> ## Content Index > Fetch the complete content index at: https://aipriorities.digitalfutureslab.in/llms.txt > Use this file to discover other available public pages before exploring further. # The Invisible Edge: Labour, Scale, and India’s Geopolitical Leverage in AI - URL: https://aipriorities.digitalfutureslab.in/the-invisible-edge-labour-scale-and-indias-geopolitical-leverage-in-ai/ - Published: 2026-10-03T18:12:32.000Z - Updated: 2026-10-03T18:12:32.000Z - Author: Hari - Tags: #Import 2026-10-12 10:52 > Sarayu Natarajan and Astha Kapoor ## Abstract Artificial Intelligence is reshaping global geopolitics by redefining economic power, technological sovereignty, and security paradigms. For India, AI is a strategic imperative and an arena of opportunity as it enables key advancements in economic growth, governance, defence, and innovation. India occupies a dual position as both a vast consumer and an emerging producer of AI, offering it a distinctive geopolitical vantage point. As a consumer, India represents a diverse, large-scale testing ground that can shape global norms around inclusion, ethics, and affordability. As a producer, its growing ecosystem of startups, public digital infrastructure, and skilled workforce allows it to contribute to global AI value chains. This essay examines how India’s scale, capability, and democratic ethos can position it as a credible and balancing voice in shaping the emerging global AI order. ## Introduction Artificial Intelligence (AI) has moved from being a frontier technology to becoming central to geopolitics. Countries are positioning the development and deployment of AI as drivers of growth, development, and strategic autonomy. This pursuit was visible at India’s recent AI Impact Summit, where the Government of India simultaneously courted American Big Tech capital while advocating for Global South solidarity, seemingly seeking both the resources and the values. Yet, in the context of a growing scramble for AI supremacy, India, like many other countries, appears long on desire and short on strategic vision, particularly when viewed through the lens of geopolitical advantage. The focus in India has largely been on the development of frontier models, which, as things stand, has two significant and deeply entrenched players. This is especially surprising since, in the context of the Digital Public Infrastructure (DPI) discourse, India advanced a "third way" model of digital development for low- and middle-income countries through its 2023 G20 presidency. Therefore, when examining AI governance, it is necessary to unbundle India’s genuine strategic advantages and consider how they might be purposefully cultivated to craft a pathway for impact. Conceptualising AI as a lever of strategic advantage is not straightforward. Unlike natural resources, AI is not a singular technology or object. Rather, it is a set of technological artefacts, capabilities, and material and intangible assets. The cumulative control of this bundle compounds into geopolitical power. In this scenario of rapid technology evolution, and the predominance of two superpowers with appreciable capabilities in frontier technologies, we enquire into India’s role and priorities in AI. Although it has yet to demonstrate unassailable access to compute capacity or models, we argue that India does have access to strategic levers that can strengthen its global engagement. India occupies a dual position — it is potentially a singularly large consumer of various types of AI technologies, while also being a producer of these very technologies. This duality provides differentiated strategic advantages that have the potential to magnify India’s role in norm-setting around AI, with spillover benefits for other domains. This paper examines this dual role and argues that, to reap the benefits of AI, India must play an active role in shaping norms governing both its development and deployment, beyond the pursuit of frontier technological advancement. The essay is structured as follows. First, we establish the conceptual basis by examining the need to shift away from conventional and technology-deterministic views of influence and power. We then examine India’s role as a producer and consumer within global systems of dependence. Finally, we lay out a high-level strategy for India to position itself as a value-centric leader in this space. ## From Compute to Complexity: Network Dependence and Power in AI By centring structural power in understandings of AI and geopolitics, we attempt to move away from narrow framings of compute and frontier models as power to a more networked understanding of AI development. This approach allows us to acknowledge the complex interdependence that characterises AI and examine other forms of dependencies. Compute- and model-centric views of AI power treat technological capability as the sole determinant of geopolitical influence. This approach necessitates large capital investments, seemingly concentrating power in the hands of a few firms or states. Another dimension of this compute- and model-centric view relates to the U.S. export controls on China, including recent efforts to restrict export of chipmaking equipment. These views on influence are both narrow and self-serving — narrow because they make AI seem like a zero-sum game, focusing on single pieces of the puzzle; self-serving because they have an aura of dominance due to their focus on a limited set of issues, though reality may be more fractured and complicated. Susan Strange offers a useful framework here. She argues that structural power — both primary (security, production, financial, and knowledge structures) and secondary (transport systems (sea and air), trade, energy and welfare) — matters. In the context of AI, influence may lie in shaping the systems within which technology operates — production networks, data flows, labour markets, and governance norms. Structural advantage in AI systems cannot be extracted from the global supply chains within which they are embedded. For one, AI is resource-intensive, requiring natural resources (like water and silicon), data, and human input at various levels. Driverless cars cannot be trained without a large dataset of labelled media content; the algorithm must learn, through repeated iterations, to distinguish between a safe, motorable road and an obstacle. Such data is often labelled and compiled through human input. Companies making these systems are therefore reliant on global supplies of these inputs. AI can be viewed, therefore, as not merely a linear system from production to consumption, but an interconnected global system in which dominant producers of AI