Volume 8: AI Sovereignty and Dependency in the Global South

Seven volumes have traced how AI reshapes industries, employment, supply chains, resources, military power, and governance across the Americas. Each volume has revealed a recurring pattern: AI development concentrates in ways that create dependencies, and those dependencies translate into power relationships that determine which nations control their technological destinies and which become permanently subordinate.

For Latin America and the Caribbean, this pattern isn't academic abstraction. It's lived reality playing out in real time as the region navigates between competing visions of digital development. The question isn't whether to engage with AI, that ship has sailed. The question is whether nations in the Global South can chart independent courses that allow them to benefit from AI without surrendering control over critical infrastructure, data, and decision-making capabilities that define sovereignty in the 21st century.

This volume examines what digital sovereignty means in practice, why dependency traps are so difficult to escape, and whether pathways exist for Latin American nations to build genuine technological autonomy or whether structural forces make dependence inevitable.

Defining Digital Sovereignty

Digital sovereignty encompasses the ability of nations to govern their own digital systems, infrastructure, and data flows without undue external control or coercion. It means self-determination over which technologies get deployed, on whose terms, and for whose benefit. It means control over the physical infrastructure where data resides and gets processed, the legal frameworks that govern digital activities, and the economic arrangements that determine value capture.

But sovereignty doesn't require complete self-sufficiency, an impossible standard for all but the largest economies. Uruguay will never build its own semiconductor fabrication plants or develop proprietary AI models that compete with American or Chinese offerings. That's not realistic and probably not desirable. The question is whether Uruguay can maintain genuine optionality, the ability to switch providers, develop alternatives if critical suppliers become unreliable, and participate in AI development as something more than passive consumer.

The concept emerged from intersecting concerns. Indigenous data sovereignty movements emphasized communities' rights to control information about themselves. Food sovereignty debates questioned agricultural systems that made developing nations dependent on foreign seeds and chemicals. Digital activism in the Global South highlighted how platforms and infrastructure controlled from Silicon Valley or Shenzhen could be weaponized for surveillance, manipulation, or simply to extract value from local economies without returning meaningful benefits.

For Latin America, digital sovereignty connects to historical patterns of dependency where the region exported raw materials and imported finished goods, remaining perpetually at the periphery of global value chains. The fear is that AI replicates this pattern in digital form: Latin American nations will provide data, cheap labor, natural resources, and markets while technological capacity, value capture, and decision-making authority remain concentrated elsewhere.

The Infrastructure Dependency Trap

The most visible dimension of digital dependency is infrastructure: the physical systems where computation happens and data gets stored. Without local data centers, cloud infrastructure, and network capacity, everything flows through facilities controlled by external actors.

The Latin America data center market, now worth about $5 billion, is forecast to nearly double by 2029. Private investment is driving significant expansion. Hyperscale investors such as Amazon Web Services, Microsoft, and Google are pouring billions into new cloud regions and server farms. Brazil, Mexico, and Chile dominate this build-out so far, with secondary markets like Colombia, Peru, and even Panama starting to attract large facilities.

This infrastructure boom looks like progress, and in many ways it is. For decades, the lack of local facilities forced many companies to rely on servers on other continents, leading to higher costs, significant latency, and security risks. Local data centers reduce those problems while creating specialized jobs and fostering local digital ecosystems.

But who owns and controls this infrastructure determines whether it serves national development or simply extends foreign dominance in new form. When Brazil's national security apparatus relies on Amazon Web Services for data storage and processing, when Mexico's emerging AI industry depends entirely on U.S. cloud providers, when Chile hosts Microsoft and Google data centers but exercises minimal control over their operations, sovereignty becomes more aspiration than reality.

Brazil is learning that establishing tech sovereignty is easier said than done. In contrast to the country's strong rhetoric around asserting independence and pushing back against U.S.-based digital platforms, recent announcements underscore how tied up Brazil is with foreign technology. In early October, Brazil's Institutional Security Bureau, tasked with assisting the president on national security and defense policy, published an ordinance paving the way for a technical cooperation agreement with Amazon Web Services.

