India’s Semiconductor Scam: Indian Mining Tycoon and Pro-CCP Assembly Giant Swindle India’s Semi Subsidy Program
Overview of incentive program, 3 major proposals feasibility
With the politicization of the semiconductor industry reaching all-time highs due to supply chain instability and a new cold war between the US and China, governments around the world are racing to support their domestic semiconductor industries. China, Japan, the US, and the EU have all passed various levels of government support which total hundreds of billions of dollars of incentives. Semiconductors are the bedrock of the world’s economy and national security. No country can be a world power without a domestic semiconductor industry.
India, the 5th largest economy in the world, wants in on the action. India has a negligible presence in the semiconductor manufacturing ecosystem, despite its massive presence in the design ecosystem. In the 1980s, the government of India set up Semiconductor Complex Limited (SCL). SCL has a long history, but the summary is that it was only ever useful for research. Today it fabricates at most a few hundred wafers a month. Compare that to the monthly production worldwide of ~10.23 million wafers a month, and India’s market share is effectively 0%.
India has been attempting to woo a commercial semiconductor fab, either an Integrated Device Manufacturer (IDM) or pure-play foundry, as far back as the early 2000s. These have primarily resulted in failures that lead many opponents of the industrial subsidies to believe an Indian semiconductor industry is unfeasible.
In May 2014, the current Prime Minister of India, Narendra Modi, took charge. He quickly launched the “Make in India” campaign. While this campaign has seen success in some other industries, there was no progress for semiconductor manufacturing in India for his first term. His second term has been playing out differently. Nirmala Sitharaman was appointed to be the finance minister of India. In one of her first speeches about the 2019-2020 budget, the semiconductor industry was directly called out.
In order to boost economic growth and Make in India, the government will launch a scheme to invite global companies through a transparent competitive bidding to set up mega-manufacturing plants in sunrise and advanced technology areas such as Semiconductor Fabrication (FAB)
While a fab established in India could cater to the global export markets, policymakers in India focused on creating a domestic market for India as well. The prevailing belief is that “low-hanging fruit” such as low-value assembly can create many jobs and help move India up the supply chain. This is generally correct if we look at the development of various global economies. Economies often transition from assembly to packaging to fabrication and design.
The Indian government, through its Ministry for Electronics and Information Technology (MEITY), launched efforts such as the “Production Linked Incentives” (PLI) in 2020 to increase “Assemble-in-India” An article published soon after the launch of the initial set of PLI schemes summarized it thus:
The idea is that focusing on assembly units, that is, ‘last-mile industries’, would eventually help develop deep backward linkages, thus inducing industrialization. The need of the hour is to promote semiconductor manufacturing alongside assembly units in India
In December 2020, about a year and a half after the initial budgetary announcement, MEITY released “Expression of Interest for Setting up of Semiconductor FAB,” which kicked off the rush for proposals. While many discussions were happening in the background, some of those also bubbled up to credible media outlets.
Reuters reported that India would give cash incentives of more than $1 billion to each company that sets up a fab in India.
Bloomberg reported that officials of the Indian government and Taiwanese government had met to discuss a deal that would bring a $7.5 billion fab to India.
The outcome of these discussions never came to fruition. Many companies that were not involved began to bid for these subsidies. Tower Semiconductor and their Indian business partner, Next Orbit Ventures, sent a letter directly to Narendra Modi about the lack of transparency and movement in the process. Tower Semiconductor has been proposing a fab in India since 2013 and the current 65nm fab proposal since as early as 2017.
In December 2021, MEITY announced India’s Semiconductor Incentive Program. This program explicitly approved 76,000 Crore Rupees (~$10 billion) of incentives for semiconductor fabs, outsourced assembly and test (OSAT), and design-linked incentives. Initially, the amount of incentives varied based on the type of project, but eventually, the government announced that these subsidies would max out at 50% of a project’s total cost without any ceiling.
The India Semiconductor Mission (ISM) was established to evaluate the various proposals. There was an explicit pre-requisite for applicants that is noteworthy.
