Volume 4: Supply Chain & Infrastructure, The Geopolitics of AI in the Americas Series

Code may have mass, but that mass requires physical foundations. Every AI model trained, every query processed, every autonomous system deployed depends on hardware manufactured in specific places, assembled by specific hands, powered by specific energy sources, and connected through specific networks. The digital revolution we've spent three decades treating as borderless and weightless turns out to be deeply rooted in geography, constrained by physics, and vulnerable to the same geopolitical forces that have always shaped which nations rise and which decline.

The COVID-19 pandemic delivered a brutal lesson about supply chain fragility. When factories in Asia shut down, hospitals in the Americas ran short of masks and ventilators. When semiconductor production hiccupped, automakers idled plants and consumer electronics became scarce. These disruptions, painful as they were, pale compared to what a major conflict in the Taiwan Strait would mean for AI development. The most advanced chips that make modern AI possible are produced almost entirely on a small island that sits at the center of U.S.-China tensions.

This concentration of critical capabilities in geopolitically contested regions represents an unacceptable strategic vulnerability. The question facing the Americas isn't whether to build more resilient supply chains and infrastructure, it's whether the hemisphere can move quickly enough to matter.

Why Supply Chains Are Strategic Weapons

Supply chains have always been instruments of power, but their geopolitical significance has intensified as technology becomes more complex and concentrated. Nations use supply chain position as leverage, granting or denying access to critical inputs as a form of statecraft that can be as effective as traditional military coercion.

Consider semiconductors, the foundation of all modern AI systems. Only five or six countries control the advanced stages of the semiconductor supply chain: the United States, the Netherlands, Taiwan, China, South Korea, and Japan. One North American nation, one European nation, and the rest clustered in East Asia. This geographic concentration creates chokepoints that can be exploited or disrupted.

American companies lead in processor architecture and chip design, while the Netherlands supplies an irreplaceable component: ASML's extreme ultraviolet lithography systems, without which cutting-edge fabrication becomes impossible. Taiwan's TSMC handles the bulk of high-end manufacturing. South Korean and Japanese firms contribute essential materials and specialized equipment. China has poured resources into catching up but remains behind in the most sophisticated processes, though it maintains substantial control over commodity chips and upstream materials.

Interdependence appeared economically rational during periods of geopolitical calm. Today it registers as strategic exposure. Any serious conflict involving Taiwan would sever access to the majority of advanced fabrication capacity. Rising U.S.-China tensions jeopardize supply chains for components and materials. Even relationships among aligned nations create dependencies that can be leveraged when priorities diverge.

For the Americas, the reality is stark: even when AI infrastructure is assembled in the United States or Mexico, the components and tools often originate in or pass through Asia. This ties the hemisphere's economic security to stability in the Western Pacific, particularly around Taiwan.

While the Americas have successfully resourced the front-end of the AI stack through historic fab investments in Arizona and Ohio, a critical Advanced Packaging Gap remains. We can design the world's most sophisticated AI processors in California and fabricate them in Arizona, but we still lack the domestic high-volume facilities to package them. Currently, the majority of Back-End Assembly, Testing, and Packaging still routes through Asia.

To close this loop, the focus of the USMCA 2026 Review has shifted toward the Arizona-Sonora-Jalisco Corridor. By scaling Jalisco's Kutsari semiconductor park and Sonora's Green and Safe Corridor, the hemisphere is attempting to bridge this gap. Mexico is no longer just providing low-cost labor; it is becoming the specialized back-end fortress of the North American stack, ensuring that a chip born in the U.S. doesn't have to cross the Pacific to become functional.

The Digital Infrastructure Imperative

If supply chains are the circulatory system of AI development, infrastructure is the skeleton. Data centers have evolved from boring backend facilities to objects of national strategy. These massive buildings filled with servers and networking equipment are where AI models train, where cloud services run, where the computational work of the digital economy happens. Their location matters profoundly because latency, energy costs, regulatory jurisdiction, and physical security all depend on geography.

The United States dominates global data center capacity with over 5,000 facilities. This reflects and reinforces American leadership in cloud computing and AI development. Amazon Web Services, Microsoft Azure, and Google Cloud, all headquartered in the United States, have made massive continuing investments in cutting-edge infrastructure.

The energy appetite of AI infrastructure imposes hard limits on development. Training a single frontier model can demand electricity equivalent to powering a mid-sized town for a full day. These computational requirements grow exponentially, outpacing efficiency improvements from better chip design. Regions offering abundant, affordable, and reliable power hold decisive advantages in attracting AI investment.

Building AI data centers has become tantamount to building military equipment in terms of global power projection. Control over where computation happens translates directly into economic and potentially security advantages.

