India Telecom

January 24, 2005
India as an emerging design hub: Challenges ahead

Geetanjali Wadhwa & Pradeep Chakraborty

BANGALORE, NEW DELHI AND PUNE -- The India Semiconductor Association was recently launched in Bangalore. Its mission: to make semiconductor excellence the next big wave in India. Dr. Sridhar Mitta, chief mentor, stressed that India "direly needed a local, vibrant semiconductor industry. S. Janakiraman, president and CEO, R&D services, MindTree Consulting and executive member, ISA, pointed out the opportunities for India in contract electronics manufacturing, analog, RF and MEMS (micro-electro mechanical systems) foundries, semiconductor packaging and test industry, and finally in becoming the design services hub of the world.

Apparent dichotomy in numbers
At a panel discussion during a seminar organized by the ISA concurrently with the BangaloreIT.com held last October, Dr. Pradip Dutta, managing director, Synopsys, a leading DSP firm, touched upon what all India needed to do to become a major global semiconductor player. The consumption of electronics in India was forecasted to reach US $9.7 billion in 2005, as against US $150.3 billion in China, and US $452 billion in the US. Next, 85 venture capital firms invested US $162 million in tech companies as per 2003 data. In comparison, 700 venture capital firms invested US $9.2 billion during the same period in the US. Apart from this apparent dichotomy in numbers, PC penetration, Internet connectivity and business IT spend were all significantly lower than the global average.

Although India had over 10,000 IC designers, 250,000 annual engineering graduates, prestigious engineering institutes, and owned one out of seven patents worldwide, the ground situation was different. He said: "We need to see how many architects were there and how many were VLSI trained! Next, what sort of depth and breadth do we have in R&D and manufacturing? India is an important part of the ecosystem, and the 'I' in the BRIC (Brazil, Russia, India and China) economies. It has shown the most rapid potential among the BRICS and has the ability to manage 6 percent growth over the next 50 years. India would go on to become the third largest economy after China and the US by 2050."

'Scars on the back' necessary
Dr. Bobby Mitra, managing director, Texas Instruments India, also, a panelist in the discussion, touched upon some other points regarding how India could become a major semiconductor player. He said that talent was one area that was extremely experience intensive. People literally needed to have "scars on their backs." Next, microelectronics had a high entry barrier. However, it paid at a very high value as well.

According to him, some other areas of entry barrier were tool cost and the cycle time of getting a sample in and out of the country. "The game ends when the chip gets into production," Dr. Mitra said. Systems and applications posed another entry barrier. He added: "We need strong talent end-to-end - both front-end and back-end. The small percentage of chip designers in India are systems savvy. Yet, it is a paradox of what is going on here. Indian engineers have delivered thousands of products in India. Companies are now pushing the envelope in chip design." He added that the results had been outstanding in pockets. He noted: "India was going strong without a tailwind in semiconductors. For that to happen, microelectronics must become a national agenda. It must be among the top two areas under the big IT umbrella. Another area that would make us strong is very strong execution."

Challenges in developing a robust industry
Do we have it in us to become the design hub in the region? Somshankar Das, CEO and president, e4e Inc., said that India was a latecomer and had no established track record, though the semiconductor industry has been in Asia for the past 25 years. He was speaking about developing a robust semiconductor industry in India, at the ISA seminar during BangaloreIT.Com. He added that the other challenges included high cost of capital investments, poor infrastructure and little market pull.

Tapan Joshi, vice president, marketing, eInfochips Ltd., said two major challenges needed to be overcome by India's technology industry on its path to becoming a design hub. One, the emergence of indigenously designed electronic products specifically for the Indian and regional markets. This will spur development of product definition and architecture skills necessary for any product design activity. Two, electronic design services firms must move up the value chain and develop 'whole products', in addition to the current focus on engineering services. He said: Currently, most technology companies are either captive design centers of multinationals or third party service providers with a few IPs (intellectual property). Overcoming these two hurdles will contribute significantly toward the establishment of India's technology industry as a serious design hub.