The applicant Companies / Consortia / Joint Ventures should own or possess production grade licensed technologies for proposed technology process and demonstrate the roadmap to advanced nodes technologies through licensing or development.
In February 2022, MEITY released information on various applicants and the categories they were pursuing. For silicon fabs, there were three applicants.
International Semiconductor Consortium (ISMC)
ISMC is led by a Mumbai-based Venture capital group, Next Orbit Venture Fund (NOVF). The group proposed a 65nm Analog fab. Initially, they considered Dholera, Gujarat, but later decided on Mysuru, just a short distance from Bangalore, the technology capital of India.
Tower Semiconductor offered to transfer the 65nm process technology required and 50% of the fab’s utilization. This is most of the way to making the project economically feasible. The total cost of the fab is ~$3 billion, with 50% of the cost coming from the Indian national government, another ~20% to 25% from the state government in the form of incentives of land and infrastructure, and the rest from debt and equity. There are also “two reputed defense Public Sector Undertakings (PSU),” Reliance Industries and Hindustan Computers.
The ISMC proposal has had negative rumors surrounding it that MEITY is not convinced with NOVF’s ability to handle such a large amount of money. This makes no sense, given that there are multiple stakeholders, including reputable public and private firms. Another potential reason for the lack of approval is that the Indian government has been waiting for Intel’s acquisition of Tower to close. The state where the facility would be located, Karnataka, also has elections coming soon.
The last speculated reason for the lack of approval is that 65nm is not seen as important enough. For some reason, there is an over-fixation on the 28nm process technology rather than what is achievable today. It’s not as advanced of a fab as some other proposals, but it is far more attainable. India cannot expect to get to advanced nodes without prior process technology experience. The Indian government highlighted the importance of “45nm and above” process technology.
The 8-10 month “deadline” for approval has passed, and there have been no official decisions on any applications. In July, Tower Semiconductor’s CEO, Russell C. Ellwanger, met with Indian ministry officials and media. He allegedly stated that the India Semiconductor Mission, ISM, lacked the necessary executive experience within the semiconductor industry, which is cited as a potential cause of delay.
MEITY says that ISM has an advisory board, but the 17-member board formed in April had just one member with commercial fab experience. As per media reports, they did not even have formal meetings for months after its formation. Two more members with fab experience were added later. The ISM now has an advisory committee of 21 members. Only three have real fab experience. A few others have design and packaging experience. Most of the members are more political appointments.
The three people with real fab experience are tied to real jobs in the industry, so they have limited time and capabilities to offer. These semiconductor experts likely have had their opinions overpowered by bureaucrats. Some speculate that a decision for this project will come in February next year.
Innovative Global Solutions and Services (IGSS)
IGSS is a Singapore-based consortium that has proposed setting up a high-tech semiconductor park. IGSS is an ex-fab manager for Global, Singaporean fabs. They have done some work on GaN on Si technologies and fab optimizations. Oddly, they did not apply to do a compound semi fab, but rather a logic and analog fab.
The investment was initially reported to be Rs 25,600 crore or about $3 billion. The proposed semiconductor park would have fabs producing wafers ranging from 28nm, 45 nm, and 65 nm and an industry ecosystem infrastructure that hosts semiconductor circuit designers, material suppliers, equipment suppliers, and OSAT players.
Later reports quoted the cost as $3.5 billion with 28nm to 55nm technology transferred from an IDM majority owned in the US. The name of this IDM has yet to be revealed, but only one US-based IDM has fabricated 28nm, Intel. They may mean GlobalFoundries or IBM, which have the technical know-how, but neither is an IDM.
Assuming that the IDM partner that IGSS claims to have, is indeed giving it “production-grade licensed technology” for 28-55nm, the feasibility of this project is more reasonable.
The last we heard about this proposal was that they were asked to find a strong Indian investor. The policies had no such prerequisite for funding, but it makes sense to require it.
Foxconn and Vedanta
A joint venture between Foxconn and Vedanta has been formed. Vedanta is an Indian multinational corporation primarily in the mining industry. This joint venture has received the most media attention of the three major proposals.