models consume materials and services from many parts of the world. These dependencies are shaped by the cost advantages some countries offer, or standards of quality driven by existing industries. For example, Toto, a Japanese designer-manufacturer of bathroom fittings, has capabilities in ceramics processing and is therefore significant in the semiconductor space. Consumers of AI also shape the trajectory of AI development through demand. For example, a U.S. company cannot develop driverless cars without contextualised, labelled images, which are often produced by workers in various parts of the world. Similarly, the elasticity of demand for driverless cars may be shaped by local safety regulations and the availability of cheaper alternatives. Robert Keohane and Joseph Nye’s work is helpful here in analysing the complex interdependence of actors. Their work further encourages a shift from deterministic approaches focused on frontier models and computing power. Keohane and Nye argue that power emerges from network position rather than linear dominance. Hence, factors like the centrality of a node and the vectors of dependency matter in evaluating each actor. The global system around AI can thus be imagined as a network, with state and corporate actors forming its various nodes. Influence in contemporary geopolitics also flows from states’ embeddedness in transnational and regional networks. These include regulatory, technological, and knowledge systems and networks. For India, this perspective reveals overlooked sources of leverage in the global arena — scale, labour, and governance capacity. It shifts attention away from who builds the most advanced models to who shapes the ecosystems through which AI is deployed and developed. Cumulatively, a networked view of influence is analytically and strategically useful to better understand India’s position. ## Networks of Dependence: Middleware and Labour as Levers India’s role as a producer lies in two areas: first, it acts as a middleware power. India has a large pool of technical talent that translates frontier models developed by firms into deployable, context-specific applications. This includes adapting models for a variety of settings and needs, ranging from tools for multilingual environments to AI applications for public service delivery and welfare. This role is apparent from the numbers. The India Skills Report points out that India now contributes 16 per cent of the world’s AI talent. Moreover, Indian talent is relatively highly skilled — the Stanford AI Index report 2025 reveals that India has a high relative penetration of AI skills (2015–24), second only to the United States. This indicator measures the intensity or degree of AI skills in a particular country. India also hosts over 1800 Global Capability Centres (GCCs), of which nearly 30 per cent are focused on AI. Taken together, these indicators suggest that India’s role as a middleware state cannot be underestimated. Second, India provides labour infrastructure for AI. Indian labour is central to data annotation and model training pipelines. Indian workers also do testing and system maintenance. This labour is foundational to AI development, even if often undervalued. Examples abound: seemingly automated tasks, such as virtual assistance, may rely on invisible workers transcribing files, testing systems, and acting as human failsafes when AI fails. Human workers also play an important role in mitigating toxic content and bias by sifting through vast datasets to identify problematic material. Moreover, the quality of labels for datasets and the fidelity of data itself can distort the results of algorithmic audits, affecting critical AI evaluation parameters. Hence, labour is central, not peripheral, to the making of AI. Indeed, some estimates suggest that half of the global data labelling workforce is based in India. There is considerable evidence suggesting that these workers experience low pay, poor working conditions, atomisation of tasks, and limited access to social welfare. Further, they have little recourse to grievance redressal systems. Although there is limited speculative or counterfactual research, it is safe to argue that constraints on the supply of labour or capability to deploy AI would stymie the global diffusion and adaptation of AI. This asymmetry presents a political economy conundrum: changes to labour welfare regulations and laws around certification of data for AI could, for example, alter the cost structures of ensuring high-fidelity, high-quality data sets for global AI firms. Cumulatively, the two themes highlight the centrality of India and Indian firms to the global economy. If these capabilities were withdrawn, global AI systems would encounter a supply crisis marked by rising costs, declining model quality, failing systems, and slower deployment cycles. This, in turn, provides India with crucial bargaining power. A few risks are worth considering in this context. First, the annotation industry is diffuse and fragmented, and no single firm predominates, making alternatives easily available. Second, despite its relevance, it accounts for a small portion of costs and profit pools, making Indian labour —whether in IT services or data annotation — potentially replaceable. Third, as AI becomes more able to produce synthetic data, some of these capabilities may be brought in-house. There are no foolproof solutions to these challenges; however, strategic investment in Indian human capital to deepen reliance and secure realistic profit pools may be a worthwhile strategy. ## Leveraging Scale and Diversity for Norm Setting India’s position as a massive consumer of AI technologies is itself a distinct axis of geopolitical power. With a population exceeding 1.4 billion people, a rapidly expanding digital economy bolstered by DPI, one of the world’s fastest-growing smartphone user bases, and a growing user base of large language models, India represents an unparalleled market for AI deployment. The growth of DPIs such as UPI has led to enormous consumer data also being available digitally, which is critical for any frontier model. Further, India’s linguistic diversity, with 22 constitutionally recognised languages, is a major attraction for frontier model developers seeking to test their models. States that constitute large, diverse, and strategically consequential markets have historically wielded considerable influence in shaping the standards, regulatory frameworks, and ethical contours of emerging technologies. India’s consumer scale, in this reading, becomes a form of structural leverage in the global AI order. Beyond scale and linguistic diversity on the