Why India’s semiconductor dreams rest on rethinking labour policy

To prepare for the predicted dominance of artificial intelligence, the Indian government – supported by industry bodies and experts – is leaning heavily on the idea of “upskilling” India's workforce.

“History has shown that work does not disappear due to technology. Its nature changes and new types of jobs are created,” said Prime Minister Narendra Modi at the AI Action Summit in February 2025. “We need to invest in skilling and re-skilling our people for an AI-driven future.”

From attempts to train Indian workers in semiconductor production processes to declaring India the “skills capital” of the world in technology, the government is pitching skilling as the key route for the country to participate in the promised “AI revolution”.

It is important to examine the assumptions underpinning these notions, particularly those related to how a Global South country can effectively participate in global supply chains. By reframing the idea of skills and skilling – for instance, through the co-production of skill-notions and value-capture, and through the idea of collective skills – India can open up policy options that are not visible when it only views skills as individual, objective, and comparable.

Any assessment of skill policy in relation to technological change needs to ask what skill even is, if it can be defined, and why this matters. Some approaches to studying skill, as described by the sociologist Paul Attewell, assume that skills are objective and can be quantified and compared with one another. Work that is seen as non-routine, abstract, or subject to uncertainty, is seen as complex, and people think of skill as the ability to do complex work.

But skill cannot be an objective attribute because there are social dimensions to which activities are recognised as skillful or complex. The historian Nina Lerman shows that social categories like race and gender come to be concretised along with ideas about skills: tasks performed predominantly by women, like tailoring, are not seen as analogous to “technical” activities like engineering. The recognition of work as skillful also affects and is affected by caste, class, and migration status.

There are patterns to how skill is defined at given places and times, and those patterns are co-produced with economic shifts. How then are notions of skills in AI policy co-produced with the pattern of economic value-capture along the semiconductor value chain? What work is recognised as skilled and under which conditions?

Today's semiconductor supply chain is globalised and concentrated across the design, manufacturing, and assembly of chips. US firms like NVIDIA and Intel dominate chip design, while manufacturing is led by Taiwan's TSMC using machines exclusively made by ASML in the Netherlands. The assembly stage is less capital-intensive than other stages and is often outsourced to the Global South.

One assumption in India's upskilling and reskilling policies is that adjusting to AI-affected roles represents a type of upskilling rather than potentially deskilling, and that a change in skills can be compared ordinally. Another assumption is that if workers in India were to upskill and reskill, it would be able to secure an acceptable existence in a global economy where AI is the dominant technology.

These assumptions frame skill as both the reason and the solution to problems arising from AI. For instance, an expert group on skills constituted by the Ministry of Electronics and Information Technology, states, “Job losses due to AI-driven automation are more likely to affect low-skilled workers.” And in the words of a Deloitte executive, “By reskilling the existing workforce...we can ensure a steady pipeline of professionals ready to lead AI-driven innovation.” It is also no surprise that the people who run “skilling centers” in India refer to students as engineering and “below engineering.”

Ironically, when a large number of people are “upskilled” – enough to drive the cost of such skilled labour down in a segmented global labour market – those skills are no longer seen as valuable because they are so easily available, and therefore no longer “above” other skills.

One study of the semiconductor supply chain captures this effect spectacularly when it says, “Worker skill requirements go down along the value chain (that is, design is more skill-intensive than manufacturing, which is more skill-intensive than assembly).”

It is not that the more skilled activities capture more value, but that the activities that capture more value are seen as being more skilled. This effect can be seen when Global North firms refuse to transfer knowledge or skills while operating in the Global South, because if those skills became too diffuse, that work would no longer capture as much value and be considered as skilled.

This effect might stymie India's hopes to achieve skill transfers by attracting semiconductor assembly investments; in Vietnam, the same strategy came up against foreign-owned companies' unwillingness to transfer any skills to the domestic Vietnamese industry, instead drawing Vietnamese engineers away from the domestic industry.

In 1980s Brazil, foreign electronics companies employed far fewer workers in skilled positions than domestic electronics companies; the absence of foreign firms in the computer industry might even have contributed to the creation of domestic expertise in this industry.

Additionally, higher pay scales helped divert Brazilian engineers away from domestic firms where they worked in R&D activities, to foreign firms where they worked primarily in marketing, quality control, and equipment testing. A 2008 study of the electronics industry in Mexico also reveals low levels of knowledge transfer to local entities by foreign companies and regional differences in such transfers.

These diffusion-devaluation dynamics are visible in other industries and are felt more by socially disadvantaged groups. For instance, in Mexico, foreign companies employed educated and skilled women workers in repetitive manufacturing and assembly work for decades.

When these workers responded to changing industrial demands by demonstrating flexibility and performing complex production activities, the companies still classified their work as unskilled, depressing their wages. Similarly, a number of scholars have studied how the idea of women's “nimble fingers” is used to devalue rather than value their work in the electronics industry.