Dr. Vivek Mansingh, managing director, Portal Information Technology India Pvt. Ltd., noted that the availability of talent (senior technical people who are of global standard in skills), conducive infrastructure and the need for more success stories were necessary. Ramakrishna Dutt, managing director, Quasar Innovations Pvt. Ltd., which is the first Indian company to develop a full-feature GSM/GPRS phone added that logistics, infrastructure, government clearances and bureaucracy proved to be a hindrance for India in its growth as a design hub.

Anirudh Mathuria, country head, SiNett Semiconductors, a fabless IC firm, highlighted three major challenges. He said: "The first is getting the experience of designing complete chips from scratch. There is no substitute for the actual experience of facing problems in silicon. The more you get to the design, the more expertise you will gain. The second challenge is to get the silicon manufactured easily (local manufacturing will help in curtailing the cycle time) and debugging it. This will build the expertise to do the complete job. Easy access to tools, which may be quite expensive, is the third challenge."

Kamal Aggarwal, vice president, marketing and strategy, SoftJin Infotech, an India EDA (electronic design automation) software development services company, added that the key challenges before India were the availability of trained manpower trained in VLSI (very large-scale integrated design) and electronic design at an engineering level, system level electronic product design perspective for conceiving full-fledged products, and a need to shift from a services-oriented business mindset to a product-oriented mindset in order to capture a larger pie of the value being created.

Softjin recently released free source codes of GDS-II and OASIS data exchange libraries and tools. It released a free suite of IC design layout data exchange libraries and tools for use by IC designers and EDA product firms. This includes GDS-II and OASIS (Open Artwork System Interchange Standard) readers, writers and GDS-II-to-OASIS translator, in source code form. The software suite named 'Anuvad' includes one of the earliest available tools to handle the OASIS format, and the only one available for free use in source code format.

The volume of the data to represent the IC design during the layout design phase has exploded in size, with an enormous growth in the size and complexity of the ICs as well as nanometer accurate manufacturing processes. GDS-II has been the industry standard format for representing the layout data and exchanging the layout data between chip design teams, silicon foundries and mask companies. OASIS, promoted by Semiconductor Equipment and Materials International (SEMI), is a new and much more compact format that is slated to replace GDS-II as the standard data format used for exchanging layout data.

On the right path - foundry+fabless or fabless only
Regarding what needed to be done to put India on the right path, e4e's Das added that the fabless company plus foundry company model was well established since 1987. There was not an urgency to build fabs right away. "Even IBM finds a 12-inch fab expensive," he said. Next, global risk sharing business models have evolved since the mid-90s. Two factors in India's favor were the opening of the Indian economy, which led to growing trade, and the emergence of India as a mass market for consumer electronics and telecom.

"We must start here. Too often, people start with the supply side and wait for demand to catch up; often times, it is too late," Das said. According to him, the drivers of demand were overseas and local markets. One, there was a need to establish India as a base for volume electronics manufacturing (selective enclaves linked efficiently to the outside world). Two, local markets, specially, consumer electronics and telecom, were emerging. "Semiconductor success depends heavily on having a contract electronics manufacturing capability close at hand," he pointed out.

Regarding the supply side, Das added three key factors needed to be considered for nurturing the overall ecosystem. These were the creation and deployment of the right talent pool locally, product definition and design, and manufacturing (fabrication, assembly and test). There was a need to increase the establishment of US-based fabless semiconductor firms in India, as it would be a 'tacit encouragement of a reverse brain drain.' This had happened in many cases before, in Taiwan, Korea and China. There was an increasing venture capital interest in cross border fabless companies as well.

As for building fabless semiconductor companies in India, Das said India had a major advantage, as Indian talent was a large part of the global semiconductor industry. There could be some element of reverse brain drain. As for other advantages in favor of India, it had local IC design service firms, who were creators of selective IP as well. Development of smart chips with embedded software was ongoing. Next, the US-funded cross border semiconductor firms were setting up development centers in India. The only missing focus was products for the domestic market. However, there were some good examples, like the Indian electronic voting machine.