consumer side, India’s complex democratic landscape and the diversity of lived experiences with regard to technology make the country a microcosm of the different types of issues faced by users on platforms. Combined with its emphasis on welfare-based technological deployment at scale, this points to a distinctive approach to inclusive public interest versus the market-led views most considered in the context of technological rollouts. Therefore, in an environment where global AI governance is fairly nascent and fragmented, the practice of norms (as distinct from legally binding norms) is likely a useful governance mechanism. As seen in various AI Summits and the UN Global Dialogue on AI Governance, current global efforts are largely directed towards non-binding, norm-setting processes. There is evidence from China of ethical norms being embedded in AI governance, including those of data protection, human control over AI, and regulation of monopolies. Norm-based governance is especially critical in a time when governments across the world are pursuing the “myth of deregulation,” and quasi-regulatory action increasingly operates beyond the realms of law and policy. In this environment, India’s scale and diversity can create the evidence needed to influence frontier models. First, the country’s large user base can compel systems to adapt to the peculiarities of the population and embed regulatory and safety expectations within these models. Second, Indian users can generate dense and diverse evidence on both the uses and harms of AI, thus proving more powerful in directing necessary safeguards. Finally, India's experience of welfare delivery at scale, juxtaposed with market-driven models of deployment, offers an opportunity to embed public-interest values in the design and governance of frontier AI models. ## India’s Geopolitical Strategy India's pursuit of AI sovereignty has been defined, somewhat narrowly, by its ambition to compete in the development of frontier models. Sarvam AI, an indigenous large language model oriented toward India’s linguistic diversity, was unveiled with considerable fanfare at the AI Impact Summit. Yet the occasion also carried the implicit acknowledgement that India’s models are unlikely to rival those produced by OpenAI or Anthropic in the near term. This gap is not incidental. It reflects structural constraints that are unlikely to dissolve quickly: the absence of deep pockets for compute procurement, the long-term institutional investment required to build world-class talent, and the chronic underutilisation of India’s considerable data resources. In competing on terrain defined by others and on terms it cannot yet meet, India has lost its strategic direction. Its most significant advantages in AI are not found at the frontier; they are embedded, often invisibly, in the global AI supply chain itself, and in its latent capacity to shape the governance frameworks through which AI is deployed worldwide. The pursuit of frontier parity has obscured these advantages rather than complemented them. This is compounded by Indian industry’s limited participation in global AI governance conversations. India’s role in data annotation, multilingual diversity, and safety audits is undervalued, despite its labour-intensive contributions, traditionally seen as less glamorous at the geopolitical level. Yet, as argued in the preceding sections, the systemic dependence of global AI firms on Indian labour and middleware services creates genuine vulnerabilities in the global AI supply chain. India has not seriously considered the switching costs associated with these services — or global dependence on them — as sources of geopolitical leverage. The dependence of global AI firms on Indian labour and middleware services creates vulnerabilities within the global AI supply chain, as demonstrated in earlier sections. If India were to regulate annotation standards, impose welfare or collective bargaining standards on AI-related gig work, or enable bottom-up, community-led evaluations, these actions would propagate through the global AI development pipeline with significant effect. To translate these structural advantages into a coherent geopolitical strategy, India must draw simultaneously on two forms of capital. First, it must build solidarity capital across the Global South. India is uniquely positioned to work with these countries that share similar concerns and developmental ambitions to coalesce around India’s perspectives on AI governance. For this, India will have to demonstrate values-based engagement with the AI governance discourse and create space for collaboration and collective negotiation on global forums such as the Global Dialogue. This will need to be buttressed by alliances and partnerships with other middle-income countries such as Brazil and Kenya, which may have shared interests. Second, India can also advance a welfare-focused discourse within the AI governance ecosystem, much like it did in the DPI discourse. Notions of publicness of interest, governance, and investment are critical to AI governance, and India is well positioned to bring these concerns to the forefront. While pursuing solidarity and advancing a welfare-oriented approach play to India's strengths, recognising existing dependencies within the AI supply chain is equally important. The argument of this paper is therefore not that India should abandon investments in compute infrastructure or frontier models. Rather, it should broaden its strategic focus to recognise and build upon the comparative advantages it already possesses. ## Conclusion India stands at a critical juncture. While it doesn’t have access to the traditional sources of power in the AI supply chain — compute, frontier development, semiconductor manufacturing — it does have the space to craft a more networked and differentiated form of power that can be equally significant. To do so, India’s aspirations must move beyond competition in building frontier models towards cultivating instruments of normative power that can be exercised through domestic AI governance frameworks (such as labour laws, or welfare laws), solidarity-based Global South collaborations (drawing on precedents such as India’s position on generic drug production), and a strategic use of its large consumer and labour base to shape the values embedded in AI (for example, through consumer awareness campaigns that influence user behaviour). If India recognises the weight of these comparative advantages and the leadership role they afford, it has the potential to rebalance power within the AI ecosystem. The question is whether decision-makers will seize this opportunity.