The recent digital transformation accompanying artificial intelligence (AI) and frontier technologies is creating new patterns of technological dependency that echo historical socio-economic relationships between developed and developing nations. While the technological mechanisms have evolved, the underlying dynamics continue to reproduce the structural asymmetries of earlier phases of the global system.
As economic historians and sociologists Giovanni Arrighi and Immanuel Wallerstein have observed, global economic relations have historically positioned developing countries as providers of cheap labor, raw materials, and markets for technically advanced goods produced in industrialized core nations. The contemporary AI revolution appears to be replicating these core-periphery dynamics through multiple extractive mechanisms: the exploitation of labor from assembly line workers and data annotators working at below-subsistence wages; the extraction of rare earth materials like cobalt and lithium to manufacture AI hardware; and the appropriation of data generated by billions of users worldwide to train large language models (LLMs) that are then sold back to these same populations as commercial AI services.
Our previous analysis in The Diplomat demonstrated how Southeast Asia and its 700 million internet users face a critical technological sovereignty challenge. The region’s vast data resources are systematically extracted, stored, and utilized by United States and European technology giants like Amazon Web Services, Microsoft Azure, Google Cloud, and OVHcloud. In response, countries like Vietnam, Indonesia, Malaysia, and the Philippines have implemented data sovereignty and localization laws and policies requiring foreign tech companies to cede ownership and store citizens’ data domestically. Yet these regulations fall short of what is required, as no country in the region has successfully broken free from these extractive data flows.
In this article, we build on our previous assessment of Southeast Asia’s digital sovereignty challenges and examine why data localization policies may fail to deliver technological autonomy and sovereignty. Why are these regulatory attempts likely to fall short?
While data represents the phenomenological surface or visible flows that governments may be able to regulate, there are deeper structural and material dependencies that transcend data location entirely. The concentration of resource-intensive AI computing infrastructure itself (specialized GPU chips, cloud architectures, and processing capabilities) represents a more fundamental chokepoint. Yet the material requirements of AI computational infrastructure indicate that current global economic and geopolitical arrangements systematically prevent peripheral regions from developing autonomous technological and scientific capabilities.
Data Sovereignty and Localization: Surface-Level Solutions and Gains
Recent scholarship shows that data sovereignty and localization policies, while politically popular, address only the surface layer of technological dependency without challenging the underlying power structures that maintain foreign control over critical digital infrastructure. While data sovereignty emphasizes legal control and data localization focuses on geographic storage requirements, both approaches result in what one can term “sovereignty simulacrum.” This creates the appearance of technological independence while leaving societies and socio-technological systems vulnerable to structural and material forms of control through system dependencies. In the field of AI, the “computational power” and “intelligence” remains foreign-controlled even when data stays local.
As Emily Wu argued in “Sovereignty and Data Localization” (2021) for Harvard’s Belfer Center for Science and International Affairs, data sovereignty and localization policies “are ineffective at improving security, do little to simplify the regulatory landscape, and are causing economic harms to the markets where they are imposed. In order to move away from these policies, the fear of sovereignty dilution must be addressed by alternative means.”
Across Southeast Asia, recent policy implementations demonstrate these contradictions in practice. Indonesia’s 2022 Personal Data Protection (PDP) Law and Electronic Information and Transactions (EIT) Law aim to achieve “digital sovereignty” through the introduction of laws, content, and data control in order to “maintain political, economic, social and cultural stability.” The legislation mandated domestic data storage, pushing Amazon Web Services to open a Jakarta data center in an apparent regulatory victory.
However, this achievement proves hollow when examined closely. Indonesian businesses and government agencies may store their data locally, but they remain entirely dependent on foreign AI systems to process and extract value from that information. This creates a more sophisticated dependency where the most valuable components – like the algorithmic processing capability and underlying computational infrastructure – remain under foreign control, even though the raw data is indeed in-country. Indonesian organizations, from startups and corporations to government institutions and NGOs, developing or using AI systems must still rent computing power from U.S tech giants, relying on foreign GPU chips, cloud architectures, and data centers, making them subject to foreign policy decisions and technological dependencies that transcend algorithmic control.
Vietnam’s Cybersecurity Law regulated (2018) and Decree 53/2022/ND-CP (2022) demonstrate similar limitations. The regulation forced platforms like Facebook and TikTok to lease server space in the country, showcasing regulatory power against global tech giants. However, Vietnam’s broader push for digital sovereignty reveals the same structural contradictions.