The paradox is brutal: the agency responsible for national security outsources data storage and cybersecurity infrastructure to a foreign corporation subject to U.S. laws including the CLOUD Act, which requires U.S. companies to share data hosted on their servers, regardless of location, when subpoenaed, and FISA, which allows U.S. intelligence agencies to request user data without warrants under national security grounds.

The Cloud Computing Paradox

The Latin America cloud computing market reached USD 46.70 billion in 2024 and is expected to reach USD 184.00 billion by 2033, exhibiting a growth rate of 14.70% during 2025-2033. This explosive growth reflects genuine digital transformation as businesses migrate operations to cloud platforms for scalability, cost efficiency, and access to sophisticated tools they couldn't build independently.

In recent years, major cloud providers have dramatically expanded regional presence: AWS expanded its São Paulo Region with additional availability zones in March 2024; Microsoft Azure launched a new dedicated cloud region in Santiago, Chile in June 2024; Google Cloud partnered with Banco Santander in September 2024 to develop AI-powered financial services; and IBM Cloud entered a strategic alliance with Mexican telecom giant Telmex in January 2025 to deliver hybrid cloud solutions across Latin America.

These investments create genuine benefits. Internet adoption in Latin America has surged from 43% in 2012 to nearly 78% today, overtaking major economies including China. Cloud platforms enable startups, improve government services, and allow firms to compete globally without massive capital investments in on-premise infrastructure.

But the flip side is growing dependency. When government services, financial systems, healthcare records, and business operations all run on cloud platforms controlled by foreign corporations, sovereignty erodes even as efficiency increases. A 2025 NTT DATA–MIT Technology Review study revealed over 80% of regional companies are in advanced cloud adoption, which signals digital maturity but also deepening lock-in to systems designed, controlled, and ultimately governed by entities outside the region.

The challenge is structural: cloud services exhibit powerful network effects and lock-in dynamics. Once an organization builds systems around AWS, Azure, or Google Cloud, switching becomes prohibitively expensive due to data migration costs, retooling requirements, and staff retraining needs. The longer the relationship persists, the deeper the dependency becomes, until the theoretical possibility of changing providers becomes practically unrealistic.

Brazil's Sovereignty Struggle

Brazil offers the most instructive case study in the region's sovereignty struggles because it combines genuine ambition, significant resources, and repeated confrontations with the limits of what mid-sized economies can achieve against concentrated global tech power.

In early July 2025, Washington imposed a 50% tariff on Brazilian exports, officially framed as applying political pressure on Brazil's Supreme Federal Court investigating former President Jair Bolsonaro, but documentation revealed a broader rationale: the claim of "unfair competition" posed by Brazil's state-run instant electronic payment system, PIX, against major U.S. companies like Visa and Mastercard.

PIX represents Brazil's most successful digital sovereignty initiative: a public digital payments infrastructure that processes transactions instantly without fees, dramatically expanding financial inclusion while reducing dependence on foreign payment networks. But PIX's success triggered retaliation from firms whose business models it threatens, demonstrating how sovereignty initiatives face not just technical challenges but geopolitical resistance.

Around a year ago, in August 2024, Brazil launched its pioneering Artificial Intelligence Plan (PBIA), aiming to advance local and public AI solutions with relative independence from US tech giants, including by developing its own data centers. Less than a year later, little remains of this initial goal of digital sovereignty.

What happened illustrates the mechanisms through which sovereignty ambitions get undermined. In September 2025, President Lula da Silva's government introduced Provisional Measure 1,318, which established the Special Taxation Regime for Data Center Services (ReData), aiming to attract investment and enhance technological capacity through tax incentives, including elimination of federal taxes on equipment and promotion of renewable energy use.

The policy looked like sovereignty-building: incentivize local data center development, reduce dependence on foreign facilities, build indigenous AI capacity. But critics argue the anticipated tax benefits will primarily advantage large multinational technology companies, which could build or expand their data centers at significantly reduced costs, thereby widening the gap with local providers and increasing reliance on foreign infrastructure.