Think well before building fabs
As for the manufacturing element of the industry, Das said that a 12-inch line costs US $2.5 billion, while an 8-inch line costs US $1.5 billion. "Even IBM thought twice before building a fab." There was a need to think about a leveraged model for wafer fabrication. Two options were, one, partner with Asian countries that have fab experience and capital, but relatively less talent and limited markets. India provides partial investment, talent (if required) and access to markets in India - local and export centric; and two, study and implement the existing co-operation models, for instance, those between Taiwanese fabs and firms like Xilinx, Altera, etc. Such models were built around financial risk sharing.

On the same note, Das added it was possible to set up local IC assembly operations, especially with an adjacent, export-oriented electronics contract manufacturing industry. Testing activities would depend on what kinds of volumes emerged for what types of products. A practical strategy would be testing of high-volume products, again an adjunct to local electronics contract manufacturing.

India as a base for volume electronics manufacturing
Regarding the steps that need to be taken for establishing India as a base for volume electronics manufacturing, Joshi at eInfochips said, that the emergence of OEMs servicing the Indian and South Asian markets with indigenous products was a first step in this direction. This would lead to the growth of volume electronics manufacturing at low costs. To do this, a healthy supply of low cost components and associated services like industrial design, certification labs, testing facilities were pre-requisites. A volume electronics-manufacturing base could then be leveraged to service other geographies as well.

Portal's Dr. Mansingh said that hardware manufacturing needed infrastructure (roads, facilities, water, power, etc.), ports, customs processes that could facilitate trade at global speeds. Quasar's Dutt re-iterated the need to provide the requisite and appropriate infrastructure, besides setting up a conducive, investor-friendly environment, not only for electronics manufacturing but for related industries as well.

SiNett's Mathuria added that setting up an electronics manufacturing setup was quite costly. Also, it was not a one-time cost. It required regular investments to keep pace with new technologies. "First, we need to understand the market we want to go after and then try and build the manufacturing base for that particular market. This would ensure financial viability of the project. From a business perspective, you have to keep the line running continuously. You must have enough designs for that," he said.

Second, there was need to have access to all raw materials. This included trained labor as well. We must be ready when the plant comes up. Next, supply chain management (SCM) would be an important factor to determine the overall cost of operation. Finally, India should have a technology roadmap, and the capacity and intention to execute it. SoftJin's Aggarwal added that key steps required from the government's side included providing the basic infrastructure improvement in terms of roads, ports and airports, labor law reforms in order to allow more flexibility in resourcing, and simplification and continuity in the central and state level tax/duty structure.

Fabless companies can drive semicon
The establishment of fabless semiconductor companies in India is one way to drive the semiconductor industry in India. Joshi said that fabless semiconductor companies in India were well-suited to cater to Asian OEMs. Such companies could drive the entire industry from several perspectives - moving up the value chain, generating spillover design services opportunities, creating a healthy pool of 'spec to silicon' chip design engineers, creating a market for EDA tools and other ancillary services like packaging, testing, high-speed board designs, and lastly, generating wealth for all stakeholders. "Consider Silicon Valley, which was dominated by large, vertically-integrated semiconductor companies like Intel. The emergence of fabless semiconductor firms energized the entire ecosystem and grew semiconductor industry sales," he noted.

Dr. Mansingh added that it would drive the design and IP creation (to a lesser extent) activity only. The good news was that India was finally going to have a fab in Hyderabad. Dutt at Quasar said that the establishment of fabless semiconductor units would provide necessary value addition as an economy from having design houses. Manufacturing units provide volumes but semiconductor companies would prove to be an important aspect of the entire system. These would impart value-added inputs and aid in the development of skills.

Mathuria, the head of a fabless IC company, noted that fabless companies would allow the local talent to design chips. Thus, a talent pool would be created, which could be tapped for designing and production of chips for the local market as well. Secondly, if the cost structure could be kept very competitive, fabless companies would use this facility to do mass production as well, for which they are currently using either Chinese or Taiwanese companies.

