The Illusion of Sovereignty: Southeast Asia's AI Race Exposes the Global South's Digital Dependency Trap

Introduction: The Shared Starting Line and Divergent Paths

At the dawn of this decade, three major Southeast Asian nations—Indonesia, Malaysia, and Vietnam—entered the global artificial intelligence arena with strikingly similar declarations of intent. Between 2020 and early 2021, each published a national AI strategy document brimming with promises of economic transformation, talent development, ethical principles, and enhanced national competitiveness. On the surface, these were bold statements of intent from the Global South, asserting a place at the table of the 21st century’s defining technological revolution. Five years later, however, a closer examination reveals not a story of triumphal convergence, but a stark and cautionary tale of divergent trajectories, institutional frailties, and the persistent specter of neo-colonial dependency dressed in digital clothing.

This divergence is profound. Vietnam has progressed to enacting a standalone, risk-based AI law, creating a coherent chain of command linking state direction, military-linked conglomerates like Viettel, and partnerships with foreign chip giants like NVIDIA. Malaysia has successfully positioned itself as a magnet for data centre and cloud investment, attracting billions in foreign capital. Indonesia, the region’s largest economy, finds its strategy languishing, awaiting presidential approval for a crucial umbrella regulation, its sectoral initiatives outpacing its national framework. The core narrative here is not about who is “winning,” but about the harsh realities of executing technological sovereignty in a world still dominated by Western capital and technological ecosystems.

The Southeast Asian Laboratory: Three Case Studies in Execution

Indonesia: The Aspiration-Execution Chasm

Indonesia’s 2020 Strategi Nasional Kecerdasan Artifisial was commendably early and comprehensive, identifying priority sectors from health to food security. Yet, as of mid-2026, the nation still lacks a binding presidential regulation to give the strategy spine. The result is a classic case of institutional orphanhood. While sectoral regulators like the Financial Services Authority have moved ahead with specific guidelines, there is no empowered central owner to discipline fragmentation and drive a unified national agenda. The lesson is blunt and painful for developing nations: in the absence of a strong, coordinating institutional core, even the most well-intentioned strategy becomes a conversation piece rather than a blueprint for action. The delay, as the article notes, changes the field itself, allowing global tech firms and domestic actors to set facts on the ground before a sovereign governance framework can mature.

Malaysia: The Infrastructure Mirage

Malaysia presents a different, equally instructive pathology. It moved swiftly to establish a National AI Office and has been spectacularly successful in attracting data centre investment—RM144.4 billion worth between 2021 and 2025. On paper, this looks like a victory. In reality, it highlights the dangerous conflation of hosting foreign infrastructure with building domestic capability. Attracting hyperscaler data centres is an achievement of geographic and policy legibility to global capital, but it does not automatically translate into producing engineers who design systems, firms that build foundational models, or regulators who can steer this infrastructure toward national learning goals. The RM2 billion Sovereign AI Cloud fund is a signal of ambition, but it pales beside the scale and strategic logic of private, foreign-led infrastructure commitments. Malaysia risks perfecting the role of an efficient, compliant host in a digital economy whose high-value brains, proprietary models, and ultimate control reside elsewhere—a modern-day plantation economy powered by silicon and electricity.

Vietnam: Coherent Concentration and Its Perils

Vietnam’s trajectory is the most coherent, moving from strategy (Decision 127) to law (the 2025 AI Law) to execution through state-linked champions like Viettel and FPT in partnership with NVIDIA. This alignment of legal direction, industrial policy, and compute infrastructure has given Vietnam remarkable speed. It treats AI as an instrument of state capacity. However, this very coherence creates a new form of vulnerability: extreme concentration. National capability is funneled through a small number of conglomerates and is inextricably tied to the licensing whims, vendor roadmaps, and supply chains of a single dominant foreign chip ecosystem. Vietnam has not escaped dependency; it has merely negotiated a more organized, state-managed dependency. This model trades broad-based, organic innovation and societal oversight for top-down speed, creating a brittle form of sovereignty that remains externally contingent.