Brazil ends up subsidizing foreign tech giants to build infrastructure they control, deepening rather than reducing dependency. The government gains local data centers but not genuine sovereignty because ownership, operational control, and strategic decision-making remain external.

Brazil's domestic AI regulatory framework is embedded in the discourse of "digital sovereignty," but while Brazil highlights sovereignty and autonomy in both AI development and regulation, its regulatory model still largely references EU standards, leading to a rule-taker role in AI global governance. Even Brazil's ambitious AI legislation, approved by the Senate in December 2024, adapts rather than creates frameworks, borrowing heavily from European models because developing entirely novel governance approaches without templates to follow proves prohibitively difficult for mid-sized economies.

The Data Extraction Model

Infrastructure dependency enables a broader pattern: data extraction where Latin American activities generate information that flows to platforms headquartered elsewhere, where algorithms process it, where insights get monetized, and where value accumulates.

Social media provides the clearest example. Brazil is among the top markets in social media adoption, the fifth market in terms of streaming adoption, and early adopters of most digital services, with around 85% of the people connected. This creates enormous data flows: billions of posts, messages, searches, and transactions daily.

But platforms that mediate these interactions, Facebook, Instagram, WhatsApp, TikTok, YouTube, X, are all controlled by foreign corporations that determine what data gets collected, how it's processed, what gets prioritized in feeds, and how the economic value generated through user attention translates into revenue. Latin American users provide the raw material; platforms extract the value.

Initiatives such as Google News Showcase in Brazil have led to financial dependence for local media outlets. Since 2020, this program has established partnerships with Brazilian media organizations to pay for news curation. While marketed as support for local journalism, it has resulted in economic reliance, with many media outlets' income now tied to contractual agreements with Google. This enhances Google's role as dominant intermediary in the information landscape, diminishing editorial independence.

The model extends beyond platforms to AI systems themselves. When Latin American hospitals adopt AI diagnostic tools developed in the United States, trained on predominantly North American patient data, they become dependent on systems that may not work well for local populations and that extract information about local patients to improve models that remain proprietary to foreign companies.

Are There Pathways to Sovereignty?

Given these structural challenges, we must ask honestly: Can Latin American nations build genuine AI sovereignty, or is dependency inevitable for mid-sized developing economies?

The pessimistic case is straightforward. AI development requires resources that concentrate overwhelmingly in wealthy nations: massive compute infrastructure, specialized chips, proprietary training data, rare technical talent, and patient capital measured in tens of billions. China and the United States together account for the vast majority of global AI investment, research output, and commercial deployment. The gap between them and everyone else isn't narrowing; it's widening as network effects and economies of scale compound advantages.

Latin American nations lack the fiscal capacity to match these investments. They lack the technical ecosystems that produce AI researchers at necessary scale. Their markets, while substantial in aggregate, fragment across borders in ways that prevent achieving the scale U.S. and Chinese firms enjoy. Their universities, though producing quality research, operate with budgets that pale compared to Stanford, MIT, or Tsinghua.

From this perspective, sovereignty talk is fantasy. Latin America will consume AI developed elsewhere, run on infrastructure owned elsewhere, governed by rules written elsewhere. The only question is whether to align primarily with the United States or China, accepting dependency on one or attempting to play them off against each other for marginally better terms.

But a more nuanced view identifies spaces for strategic autonomy even within structural constraints.

First, sovereignty doesn't require matching U.S. or Chinese capabilities across the full AI stack. It requires controlling critical nodes that provide leverage and options. Brazil's PIX demonstrates this: it doesn't require Brazil to compete with Visa globally, just to create domestic alternatives that reduce dependency for the most strategically important use cases.

Second, regional cooperation can achieve scale that individual countries cannot. Brazil's national data center plan alone, if implemented, could push the country's data center storage from about 1 GW of installed capacity today to 8 GW. If Brazil, Mexico, Chile, Argentina, and Colombia coordinated data center development, renewable energy investments, and technical training programs, they could build genuine capacity that serves hemispheric needs rather than depending entirely on facilities controlled from Silicon Valley or Seattle.

