The government has announced subsidies totalling ₹76,000 crore (around $10 billion) to build a semiconductor ecosystem in India. Combined with matching private investments, this would amount to $30 billion in investments. Unfortunately, even this staggering amount falls short of what is needed to build a semiconductor industry, let alone a single fab. TSMC, a Taiwanese company that is the leader in advanced semiconductor manufacturing, has announced that the new fab it is building in Arizona, US, will cost $40 billion.
At the height of the Cold War era in the 1960s, Soviet premier Khrushchev built an entire town for scientists and engineers tasked with replicating semiconductor manufacturing like in Silicon Valley, US. The attempt failed, and to this date Russia isn’t a significant manufacturer of semiconductors. The lesson: replicating the entire supply chain locally is futile. Instead, we should follow the strategy of gaining control of strategic components of the supply chain, a strategy adopted by Japan, Taiwan, and Korea.
The thrust areas
In India, the policy thrust is currently on self-sufficiency (atmanirbhar) and national security. However, given the complexity of semiconductor manufacturing, no country is truly self-sufficient. The life-cycle of chip manufacturing involves many stages. A semiconductor integrated circuit (IC) is designed using Electronic Design Automation (EDA) tools. A foundry etches this design on a silicon wafer using highly complex machines.
Specialty chemicals are used to etch and deposit layers on the silicon wafer to build an integrated circuit. A modern production fab achieves nearly 100 per cent yield, i.e., the percentage of defect-free chips per wafer. A supply chain for manufacturing ICs involves companies that are often geographically apart. No country, including the US, has all vendors in the supply chain within its borders. Thus, total self-sufficiency should be seen as a mirage.
Our primary driver for creating an indigenous semiconductor ecosystem should be to reduce the cost of direct or indirect import of semiconductor components. The domestic consumption of semiconductor components in 2020 was $80 billion and by 2030 it is expected to be $110 billion, rivalling what India pays for the import of oil. This makes a strong case for a strategy of building export-oriented chip design companies (called fabless companies,which do not own a semi-conductor foundry) in India.
The profits from sales of semiconductor ICs primarily flow to fabless companies and, fortunately, the design of ICs is done entirely in software, which is a strength for us. Also, India has a large pool of IC designers trained by multinationals operating in India. This highly skilled workforce, coupled with our hardy entrepreneurial class, is a significant advantage. If incentivised well, this will play out as it did for our software industry. Fabless company start-ups, with their innovations, can build a world-class export industry. Most such companies will employ hundreds of engineers working for two to three years before the first chip is available to earn revenue. The incentives offered to them should be tailored to this reality. Design-linked incentives offered will not incentivise the correct type of fabless companies.
The government should also create a pool of capital that can help large Indian conglomerates to acquire and relocate the headquarters of mature fabless companies from abroad. These mature companies are valuable in growing ecosystems due to their significant IP portfolios and worldwide customer network. And the human talent they nurture would become future entrepreneurs.
Instead of capitalising on India’s primary resource — the talented software and hardware designers — the current policy seeks to incentivise manufacturing factories for semiconductor components, display panels, specialised components like silicon photonics, and packaging/testing ICs. The fourth one offers design-linked incentives for fabless companies.
A company proposing to build a project within the first three schemes is offered a subsidy of 30-50 per cent of the total project cost. Incentives for a fabless company involved in the design and productisation of components are provided a subsidy capped at ₹15-30 crore per company.
Even these large investments will be insufficient for setting up cutting-edge semiconductor fabs. At the same time, the fabless design industry, which has much greater potential, is relegated to an afterthought. Moreover, several components of the proposed incentive package seem less important. There may not be enough value in incentivising packaging and testing companies. Domestic assembly and consumer electronics design is highly desirable but not critical to creating an ecosystem for the semiconductor industry.
Take the long view
Instead of pursuing a pipe dream or wasting money on minor innovations, India should take the long view. The success of the auto-parts industry points the way. These manufacturers did not start by producing the most complex parts used by foreign car OEMs like BMWs. Instead, they focused on building parts of lower complexity where they had a cost advantage and, over time, moved to manufacture more value-added parts.
Similarly, building fabs using 130 nm to 90 nm process technology instead of cutting-edge 4 nm to 7 nm technologies should be the focus. This older technology is used to develop the bulk of mixed-mode logic circuits used as sensors, input, output data, and power electronics. It will be easier to capture a share of the world market here, as there is zero loyalty to brands and we can provide quality parts at the lowest price.
Successful establishment of such fabs can be the stepping stone to building more advanced fabs. This will help reduce imports since these ICs are the most used in domestically produced goods such as TVs/ refrigerators.
Given the supply chain disruption, India has a window of opportunity. Using a combination of mass production of routine ICs that can substitute imports and increase our exports in combination with the development of fabless companies designing state-of-the-art product innovations, India can kickstart the semiconductor industry. In summary, subsidy will be a catalyst only when it is focused right.
The writer is a serial entrepreneur, and a member of India and US-based angel investor networks. Views are personal