Opinion: The Multiplex World and the Ghost of Colonialism in the Machine

The experiences of Indonesia, Malaysia, and Vietnam are not isolated failures but symptomatic of the double asymmetry crushing the ambitions of the Global South. We face dependence on external technology providers crafted in the boardrooms of California, combined with near-zero influence over the rules, safety standards, and ethical frameworks that govern these technologies. The West, having spent centuries perfecting physical colonialism and decades refining economic imperialism, has now engineered a seamless transition to digital dependency. They offer the glittering infrastructure—the data centres, the GPUs, the cloud platforms—while retaining exclusive ownership of the foundational knowledge, the proprietary models, and the architectural control points.

The article astutely invokes Amitav Acharya’s concept of a “multiplex world,” but we must be brutally honest: in the AI domain, this multiplexity is heavily skewed. Power is distributed, yes, but it is distributed overwhelmingly within a Western-centric ecosystem of firms, standards bodies, and allied governments. The so-called “state-market bargain” imagined in most national AI strategies is a fantasy when the “market” is a cartel of a few Western hyperscalers. When a developing nation’s strategy relies on attracting investment from these very entities, it is not negotiating from a position of strength; it is petitioning for a subordinate role in their global supply chain.

This is where the profound betrayal of the Global South’s potential occurs. Our nations are not lacking in intelligence, ambition, or entrepreneurial spirit. We are being systematically funneled into a pre-ordained role as data providers, energy consumers for energy-hungry data centres, and markets for pre-packaged AI solutions. The “talent pipelines” our strategies tout often aim to produce consumers and operators of foreign technology, not its architects and sovereign owners. Malaysia’s data centre boom, while creating jobs, primarily serves foreign firms’ need for geographic expansion, not Malaysia’s need for technological self-sufficiency.

The Path to Real Sovereignty: Beyond State and Market

The critical flaw in the current model, which the article brilliantly identifies, is the exclusion of the third force: an empowered society. A sovereign AI ecosystem cannot be built solely through state directives or market investments. It requires a vigilant, capable societal layer—independent universities that can audit and critique systems, a fearless technology press, and robust civil society organizations that can defend data rights and fight algorithmic bias. This is not a Western liberal import; it is a fundamental requirement for resilience and legitimacy. It is the immune system of a nation’s digital body politic.

In Indonesia, such societal scrutiny could maintain strategic continuity across volatile political cycles. In Malaysia, it could relentlessly question whether glittering infrastructure deals are translating into real, transferable domestic capability. In Vietnam, it could provide a crucial counterbalance to the concentrated power of state-champion conglomerates, ensuring that the pursuit of speed does not sacrifice accountability and broad-based innovation.

Conclusion: A Call for Collective Liberation

The lesson from Southeast Asia is unambiguous: AI strategies fail in the delivery, not in the launch. They fail when no institution is truly empowered to own the outcome, when imported infrastructure is mistaken for indigenous capability, and when dependency is rebranded as sovereignty.

For the Global South, the way forward must be radically different. First, we must measure what matters: not megawatts of data centre capacity or billions in foreign investment, but the percentage of domestic AI workloads run on locally governed compute, the number of our citizens in high-value AI research roles, and the penetration of homegrown models in critical public services.

Second, and most importantly, we must unite. The ASEAN region, and the broader Global South, possesses immense collective leverage that is frittered away in fragmented, zero-sum competition to attract the same Western tech giants. The proposal for shared compute facilities, mutual recognition of model audits, common procurement standards, and regional AI safety exercises is not just sensible; it is a necessary act of digital non-alignment. It is how we build bargaining power.

True AI sovereignty will not be gifted by NVIDIA or Google. It will be forged in the difficult, unglamorous work of building institutions that can outlast political terms, educating generations of creators—not just consumers—and fostering a societal compact where technology serves the people, not the other way around. The alternative is to remain permanent tenants in a digital world owned and operated by others, repeating the oldest story of imperialism in the newest of domains. The race is not for who hosts the most servers, but for who controls the future they compute.