Due to signs of saturation in the growth of most industrial sectors, including the ready-made garments (RMG) industry, declining local value addition, and rising graduate unemployment, Bangladesh has been under tremendous pressure to find the next sector for job creation and economic growth. Among many options, the semiconductor industry appears to be beckoning Bangladesh. Despite the opportunity, the entry and scalability barriers of this 78-year-old industry deserve in-depth analysis for the effective and efficient use of national resources. It is worth noting that Bangladesh must invest in thinking and analysis to scale up entry and expand it further in an effective and efficient manner.
The invention of the transistor -- an electronic switch and amplifier -- gave birth to the semiconductor industry in 1947. Over the past 78 years, this invention has grown into an $800 billion industry, with no sign of slowing down. Semiconductors, such as microchips, power nearly all industrial products, from smartphones to missiles. More importantly, this industry offers job opportunities to a diverse group of people, ranging from high-school graduates to those with PhDs in physics or electrical engineering. Hence, due to wage differentials and a growing number of graduates, Bangladesh has reason to feel urgency in leveraging this industry to create jobs and drive economic growth. The obvious question, therefore, is what it takes to scale up and expand. So far, the underlying reasons for optimism have been the large global semiconductor market, a growing pool of science and engineering graduates, the success of Bangladeshi-born semiconductor experts in the USA, and wage differentials. Of course, these are precursors -- but are they sufficient?
Correlation with technology and engineering competence: Semiconductor is a science-, technology-, and engineering-intensive industry. It began with a Nobel Prize-winning invention in physics. Does this imply a linear correlation between relevant scientific, technological, and engineering competence and business success in the semiconductor industry? Perhaps not.
It is worth noting that transistor inventor and Nobel Laureate Dr. Shockley's semiconductor venture in California failed to scale and went bankrupt. Ironically, a radio repair shop (Sony) in war-ravaged Tokyo achieved significant success by licensing transistor technology from Bell Labs. Furthermore, despite past success and a vast patent portfolio, Intel has been struggling to remain competitive. On the other hand, Taiwan, starting from a relatively weak science and engineering base, has grown into a major success story. Similarly, despite India's track record of producing high-caliber engineering graduates from the Indian Institutes of Technology -- and their success in Silicon Valley and top-ranking American engineering schools -- India's last three attempts to develop a semiconductor industry have not scaled into a national economic growth engine. Such realities raise an important question: what else is needed to leverage science and engineering competence to turn semiconductor opportunities into an economic growth driver.
Lesson from RMG and information technology: Bangladesh's RMG sector has successfully created jobs and driven economic growth for 40 years. Did it begin with Bangladesh's skill in tailoring or competence in textile engineering? Perhaps not. Its roots lie in the Multi-Fiber Arrangement (MFA), an international trade pact active from 1974 through 1994 that governed global textile and clothing markets by setting quantitative export quotas. This created an entry opportunity for Bangladesh into global RMG trade. To capitalise on it, foreign buyers and RMG producers supported the initial growth phase of the sector. Unfortunately, no such quota-based opportunity exists in the global semiconductor trade.
The rise of India's export-oriented software and information technology (IT) service industry tempted Bangladesh to follow a similar path. Consequently, Bangladesh made massive investments in expanding computer science education and IT skill development programs. It also invested significantly in improving power supply, internet connectivity, and high-tech parks. Despite these efforts, IT service exports have not scaled to become an economic growth engine comparable to RMG. The underlying reason appears to be that the scale, scope, and externality effects of early entrant India have neutralized Bangladesh's wage advantage in skilled IT labor. Bangladesh has faced similar scaling barriers in call center services and business process outsourcing. In retrospect, the large size of the global market, technological competence, and wage differentials are not sufficient to scale initial success into a major driver of economic growth and job creation.
Bangladesh's semiconductor footprint and its scalability: As early as the 1980s, Bangladesh entered the global semiconductor value chain through a Japanese firm's investment in the Chittagong Export Processing Zone for testing and bonding special-purpose LED lighting devices. Unfortunately, this did not scale in the way Intel's entry into Malaysia -- through a bonding facility for memory chips in Penang -- did. After a long gap, in the 2000s, there was a modest attempt to develop a fabless company through power management integrated circuits, but it did not scale. However, it planted the seeds for ULKASEMI, which now offers high-quality semiconductor design services for OEMs, fabless design houses, and electronic system design companies. Subsequently, a few other design service providers, notably Neural Semiconductor, emerged, creating jobs for more than 1,000 professionals and generating approximately $10 million in revenue.
The semiconductor value chain comprises five major segments: (i) microchip R&D and intellectual property creation for next-generation specifications, (ii) design services to translate specifications into detailed designs, (iii) fabrication (printing designs on silicon wafers, known as foundry), (iv) testing, assembling, and packaging, and (v) equipment, gases, and chemicals. As noted, despite early entry and foreign investment, bonding and testing did not scale. On the other hand, existing design services do not appear to be highly scalable. First, the design services market itself is not very large. For example, India's global success has so far created around 25,000 jobs for designers. Besides, Bangladeshi firms cater only the overflow of demand of multinationals, as opposed to creating the market for it through microchip evolution.
Prospect of attracting multinationals: In the 1960s and 1970s, due to eyesight-related health concerns, multinationals sought locations in less developed countries to set up bonding and testing facilities. At that time, infrastructure, logistics, and favourable tariff structures were sufficient to attract such investments. Unfortunately, that opportunity has largely disappeared. To create a business case today, India has offered up to 75 per cent capital subsidies and 4-6 per cent production-linked incentives to attract investments such as Micron's. Specifically, India offered $1.95 billion in subsidies to attract a $2.7 billion assembly and testing plant from Micron, creating only 5,000 relatively low-paying jobs. Such a subsidy race appears infeasible for Bangladesh to win. Moreover, value addition in outsourced semiconductor assembly and testing (OSAT) is as low as 6 percent. Prospects for attracting foreign direct investment in other segments of the value chain are equally challenging.
As noted, multinationals established assembly and testing facilities in several less developed countries in the 1960s and 1970s, including Malaysia, the Philippines, Thailand, and South Korea. However, these did not evolve into national growth engines. The success stories of South Korea, Japan, and Taiwan stem from scaling up humble beginnings of domestic startups. Notably, in Japan and Taiwan, these initiatives created new markets rather than directly competing with established multinational firms in the USA and Europe. For example, Taiwan's flagship company, TSMC, entered by serving small fabless firms that were overlooked by American companies. Similarly, Japan's Sony pursued opportunities that American firms had rejected. It may, therefore, be reasonable to conclude that the success of Taiwan and Japan in building their semiconductor industries stemmed in part from capitalizing on the decision-making failures of American firms.
Despite the attractiveness of the semiconductor industry, the scalability of Bangladesh's early success in design services appears limited. At the same time, entry barriers in other segments, including OSAT through multinationals, remain very high. Although the expertise of Bangladeshi-origin professionals abroad, the growing supply of local science and engineering graduates, and government support through tax incentives are encouraging, they are not sufficient to scale and expand into other segments to transform the sector into a powerful engine of economic growth. Bangladesh's challenge, therefore, is to identify discontinuities in technology, innovation, and business model dynamics -- similar to those leveraged by Japan and Taiwan -- to enable a modest entry that can scale into a major success through a flywheel or snowball effect.
Rokonuzzaman, Ph.D is academic, researcher and activist on technology, innovation and policy., zaman.rokon.bd@gmail.com