Tea0 Setting up a semiconductor fabrication plant in India is not mere arrogance. There is a growing market. There are also strategic reasons: dependence on imported semiconductors makes India more vulnerable to coercion. That’s why the 2022 semiconductor mission of the government is commendable. But even today there is uncertainty about whether the FAB will happen in India or not. In this context, it is important to understand why earlier attempts failed and to examine alternative approaches.
earlier attempts
The first serious attempt in the form of a Special Incentive Package (SIP) was made in 2007, but found no response. The second attempt in 2012 in the form of a modified SIP fared better. After two years of extensive outreach with practically all major FAB companies in the world, India moved closer to being FAB. The two associations were approved by the Cabinet with an attractive set of incentives. Jaiprakash Associates formed one in partnership with IBM and Israeli company Tower Jazz, while the other was led by Hindustan Semiconductor Manufacturing Corporation with ST Microelectronics. The two fabs together involved an investment of $10 billion, and the government offered nearly $5 billion in incentives in the form of cash and tax cuts. The locations for the FAB have been finalized and land has been allotted. But ultimately both failed to raise the resources.
Semiconductor manufacturing represents the final frontier of human technological advancement. The frontier is moving forward following Moore’s law that the number of transistors in a unit area doubles every 18 months. But progress in miniaturization comes with high complexity and cost. As a result, the industry has seen a decline in the number of participants.
China started late in the semiconductor fab industry. But over the past two decades backed largely by government financial aid, it acquired hundreds of loss-making fabs from around the world and built its own fab industry. Aided by low manufacturing costs and a large-scale electronics manufacturing industry, China’s chip production has grown rapidly. By the time the US, the traditional leader of the game, found out, China had become one of the major producers of the chips. Aided by its market dominance in the rare earths needed for chip production, it has strategic dominance over chip making. Over the past year, the US and its Western allies have blocked the transfer of the latest FAB-related technology to China. But it may have been a case of the stable doors being closed after the horses were bolted. In an effort to bring semiconductor manufacturing back to the country, the US created the CHIPS and SCIENCE Act with subsidies of about $40 billion through 2022. The European Union approved €7.4 billion for a new fab in France. India will have to contend with these countries in an intense chip war.
Investing in semiconductor fab is the riskiest. Billions of dollars in recovery are needed before the technology becomes obsolete. This economic viability requires production in sufficient quantities, often reaching levels sufficient to satisfy global demand. So it’s hard to imagine a fab that is based solely on the domestic market. Semiconductors have the advantage of a small freight-to-price ratio and a zero-customs duty regime under the Information Technology Agreement, 1996, facilitating production in one location and global sales. This is the reason why no company is taking interest in setting up greenfield fabs.
Developing an ecosystem for chip manufacturing in a greenfield location is a major challenge. Chip manufacturing requires hundreds of chemicals and gases, requires people to be trained, and provides clean water in abundance. But above all this is the art of chip making. Despite the best equipment, poor quality and low yields can cause fabs to fail.
There are other issues, such as whether to install a logic/processor, memory or analog fab. An electronic device and its functionalities are characterized by their logic chips, which are strategically important and generate highest profit. To create them requires the most advanced set of technologies. Analog chips are essential, but have minimal strategic value. Memory fabs use the most advanced feature nodes, while analog fabs can be as large as 130 nm. Logic fabs are the most expensive and analog fabs are the least. A relatively easy option is Assembly, Testing, Packaging and Marking (ATMP), so that the fab ecosystem can be fully developed before the fab is installed. But ATMP has little value in terms of making the actual chip.
lessons from china
India’s strategy has been to set up a new logic fab. China, which acquired loss-making fabs and then set up its own logic fabs, offers lessons. There are many advantages to acquiring existing fabs: they are reasonably priced, have stable technology, a supply chain ecosystem, an established product line and market. They will enable India to build a great ecosystem and train human resources. Very little subsidy would be required, and the money saved could be used for advanced R&D in fab technologies that would help create state-of-the-art fabs in the next few years. Another strategy could be setting up an ATMP. Tesolv, now acquired by Tata, set up the ATMP in 2013-14. This ATMP is successfully packaging chips up to 7 nm feature size. There are over 100 ATMPs in China.
China started the fab journey about 20 years ago. As the Chinese saying goes, the best time to plant a tree was 20 years ago, but the second best time is now.