Anil Agarwal, Vedanta’s founder, chairman, and owner is incredibly well-connected politically. Foxconn is technically a Taiwanese firm, but it is also listed and regulated in Chinese financial markets, with much of its operations embedded in China. Furthermore, the firm’s leadership is deeply tied to the Chinese Communist Party.
After many months of rumors, the joint venture announced that Gujarat was chosen as the site for the fab. The Gujarat government’s press release states that the investment would be INR 1.54 lakh crores (~$21 billion), out of which INR 60,000 crore (~$7.2 billion) is for a semiconductor fab and OSAT. This ratio of spending doesn’t make much sense. In contrast, the Vedanta press release states that the initial investment would be ~$2 billion and go up to ~$20 billion. It also mentioned the 28nm process node specifically.
Most of the criticisms of this project come from the political side, clamoring that Vedanta may receive $10s of billions of incentives from the Central and State governments combined, which is more than the annual budget allocated to certain social welfare programs. Those criticisms don’t make much sense as India needs to kickstart a semiconductor industry, and these two spending categories are not mutually exclusive.
Every statement from Anil Agarwal gets quoted by the media without considering whether it makes technical and practical sense. These media appearances often have contradictions to earlier statements. For example, in an interview in August, he said, “We may not go for the 28nm chip, we may go for 45nm” and that “28nm is still in the innovation stage.”
First products should be out in the market in two years.
This claim is quite literally impossible. Even TSMC, the most well-run and highest volume foundry in the world, has only ever taken as little as ~3 years for a brand-new facility from breaking ground to production.
Foxconn’s history with government incentives is far more troublesome. For example, they had a proposed display fab in Wisconsin that never panned out. Foxconn took some incentives and never built anything close to what they claimed they would. While much of this was clawed back, it serves as a stark warning for future incentive programs. Foxconn had minimal experience in display fabs at the time.
It’s noteworthy that this proposal also faces some of the same issues. Foxconn has no experience running even a 130nm fab, let alone a 45nm or 28nm process technology. No company has ever skipped forward that many process nodes in development. The semiconductor industry consists of incremental learnings stacked on top of each other, not moonshot leaps.
Foxconn’s semiconductor experience is entirely limited to their acquisition of an ancient 6” fab running processes that are decades old. While the JV was asked a bit more about their fab experience, the speculation is that licensing an IMEC research policy was enough to handwave away these concerns.
India’s wooing of Taiwanese players with a $7.5B fab would make sense if it were TSMC, UMC, PSMC, or VIS. Those fabs have actual experience with high-volume manufacturing and closer ties to the Taiwanese government than the Chinese government. Foxconn is hardly a good partner. Foxconn has been hunting for a desperate government to fund its push into the semiconductor industry for years, from China to Saudi Arabia to Malaysia.
There has been some murmuring that this is a quid pro quo for increasing India’s share of assembly in electronics. The Indian government has recently approved another INR 357.17 crore of incentives to Foxconn for mobile assembly. India has aspirations of becoming the largest in the electronics assembly world, and convincing Foxconn to move over is a big part of this.
All approvals are in place.
Assuming Anil Agarwal isn’t lying, then India has just been duped into funding an unfeasible semiconductor fab. A well-connected Indian mining tycoon and Foxconn, a pro-CCP assembly company with no relevant semiconductor experience, would operate this fab.
The other two proposals are much more feasible for bringing an economically viable semiconductor fab to India as a launching point for domestic industry. Furthermore, they would be connected with Israel, the USA, and/or Singapore rather than India’s most significant geopolitical rival, China.
The chosen location of this fab is based primarily off of political whims. Gujarat is the epicenter of the BJP government which has run India since 2014. While we believe Gujarat is India's industrial and shipping heart of India, it is not where India’s semiconductor expertise are located. Bangalore, Chennai, Hyderabad, and Pune are all much better locations for a fab in that context.
Anil Agarwal and Foxconn have certainly greased the right politicians and bureaucrats if “all approvals are in place” for their fab.
The rest of this post are some other notes on these proposals, links, and our general opinion on the best course of action for the Indian semiconductor incentive programs. We also briefly discuss Polymatech and Tata’s OSAT proposal.