JPMorgan report on chip manufacturing

Chip design is also an opportunity: India clearly has an opportunity to build an eco-system that can lead it to the same path as Taiwan, a powerhouse in chip design. India already has a very large and growing pool of experienced IC design engineers, and hundreds of expat engineers are returning to India every year (like Taiwan in the 80s). The investment in eco-system can start at the grassroots level with a modification of education to make it targeted more at problem solving and creativity, and continue with infrastructure and financial support for R&D centers that focus on enabling technologies of tomorrow.

India’s semiconductor manufacturing: Too late—just don’t bother The Indian government is drafting a semiconductor policy, which is likely to be manufacturer-friendly, following the trend of many Asian countries. The chip manufacturing unit requires all sorts of infrastructure including basic things such as uninterrupted water and power supply in addition to the land provided at attractive rates by the government. The policy has seen delays due to a standoff between the finance and IT ministries on the issue of the fiscal incentive package for investors setting up fabrication units. The finance ministry had suggested that investors could choose a mix of incentives (equity participation, interest subsidy, tax exemption, duty reduction) for setting up their project, with the overall cost of incentives restricted to 20-25 percent of the total capital expenditure on the project. This policy and incentives are still under review by the ministries.

In order to attract foreign and domestic investors, the Indian government is also encouraging Electronic Hardware Technology Parks (EHTP) to foster electronics manufacturing in the country. Some of the benefits of the EHTPs are:

• Foreign equity is permissible up to 100 percent.
• No duties are levied on the import of capital goods, raw materials and components.
• Access to the Indian domestic market is allowed up to 50 percent of the free-on-board value exports.
• An EHTP unit can be set up for both software and hardware operations in an integrated manner.

Anyone starting an IC fabrication facility in India will first face the dilemma of choosing a technology level. If the fabrication is of mature technology—it will be in direct competition with the large capacities built-up by China and other countries that are themselves struggling for profitability. Profitability in such a scenario is almost unachievable. On the other hand, building a state of the art fab requires, other than financial and physical resources; access to technology, skilled manpower with experience on foundry process, and customers willingness to devote resources to switch over to a new entrant. In either scenario, the possibility of success is minimal.

According to a recent news story, the government is planning to revise the draft Fab policy to extend fiscal incentives to other technology products such as LCDs, OLED, plasma panel displays, storage devices and solar cells. This is a favourable approach compared to just focusing on semiconductors as it provides incentives to a larger array of manufacturing and assembly options within the technology industry rather than restricting it to the capital intensive semiconductor manufacturing.

IC assembly, packaging and test, challenging, but a viable option than front-end manufacturing: IC assembly involves a different and lower order of complexity than what is required in front end wafer processing. IC assembly is functionally separate from the other stages of production even when performed in close proximity to fabrication. With final testing the finished chips can be shipped directly to customers—a large part of which are in Asia itself.

IC assembly business dominated by a few players: The four dominant players in IC assembly space are ASE, SPIL, Amkor and STATS ChipPac. These companies have built up significant technological leadership in the advanced packaging/testing capacities and the high initial investment required is a natural barrier to new entrants. Advanced packaging already forms close to 70 percent of the packaging market from revenue stand point.

Turnaround after disastrous 2001-02: Following disastrous performance in 2001-02 time-frame, the industry has seen some consolidation. Unlike the foundry sector, with cautious capital expenditure, the industry has restricted capacity addition in recent years, which has turned out to be good for the industry as a whole.

Location and technology: Indian backend companies would need to develop strong packaging and test capabilities along with capabilities of direct shipment to end customers. Local IC assembly operations could come up adjacent to hubs of electronics contract manufacturing industry. Initially, in the absence of foundries, Indian houses could partner with packaging houses from Taiwan, Singapore or Malaysia for technological knowledge, especially on advanced packaging techniques.

Raw material supply security: Another important factor in the packaging industry is securing supply of substrate and raw material that will help reduce manufacturing costs and maintain profitability. Big companies such as ASE and SPIL have managed to secure substrate supplies. The absence of any sort of ecosystem is likely to be a problem area for Indian backend companies, as they will have to import most of the raw materials.