The Ministry of Public Security’s acquisition of a majority stake in FPT Telecom has been described as “a response to geopolitical cybersecurity concerns” and an “aggressive push to assert control over critical infrastructure – particularly its digital arteries.” Yet, like Indonesia, the Vietnamese organizations developing AI applications remain entirely dependent on U.S. and European computational resources and infrastructure.
Malaysia’s MyDigital program and National Fourth Industrial Revolution Policy illustrate the most striking contradiction. The latter outlines “three objectives, four policy thrusts and … 16 strategies to prepare the country to embrace the 4IR and cope with current and future disruptions of emerging technologies.” Yet there is not one single reference in the 69-page document to control or ownership of these technologies and their infrastructure.
Despite aims of becoming a “tech powerhouse in South-East Asia” through establishment of dedicated agencies like the National Artificial Intelligence Office (NAIO), with the digital industry contributing “25 percent of Malaysia’s GDP by the end of 2025, up from 17 percent 10 years ago,” both private and public sectors remain structurally dependent on the same Big Tech computational infrastructure that enables AI development globally. This dependency is exemplified by Malaysia’s reliance on massive foreign investments from Google ($2 billion) and Microsoft ($2.2 billion) for its data center infrastructure, reaffirming the fundamental structural restrictions Global South countries face in the AI age.
Singapore demonstrates how even substantial computational infrastructure cannot overcome fundamental questions of technological ownership. Unlike other Southeast Asian countries implementing strict data localization, Singapore emphasizes flexible cross-border data flows while maintaining oversight. As Diganta Das and Berwyn Kwek showed in a December 2024 paper, Singapore’s Smart Nation initiative provided digital infrastructure for comprehensive urban governance, encompassing real-time data collection and AI-driven public services. The city-state attracted massive foreign investment, hosting “approximately 87 data centers” by 2024 (60 percent of Southeast Asia’s total capacity) with Google, Microsoft, and Meta establishing major facilities.
However, these data centers remain foreign-owned and operated. Singapore’s AI Singapore initiative, despite substantial funding, still requires Google Cloud partnerships to access cutting-edge AI capabilities. While Singapore’s digital governance demonstrates greater flexibility than its neighbors, the fundamental computational architecture remains under foreign control.
Even the most successful data center hub in the region cannot escape the structural dependency on foreign-owned AI development tools and infrastructure that defines the contemporary global technological hierarchy.
The Path to True Technological Independence
Data localization policies in Southeast Asia have only achieved the political simulacrum of “technological sovereignty.” The aim of these policies is to keep the data at home and regulate extraction, while the entire process, infrastructure, and material framework remains structurally controlled by foreign corporations.
What we are witnessing is something a tad more sophisticated than just “digital colonialism,” as many scholars have pointed out. There is a deeper form of structural relation that maintains unequal control of power and value production. Like gravity, which invisibly shapes physical phenomena while remaining imperceptible to direct observation, the material structural relations governing computational infrastructure operate beneath policy debates about data flows and control.
In other words, beneath talk of “data flows” and “data localization” lies a spectrum of material transactions and relations that moves downward to computational infrastructure and concrete global market relations. This spectrum encompasses Amazon extracting over 50 percent of sellers’ revenue and Apple demanding 30 percent tribute on app purchases, while Amazon Web Services, Microsoft Azure, and Google Cloud control 67 percent of global cloud computing. Data compiled by Oxford revealed that only 32 countries possess AI-specialized data centers, with over 150 countries having no computational infrastructure whatsoever. The specialized GPU manufacturing, advanced semiconductor fabrication, and cloud computing architectures enabling AI development remain concentrated in core economies like the United States and Europe, creating technological and economic checkpoints that transcend basic policy.
The world-system is not only socio-economically configured. There are also deep structural patterns that stratify, divide, and rank the technological development of societies. The old dynamic of “development of underdevelopment” fundamentally covers the structured international division of labor and technological capacity and ownership. As such, if developed societies are capital-intensive, modernized, industrialized, and digitalized, it is precisely because the periphery persists as an extractive economy, exporting agro-mineral products and providing “super-exploited” cheap non-skilled labor.
AI will only deepen this divide, intensifying the international division of humanity through computational dependencies that determine technological sovereignty. Unless fundamental transformations challenge the structural foundations of this technological hierarchy, the concentration of AI infrastructure will further entrench Global South dependency in the digital age.