Third, focusing on specific domains rather than attempting comprehensive AI leadership offers realistic paths forward. Uruguay cannot compete in foundation model development, but Uruguay's public telecommunications company ANTEL, which provides internet access to 94% of Uruguayan households and fiber optic connections to 99% of those households, could develop ANTEL cloud services that host government data and systems. This wouldn't make Uruguay an AI superpower but would give it genuine sovereignty over critical public sector data.

Fourth, leveraging commodities and resources as bargaining chips with tech companies could extract better terms. Latin America possesses critical minerals, abundant renewable energy, and data center-friendly climates. Nations could condition access to these resources on technology transfer requirements, data localization agreements, and genuine partnerships rather than simple extraction relationships.

Fifth, building on existing technical capacity in specific niches creates leverage points. Brazil has sophisticated fintech and agtech ecosystems. Argentina has strong software development capacity. Mexico has deep manufacturing expertise. Rather than attempting to compete across the full AI spectrum, these nations could focus on domains where they possess advantages, developing specialized capabilities that make them valuable partners rather than mere markets.

The China Alternative?

Any discussion of Latin American digital sovereignty must confront whether Chinese technology offers genuine alternatives to U.S. dependence or simply substitutes one master for another.

China's Digital Silk Road has made significant inroads across Latin America through infrastructure investments, technology partnerships, and financing on terms Western companies often won't match. Chinese telecommunications equipment, cloud services, and smart city technologies appeal to governments seeking to avoid exclusive dependence on U.S. vendors.

But Chinese alternatives come with their own sovereignty implications. The technology may be cheaper and more readily available, but it embeds dependencies on Chinese vendors for updates, maintenance, and support. Data flows, even when infrastructure sits locally, often route through servers controlled by Chinese entities. And Chinese technology partnerships typically come with expectations about political alignment and acquiescence to Chinese interests that can conflict with national sovereignty as surely as U.S. partnerships do.

The uncomfortable truth is that playing U.S. and Chinese vendors off against each other provides tactical advantages but doesn't fundamentally solve the sovereignty problem. It creates redundancy and possibly better terms, but it still leaves Latin American nations as customers rather than genuine partners in technological development.

True sovereignty requires building indigenous capacity, however limited, that provides real alternatives when external providers prove unreliable or exploitative. That capacity doesn't emerge overnight and requires sustained investment over decades. The question is whether Latin American nations will make those investments or continue deferring to short-term efficiency over long-term autonomy.

What Comes Next

AI sovereignty and dependency in the Global South represents perhaps the most difficult challenge in this series because the structural forces working against genuine sovereignty are so powerful while the costs of permanent dependency are so severe.

We've seen how infrastructure, cloud computing, data extraction, and governance frameworks all tilt toward dependency. We've examined Brazil's struggles to assert sovereignty even with significant resources and political will. We've confronted the hard question of whether sovereignty is even achievable for mid-sized developing economies or whether the best they can hope for is choosing their dependency relationships carefully.

But we've also identified potential pathways: focusing on critical nodes rather than comprehensive capabilities, leveraging regional cooperation, building on existing strengths, and using resource endowments as negotiating leverage. These won't make Latin America independent AI powers that rival the United States or China. But they could provide enough autonomy to avoid permanent subordination.

Next week, we turn to scenarios: how might the Americas look in 2036 depending on choices made today? What happens if current trends continue? What alternative futures become possible if strategic decisions shift? The sovereignty questions discussed this week shape which scenarios are achievable and which remain fantasy.

The AI transformation is happening regardless of whether Latin America builds sovereignty or accepts dependency. The only question is what role the region plays in that transformation and whether future generations will look back on this moment as when pathways to autonomy were forged or when they were foreclosed forever.

This is Part 8 of a 10-part series on The Geopolitics of Artificial Intelligence in the Americas by Core Geopolitical Insights LLC. Follow along each week as we explore how this transformative technology is reshaping power, prosperity, and security across the Western Hemisphere. | Next week: 2036 Scenarios