Niche plays may be viable, new technology and R&D costs could be prohibitive:There are many small backend companies in Asia that are just about profitable. The key for these companies has been focus on specific, growing market opportunities, such as memory and display drivers and establishing long-term contractual relationships with key potential clients. This is exactly what any start-up in this sector will have to do to generate profits and have some sort of buffer against the cyclical nature of the industry. A start-up would need to spend significant amount of resources to enter the advanced packaging market, which forms over 70 percent of the total packaging market. Moreover, the top four dominant players enjoy a considerable lead in advanced packaging, so even intensive R&D cannot guarantee commercial competitiveness and profitability. Hence we believe that smaller players would have to target the niche market to be profitable.

IC design is a better option than IC manufacturing in India

India should focus on opportunities that require less monetary investment but more human capital. The semiconductor design process is skill intensive, and requires only Electronic Design Automation (EDA) software. Fabrication, on the other hand, needs a huge fixed investment (of the order of $3 billion) to build and operate a fabrication plant that holds a wide variety of expensive equipment and that meets extreme requirements of cleanliness. Assembly and test also requires expensive equipment, but the overall costs of plant and equipment are much lower than for the fab, as are the average skill requirements. Overall, worker skill requirements go down along the value chain and upfront investment costs are highest for manufacturing.

Reason why India can be a favoured destination

India has certain advantages over China, including superior technical education, high number of English-speakers and IP protection laws and belief boosted by the successful Indian software sector. The fact that India has no chip foundries should not matter as much since most big companies are used to designing in the US for foundries in Taiwan. Some of the design companies value the opportunity to design on a 24-hour cycle because of the pressure to reach the market ahead of competitors.

IC Design is already picking up in India

According to iSuppli, India’s design services industry grew from revenue of $511 million in 2004 to $623 million in 2005. India has about 125 companies doing design, including multinationals, domestic companies such as Wipro and Sasken, and a handful of Silicon Valley startups with Indian R&D centers. In 2005, giant multinationals like TI, Intel, Cypress, Infineon, and STMicroelectronics comprise about 70 percent of the total semiconductor design industry in India and about 30 percent was produced by homegrown companies.

In 1985, Texas Istruments became the first multinationals to open a design center in India, to work on its design automation software for internal use. TI’s India center designs at 90nm and has begun doing 65nm and also designs for DSP devices, the company’s flagship product line. TI India has been awarded the highest number of patents (225) for any R & D center in India.

Wipro’s VLSI team has evolved from job shop to full design services, with 1,400 engineers, including more than 840 IC designers. It began to win business in ASICs and SOCs five years ago and in the last two years has taped out 125 designs in automotive, consumer, industrial and other segments. About 25 percent were 130nm and 10 percent were 90nm. It recently taped out its first 65-nm design.

STMicroelectronics’ India design center was established in 1992. Among the multinationals, it has one of the largest staffs: 1,650, mostly in Greater Noida, with 50 in Bangalore. About two-thirds work on hardware design.

Intel India’s Bangalore operations include the Intel India Design Center and sales and marketing operations. Intel India employs over 2,500 people and more than half are design engineers.

Investment should also be targeted at the grassroots level, Quality education is key to employment generation

From the point of view of employment generation and environment friendliness, the IC Design industry is a much better choice than IC manufacturing. There is high demand for experienced engineers in the VLSI sector, and the demand will only grow as multinationals source more designs from India.

One challenge is the supply/demand gap for qualified professionals, despite a strong education network in India. The quality of engineering education plays a key role in creating interest for large multinational companies and developing a local IC design industry. Japan, Korea, and Taiwan score well in this area due to the existence of large number of good quality colleges. Taiwan and Japan have most likely moved closer to Western style education (focus on problem solving and creativity rather than information reproduction). While India has some great institutions, average quality is not good enough. Instead of wasting money on subsidies for $30mn a piece of lithography equipment, India can spend the same money on higher education and that includes incentives for attracting the world’s top universities to open campuses in India.

TI, STMicroelectronics and most other large multinationals are already engaged in university programs to help build design expertise in India. TI India, for example, has a University Relations program under which it works with several Indian universities and has helped to establish many DSP labs which train more than 1,500 students in graduation and post-graduation studies every year.