Aggarwal added that emergence of fabless semiconductor companies, who market their own chips, would be a useful step in realising more value for the same design effort as compared to a purely design services company. With the similar semiconductor designer talent and tool infrastructure, but bearing a greater risk, a fabless semiconductor company can hope to create more value in a shorter amount of time.

Smart chips with embedded software or more required?
Is development of smart chips with embedded software the way ahead, or more has to be done? Joshi said that if India's technology industry was successful with the three steps discussed earlier, then it would most likely occur through the development of embedded applications/devices with smart chips and smarter firmware. About 98 percent of the world's processor chips were used in the embedded applications and Pentiums accounted for a measly 2 percent unit volume each year. This clearly highlighted the importance of embedded semiconductor products and the accompanying software/firmware.

Dr. Mansingh agreed that this would be a significant part of the semiconductor business. However, this industry is huge (half the size of India's GDP), and lot more things needed to be done if India wanted to play a major role. Dutt said that smart chips with embedded software would definitely drive the future. However, simply the development was not enough. "The entire system has to be interested in driving the development of opportunities and enhancement of skills, which in turn, will fulfill in-house country requirements and address international requirement," he noted. Mathuria concurred, saying that this alone might not be enough. India must have domain expertise to implement the smartness in a practical manner. We need architectures as well to design the chip, while keeping the domain in mind.

SoftJin's Aggarwal spotlighted the fact that every future system-on-chip (SoC) would be having a significant portion of software. India, with capabilities in VLSI design and software development, could potentially capture a larger share of the SoC design by focusing on VLSI design and embedded software. This also implies that we would need system engineers - who understand hardware and software design, and are equipped to explore system architecture options that are a mix of hardware and software.

Leveraged model for wafer fabrication
There is definitely a need to think about a leveraged model for wafer fabrication. However, its impact must also be taken into account. Joshi said that having wafer fabrication in India would boost the backend design activities. It would be easier for the Indian fabless chip design companies to deal with Indian fabs and manufacture chips that can be used by contract manufacturing companies. Many market analysts predicted the growth of contract manufacturing in India in the coming decade and this could make product development more cost effective.

According to Quasar's Dutt, there was certainly a need for a leveraged model for wafer fabrication as once a leveraged model was established, it would become a catalyst in driving the growth of the industry. It would act as a propellant for advancement for industry as a whole.

Setting up local IC operations
Finally, the tricky one - set up local IC assembly operations, especially with an adjacent, export-oriented electronics contract manufacturing industry. eInfochips' Joshi said this could be achieved by partnering with Taiwanese and Malaysian packaging houses to set up shops in India. This did not require the presence of semiconductor fabs in India to begin with. However, such fabs would expedite the entire process. Nokia, Elcoteq, etc., have already announced plans for manufacturing cell phones in India. As more OEMs begin to do this, IC assembling and testing facilities would crop up around them.

Dr. Vivek Mansingh of Portal Information Technology India added that assembly operations required the ability to bring the chips from fab quickly and export those equally quickly after assembly. For achieving this our ports and customs would have to be very efficient. These operations needed infrastructure as well. Dutt said that one of the ways to achieve this would be by developing theme-oriented parks. These parks would emerge as an ecosystem of its own for this industry, as it would ascertain that an interdependency was established between the various related industries within this park. It would further ensure that the logistics problems were reduced and the time-to-market came down as the entire system was in one place. For these benefits to emerge, it should be ensured that these theme parks were completely self-sufficient.

Mathuria at SiNett felt that setting up of local IC operations would be a logical fallout left to the market forces. Fabless companies outsourced both production, and assembly and test. It was imperative that both functions, if done physically close, would provide a big cost leverage. Both should go hand in hand.

The final word came from Das of e4e Labs. He concluded at the ISA seminar: "The semiconductor industry is part of the supply chain in electronics manufacturing - it cannot be viewed in isolation. We need to create a robust demand-pull locally to build a successful semiconductor industry. Favorable conditions exist today to build fabless semiconductor companies linked to India. We need creative partnership models with the other friendly countries in order to develop a manufacturing capacity that can support the Indian semiconductor industry."