While success in the electronics and information technology industries often follows the international mobility of engineers including in India, Ireland, Argentina, and Taiwan, the counter-examples above show that this knowledge mobility is not automatically a precursor to skills-transfer across all segments of work in the industry.

In India, a similar dynamic is evident in the chip design sector. It is often stated in policy documents and media reports that India is home to 20% of all semiconductor design engineers in the world. Many chip design firms have established design centres in India, and new foreign investments in design have been announced. The government also established a design-linked incentive scheme to subsidise semiconductor design activities in India.

However, foreign companies tend to outsource primarily the labour-intensive parts of research and development activities to India because of the lower compensation that researchers can bargain for in India. There are also inadequate linkages between the needs and wants of the Indian market and more “advanced” chip design activities.

Due to this mix of factors, foreign firms then accrue intellectual property for innovations developed in centres in India. Against this background, India's inability to capture value in the chip design sector is attributed to a lack of workers' skills in “new product conceptualisation and product marketing” although these workers are highly skilled in chip design. For a Global South country, a skill is always that which is just beyond reach, a convenient scapegoat that obscures the subjugation of the country in global supply chains.

Another pervasive assumption about skills is that they reside in individuals. Some studies have developed arguments about collective skills and information. For instance, Edwin Hutchins and Tove Klausen argue, through a study of work in airline cockpits, that cognition in this setting is not individual but is rather distributed across a system of people and artifacts.

Perhaps carrying out certain productive activities at scale can be seen as a skill developed and maintained collectively by all the people who enable the functioning of large-scale productive activity. This skill is held, for instance, by bureaucrats, labour brokers, intermediaries between firms, intermediaries between firms and governments, etc., in regions where production at scale is widespread and functions relatively routinely. This is a collective skill because it is only meaningful when held along with other people. It is a skill because human interactions and understandings are necessary for large-scale and complex production systems to function.

Collective skill may consist in knowledge about the law and how it works in the given region, intuitions about how infrastructure requirements scale, experience managing large groups of migrant workers, or tacit knowledge about decision-makers in institutions.

Collective skill has been visible in the critical role labour brokers and other intermediary networks have played in the growth of app-based work in India, the Philippines, and elsewhere. It is also visible in the role that labour brokers play in managing migrant labour supply for electronics manufacturing in Taiwan.

India's semiconductor policy hinges on positioning India as an alternative to China amid US-China tensions. Often, US firms have been reluctant to move out of China, partially because of the cost of movement, but also because India is unable to provide the kind of plug-and-play atmosphere that China can for businesses. For companies that do want to move out of China, Vietnam is seen as a preferable alternative to India because Vietnam has experience with large-scale electronics manufacturing – an indication, perhaps, that there is a dimension of collective skill in this context missing in India.

The descriptor of collective skill is not entirely correspondent with the idea of state capacity or “ease of doing business” for two reasons. First, collective skill relies on non-state actors including entrepreneurs, business managers, factory supervisors, labour brokers, industry associations, and logistics service providers. Some of these actors might even work in legal gray areas while being critical to production and state approval processes. Second, collective skill emphasises the tacit knowledge of people that makes processes work, rather than the design of the administrative, policy, or production processes themselves.

Thus, a mere focus on augmenting the capacity of the state or reorienting the bureaucracy toward becoming more permissive and efficient might not be sufficient. It might even be counterproductive when it over-regulates actors who work in gray areas, like labour brokers, and deregulates activities that require regulation for building trust among all important actors for the stability of investments.

Such often-deregulated activities include especially the treatment of workers and the consideration of local environmental impacts. The notion of collective skills orients us toward trying to build knowledge and practice among all relevant actors for large-scale production. Stakeholder consultations on AI, international and national AI summits, and investment events bring together state and some commercial actors, but rarely, if ever, serve as opportunities to bring together all the actors that make large-scale electronics production possible.

It is not that India must create an economic environment that is above all conducive for foreign technology producers, but rather that while planning to develop a semiconductor industry domestically, India has chosen to focus on skill as an objective, comparable, individual, and therefore easily scapegoat-able attribute. If India were to view skill as collective and co-produced with value capture, new policy questions would open up.

Jai Vipra is a PhD student at the Department of Science & Technology Studies, Cornell University and a non-resident CyberBRICS fellow at the Center for Technology and Society, Fundação Getulio Vargas.

The article was first published in India in Transition, a publication of the Center for the Advanced Study of India, University of Pennsylvania.

Also read: US-China tech rivalry spurs semiconductor ambitions

Source: Scroll
Related Posts: The paradox of artificial intelligence Amazon loses VP helping lead development of artificial general intelligence 25 Jobs At Risk Due To Artificial Intelligence Virginia Giuffre chillingly predicted her suicide in memoir Dhurandhar 2 Opening Predicted At Rs 200 Cr India vs Namibia T20 World Cup 2026 Predicted Playing 11 Rain predicted for Delhi IND vs PAK T20 World Cup 2026 Group A predicted 11 IIT Mandi Begins Application For MBA In Data Science And Artificial Intelligence UN approves 40-member scientific panel on the impact of artificial intelligence over US objections