A relook at our semiconductor policy

In late 2021, the Centre announced an outlay of ₹76,000 crore (about $10 billion) to develop semiconductors and display fabs in India. Providing subsidies to support infant industries is not unusual. Taiwan and China have both done this successfully. Taiwan Semiconductor Manufacturing Corporation (TSMC) is today the world’s largest manufacturer of advanced semiconductor ICs. It all began with a subsidy to TSMC, founded in 1987 under the leadership of its visionary founder Morris Chang who had earlier worked for 25 years at Texas Instruments.

Similarly, the Chinese government provided funds to the Yangtze Memory Technology Corporation (YMTC) to build flash memory for cellphones, laptops and solid-state drives. Both TSMC and YMTC have succeeded spectacularly. TSMC has more than 60 per cent worldwide market share of microprocessors and advanced semiconductors produced by foundries. YMTC has shipped more than 300 million memory chips so far and has caught up with the state-of-the-art in memory devices, which store 2 terabits in a single package.

What made them succeed was more than a single-shot infusion of money. Such efforts also need an infusion of management and technical talents. India’s Modified Semiconductor Policy 2023 needs to provide multi-layered, multi-year financial support, focus on a segment instead of casting a wide net, and design subsidies appropriate to that industry segment.

Besides, an experienced CEO, good marketing talent, and senior technologists are essential for success in the semiconductor manufacturing industry. Ignoring this increases the chance of failure. The achievements of TSMC and YMTC are rightfully credited to the top executives and technology experts they managed to hire from US-based semiconductor companies like Texas Instruments and Intel.

Our policy design has so far overlooked this factor. A policy that selects projects based on the credentials of the promoters and the amount of private investments they commit would be more likely to succeed if it also includes a provision to hold back a significant portion of the subsidy contingent on the merits of a team recruited. An executive who has previous experience in developing a new process and the fabrication plant should be more highly rated than one who has been in the general management function in the semiconductor industry.

Our policy incentives essentially provide 50 per cent of the project cost, a one-time assistance. In advanced technology businesses like semiconductor manufacturing, such a one-time subsidy needs to be complemented with additional financial incentives in the future. TSMC and YMTC were assured of financial support throughout their journey until they became profitable. This is necessary because, in the emerging phase, such companies do not accumulate enough profits to fund the R&D required to keep up with rapid advancement of technology.

Our policy incentivises diverse segments of semiconductor manufacturing, offering broadly similar incentives for all segments. This may lead to sub-optimal outcomes. Consider the example of a venture that manufactures analog IC components and thus competes with Analog Devices or Texas Instruments. In such a case, the breadth of the catalogue, that is, the number of products, is the essential factor that gains a market share.

A fabless foundry business such as TSMC or a memory business such as YMTC would need different incentive structures. In the first case, one should incentivise a rapid introduction of products, whereas in the latter case, one should incentivise fast convergence to the latest technology by incentivising R&D efforts.

Effective policy measures

Two effective policy measures we do not use are to support ancillary industries to reduce the cost of inputs or to create a market pull for the products of the primary recipient of the subsidy. Consider a semiconductor fabless manufacturing plant. Since this business has a thin profit margin, it must have lower cost of inputs and labour to be competitive. Government subsidies to create a geographically local ancillary industry that offers such inputs would reduce costs.

Another way the government can help is by developing educational institutes to develop a workforce that this industry needs and supporting academic-industry partnerships. Hence, the importance of creating multi-layered schemes appropriate for the target industry segment.

The total sum of $10 billion assigned for the incentives is insufficient to build a single highly advanced semiconductor manufacturing foundry or display panel manufacturing plant. Besides, this is on offer to a wide range of segments of the semiconductor ecosystem. We need to narrow the focus by clarifying the primary reason for providing this subsidy.

We would want to subsidise display panels or memory component manufacturing if we aim to reduce import bills for electronic items. On the other hand, if we seek to gain strategic immunity from a shortage in case of war, we should build advanced foundries and R&D capabilities to design components and manufacturing processes. A sharper focus will lead to a better chance of success.

To narrow the focus and allot capital appropriately, we need bureaucrats to become knowledgeable about the industry for which they design subsidies. Thus, the short tenure of senior bureaucrats is a factor responsible for our faulty policy designs. However, depending entirely on industry experts does not work as they will have their axes to grind.

We recommend the following changes to redesign this policy:

Hire an experienced industry executive from the executive ranks of US-based semiconductor companies to head the effort.

Assign a senior bureaucrat as co-CEO with a 10-year term.

Let this team develop policies and monitor their implementation.

Because multi-layered incentive schemes will likely emerge as the best choice, this team should report directly to the Prime Minister. That would avoid turf battles that prevent designing a compelling set of subsidy schemes; and

Give the team a clear mandate about the primary objective, and establish a five-year horizon to show results.

The writer is Founder and President, Maker Bhawan Foundation, dedicated to supporting hands-on R&D in Indian technical institutions. Views are personal

COMMENT NOW