Since the 2008 global financial crisis, industrial policy has resurfaced in respectable economic discourse, after decades of being dismissed as misguided interventionism, particularly in developing countries. What is striking about this revival is that it is being led by the advanced economies that once rejected it. The push into artificial intelligence and renewable energy has only hastened the shift.
For developing countries, the return of industrial policy offers opportunities, but only if three binding constraints are addressed: a weak enabling environment, limited autonomy in policy-making, and tight fiscal space. Industrial policy is often reduced to subsidies and tax breaks, but for most developing economies, far more basic foundations must be put in place. Without reliable digital connectivity, dependable power supplies, trusted data-protection regimes, and a skilled workforce, ambitions for AI-led growth will amount to little more than rhetoric.
Policy space is also constrained externally. World Trade Organisation rules limit the use of instruments — export-contingent subsidies, local-content requirements, and technology-transfer mandates — that underpinned earlier industrialisation successes, particularly in East Asia. At the same time, major economies increasingly chart their own course. The US, the EU and China are deploying industrial policy at scale, often bending or breaking the very rules that others are expected to observe. The imbalance is stark: of more than 2,500 industrial-policy measures introduced globally in 2023, these three economies accounted for almost half.
Fiscal limits further narrow the options. In many developing economies, up to 80% of public spending goes to wages and debt service, leaving little room for long-term investment. Unlike the US or the EU, poorer countries cannot summon vast subsidy packages or bankroll multi-billion-dollar technology programmes. And while technology parks and incubators have multiplied across Africa and Asia, few have delivered meaningful results. As UN Trade and Development (UNCTAD) has noted, such zones succeed only when anchored in established supply chains. Without that grounding, they risk becoming costly white elephants — impressive on paper, but inert in practice.
Against this backdrop, the most sensible approach to AI for developing countries is not to build frontier models, but to deploy those that already exist. Unburdened by legacy infrastructure, many can leapfrog directly into emerging technologies, much as they skipped landlines and moved straight to mobile telephony. Using AI costs a fraction of what it takes to build it. Tools such as ChatGPT can be adopted without erecting data centres or assembling elite engineering teams.
Targeted applications can be transformative. In health care, AI-assisted diagnostics can help ration scarce clinical capacity. In education, digital platforms can partly offset chronic teacher shortages. In agriculture, predictive analytics can support farmers navigating climate volatility and erratic rainfall. These uses may not dazzle those at the technological frontier, but they can deliver tangible returns where they matter most.
They also represent industrial policy at its most effective: pragmatic, experimental and rooted in domestic realities. As economist Dani Rodrik has argued, success lies not in following a fixed blueprint, but in identifying sectors where public action can unlock hidden potential.
Even so, a modest innovation agenda requires financing, and domestic venture capital remains scarce in many developing economies, where private wealth often migrates abroad. Governments can help crowd in private capital by building the right institutions: blended-finance vehicles, sovereign innovation funds, targeted guarantees and regional technology hubs. Donors, too, have a role to play. According to the OECD, the information and communications technology sector receives barely 2% of total aid-for-trade disbursements — far short of what is needed to build digital capabilities.
Governments must also use digital technologies to improve efficiency, particularly in revenue collection, to create badly needed fiscal space. UNCTAD’s work on customs digitalisation, notably through its ASYCUDA system, offers a telling illustration. In Angola, one of Africa’s largest oil-dependent economies, the shift to digital customs procedures delivered striking gains: revenues rose 44% in one year and 13% the next as analogue bottlenecks were removed.
In Iraq, the returns were even larger. After key border points were digitised, customs receipts jumped by more than 120% in a single year. And in Bangladesh, one of Asia’s fastest-growing manufacturing economies, incremental digital reforms helped deliver average annual revenue growth of around 11% over several years, as compliance improved and leakages were plugged. Such examples underscore how digital tools can strengthen the state even before they transform the private sector.
International cooperation remains essential, but global trade rules also need to evolve. The WTO’s Trade-Related Aspects of Intellectual Property Rights provisions may have made sense in an earlier era, but they now constrain access to critical digital and green technologies. Patent regimes should enable broader diffusion, much as compulsory licensing once did for life-saving medicines.
Collaboration among developing countries is equally important because no single nation can afford the scale of investment required for AI or clean technology. Shared platforms such as CERN demonstrate how pooling expertise can spread costs, share risks and unlock mutual gains. A more promising model for the Global South lies in collective innovation. Many developing countries share similar disease burdens and climate exposures, benefit from abundant data, and draw on relatively low-cost technical talent. Innovating together is not only cost-effective but strategically prudent in an increasingly multipolar world.
The return of industrial policy marks a significant shift in global economic thinking. For developing countries, it is a mixed blessing. The path to industrialisation is steeper and more constrained than before, shaped by demanding technological and regulatory standards. Yet the challenge is not insurmountable. By investing in foundational capabilities, targeting high-impact applications of AI, mobilising innovative finance and using the policy space that already exists, countries can still accelerate development. Success will depend not on imitating rich-country models, but on adapting them pragmatically to local realities.
Sirimanne is senior adviser to the Secretary-General of UNCTAD; Tesfachew is senior adviser at the Tony Blair Institute for Global Change
Project Syndicate