Expert analysis of the 2026 infrastructure build-out reveals that the ultimate constraint on AI growth isn't just watts, but the Energy-Water Nexus. High-density AI racks generate unprecedented heat, requiring millions of gallons of water daily for thermal management. In the water-stressed regions of the American Southwest and Northern Mexico, data centers are no longer just tenants, they are resource competitors.

As projects like the 10GW Stargate campuses break ground, the metrics for success have shifted from simple Power Usage Effectiveness to Water Usage Effectiveness. Securing a data center site in the Americas now requires more than a grid connection; it requires a Social License to Operate in a landscape where ultrapure water is a finite strategic asset. For the first time, hydrology is as important to AI geopolitics as computer science.

North America's Integrated Position

The United States, Canada, and Mexico have built deep economic integration over decades, formalized through USMCA. This creates a foundation for coordinated AI infrastructure development that no other region can easily replicate.

The United States brings unmatched strengths in chip design, research and development, venture capital, and technology company headquarters. U.S. policy increasingly prioritizes onshoring and nearshoring of critical manufacturing, backed by substantial subsidies through the CHIPS Act and Stargate Project. The challenge is execution, translating policy ambition into operational factories within timelines that matter.

Canada offers distinctive advantages despite its smaller population. Canadian research excellence in AI is globally recognized, with pioneering work in deep learning. The country also possesses abundant clean energy, particularly hydroelectric power, which has attracted data center investment for energy-intensive AI workloads that can benefit from carbon-neutral power sources. The strategic vulnerability remains talent retention; Canada must ensure that its research excellence translates into domestic companies rather than simply feeding American economic growth.

Mexico occupies perhaps the most strategically important position for hemispheric AI supply chain integration. Its deep manufacturing relationship with the United States, geographic proximity, and lower labor costs make it an obvious nearshoring destination for AI hardware production. Key markets like Querétaro already possess data center availability zones and growing technical capacity.

Mexico's investment in Cotilicue, which would be Latin America's largest supercomputer, signals intent to develop indigenous computational capabilities and move up the value chain. But power constraints represent a significant bottleneck. Mexico's energy infrastructure struggles to meet current demand; supporting massive AI data centers will require substantial investment in generation and distribution capacity.

Latin America's Strategic Positioning

Beyond North America, Latin American nations are at various stages of developing digital infrastructure that could make them relevant players or consign them to peripheral roles.

Brazil stands out as Latin America's data center leader. São Paulo alone hosts approximately 493 megawatts of capacity across nearly 200 facilities, making it one of the most significant technology hubs outside North America and Asia. Brazil's strategic position is complicated by its multi-alignment approach. As a leading Mercosur member, it maintains deep trade relationships with neighbors and through that bloc with the European Union. It also trades heavily with China and has been receptive to Chinese technology investment, including Digital Silk Road initiatives.

Brazil has articulated a clear commitment to digital sovereignty, seeking to control its technological destiny rather than becoming dependent on either American or Chinese platforms. This could position Brazil as a genuinely independent AI power, potentially the only one in the hemisphere outside the United States and Canada. But it requires substantial investments in energy, education, and technology development.

Chile functions as Latin America's second-largest data center hub with over 50 facilities and 148 megawatts of capacity concentrated in Santiago. Chile possesses significant copper deposits, critical for the extensive wiring required in data centers and AI hardware. If Chile can position itself not just as a data center location but as a supplier of processed materials rather than raw exports, it could capture more value from the AI buildout.

Colombia and Costa Rica represent nations that built substantial business process outsourcing sectors but now face AI disruption. Both host modest data center capacity. These countries confront a strategic choice: move up the value chain by hosting AI infrastructure and developing higher-skill technical capabilities, or risk being bypassed entirely as AI automates the jobs that drove their economic growth.

Across the rest of Latin America, between 600 and 700 data centers operate, concentrated in major cities. Most nations have maintained multi-alignment strategies, avoiding exclusive allegiance to any major power. This flexibility could be an asset, allowing them to attract investment from multiple sources. But it could also leave them caught in the middle as U.S.-China competition intensifies.

True digital sovereignty in the Americas is also being forged on the seabed. The activation of the Humboldt Cable in late 2026, linking Valparaíso to Sydney, represents a paradigm shift in South-South connectivity. For decades, Latin American data traffic was routed almost exclusively through Northern Hemisphere hubs, creating structural dependency on Atlantic routes.

