incentives remains to be seen.   If the latter succeeds and gets even one working, reliable chip fabbed in India within the next few years, it could be a defining moment for India’s electronics ambitions.

The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.

Contributor

Arun Mampazhy

Arun Mampazhy

Arun Mampazhy holds a BTech from IIT Madras and an MS from University of Maryland in areas related to semiconductor fabrication. He has over a ...

Read More +