With the convergence of Chile's Humboldt and Google's Firmina cables, the Southern Cone is establishing Subsea Sovereignty. These trans-Pacific arteries allow Brazil and Chile to bypass traditional Atlantic chokepoints and connect directly to Indo-Pacific markets. This isn't just about faster internet; it's about creating a hemispheric digital architecture that can survive decoupling from legacy networks.

The Hemispheric Vision

The pieces exist for the Americas to establish a genuinely integrated AI supply chain less dependent on geopolitically contested Asian manufacturing. The vision involves each region contributing its comparative advantages to create resilient, redundant capacity across the hemisphere.

The United States and Canada would maintain dominance in chip design, advanced research and development, and venture capital investment. Both would expand domestic semiconductor manufacturing through substantial subsidies, recognizing that not everything needs to be produced domestically when trusted neighbors can provide redundancy and cost efficiency.

Mexico would absorb highly skilled manufacturing operations, moving beyond simple assembly to more sophisticated production of AI hardware components, servers, networking equipment, and semiconductor packaging and testing. This leverages Mexico's existing manufacturing ecosystem while pushing it toward higher-value activities. The proximity to the United States reduces logistics costs compared to Asian alternatives, and the USMCA framework provides regulatory predictability.

Brazil and Chile could expand their industrial bases to become reliable computational allies within the Western Hemisphere. Brazil's scale allows it to host significant data center capacity serving South American markets and potentially functioning as a computational hub for Portuguese-speaking regions globally. Chile's political stability and trade relationships position it as a trustworthy location for infrastructure serving both South American and trans-Pacific data flows.

This division of labor wouldn't eliminate all dependencies, but it would create redundancy and resilience. If one node faces disruption, whether from natural disaster, political instability, or deliberate interference, alternatives exist within the hemisphere. No single point of failure could paralyze the entire system.

Digital Sovereignty and Colocation

Embedded in these infrastructure questions is a deeper issue: digital sovereignty, the ability of nations to control their technological destinies rather than becoming entirely dependent on foreign platforms and capabilities.

Colocation facilities, where multiple organizations house their computing infrastructure in shared data centers, support trade, logistics, financial services, and increasingly AI workloads. Their location determines data residency, regulatory jurisdiction, and access speeds, all of which matter for both commercial and sovereignty reasons.

North America has a vested interest in ensuring that Central and South American nations can access sophisticated digital infrastructure without depending on Chinese or other non-hemispheric providers. If Brazil, Argentina, Colombia, and other significant economies build their digital infrastructure primarily on Chinese technology and platforms, it creates both commercial advantages for Chinese companies and potential intelligence opportunities for Beijing.

But the United States and its neighbors haven't always offered competitive alternatives. Chinese companies have provided patient capital, comprehensive technology packages, and fewer political conditions. If the Americas are serious about hemispheric digital integration, they need to match or exceed what China offers in technology, financing, training, and partnership terms.

Digital sovereignty doesn't require complete self-sufficiency. It requires having genuine options, the ability to switch providers or develop alternatives if a critical supplier becomes unreliable. For smaller Latin American nations, this means building relationships with multiple providers and maintaining some domestic technical capability even if full independence isn't achievable.

What Comes Next

The physical foundations of AI, supply chains and infrastructure, determine what's possible at every other level. Without advanced semiconductors, you cannot build cutting-edge AI systems. Without adequate power generation and cooling capacity, you cannot operate the data centers that train and deploy those systems.

The Americas possess the raw ingredients for genuinely competitive AI infrastructure: advanced research capabilities, manufacturing expertise, abundant natural resources, growing markets, and geographic proximity that reduces logistics complexity. What the hemisphere has lacked is coordinated strategic vision and the massive investment required to build out capacity at the necessary scale.

That's beginning to change. The United States has committed substantial resources through CHIPS Act funding and other initiatives. Mexico is making strategic bets on computational infrastructure. Brazil is asserting its ambitions for digital sovereignty. But these remain fragmented efforts, individual nations pursuing their own priorities without necessarily aligning into a coherent hemispheric strategy.

The risk is that this window for action closes faster than policymakers realize. Supply chain realignment takes years. Building semiconductor fabs requires three to five years from groundbreaking to production. Developing the specialized workforce for advanced manufacturing demands educational investments that take a generation to fully mature.

All of this infrastructure demands something we've treated as secondary to this point but that actually constrains everything else: energy and natural resources. You cannot run data centers without massive amounts of electricity. You cannot build advanced chips without specialized materials. You cannot construct the physical infrastructure of AI without mining, refining, and processing resources on an unprecedented scale.

Next week, we examine the resource foundations that make AI possible and ask which nations in the Americas are best positioned to supply what the AI revolution demands.

This is Part 4 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: Natural Resources & Energy Demand