/

The Geopolitics of AI and the U.S. Project of Domination

Popular movements have argued that the development of AI, as social infrastructure, must be subordinate to a project that guarantees rights and a dignified life—not to increased surveillance and corporate power.

7 mins read
OpenAI CEO Sam Altman talks with Meta CEO Mark Zuckerberg

We are living in an era of blatant manipulation, interference, and imperialist aggression against Latin America, and the Brazilian elections were a manifestation of this. At the same time, we are facing both the far right and the empire, with its tech giants. The people’s victory in Brazil had the defense of sovereignty as one of its main pillars: the people affirmed that we are no one’s backyard, understood the implications of Meta’s interference in electoral processes, and more people came to understand the methods of imperialism. There is no doubt that our sovereignty necessarily involves the digital realm, and Lula, having been reelected, is very focused on this agenda.

It is more than evident that Artificial Intelligence (AI) is one of the central issues in contemporary geopolitics; the United States responds with tariffs and threats to every attempt to build any level of technological autonomy aimed at digital sovereignty.

At this point, fortunately, the debate no longer revolves around whether or not we will use AI. It is about how to navigate this shift into a new technological era, in which data is a new factor of production and digital transformation extends to the state and to life as a whole. How can we build a strategic project capable of breaking the chains of subjugation and dependence that structure our region?

The Geopolitics of AI

Artificial intelligence is not just any technology. We must understand its scope as a general infrastructure, much like electricity.

The United States’ strategy to dominate artificial intelligence is made explicit in analyses of the day-to-day workings of U.S. policy: from documents to statements, from diplomacy to war, encompassing the fusion between the state, the military-industrial complex, and major technology companies, as well as their public and private investments.

In Latin America, this analysis must take into account the scope of the U.S. National Security Strategy, which aims to control assets that are strategically vital to the United States. To that end, the empire needs to create obstacles in the economic and technological relations between countries in the region and China.

Everything is happening very quickly. In less than a year, the Monroe Doctrine has materialized in the combination of the Shield of the Americas and Pax Silica, a U.S. corporate diplomacy initiative to secure its access to the entire artificial intelligence supply chain. Argentina, Chile, Costa Rica, El Salvador, and Panama are participating in both initiatives, as part of a far-right alignment in the service of the empire. It is clear that the electoral contest in Brazil is part of this international balance of power.

On 22 September  of this year, the ‘Joint Declaration on the Defense of Sovereignty in the Hemisphere’—to which Colombia and other countries in the region are also signatories—outlines the convergence of both initiatives. In the name of the fight against ‘narco-terrorism,’ the signatories hand over intelligence, critical minerals, and digital infrastructure to Washington’s coordination efforts, supporting the policy of unilateral sanctions defined by the U.S.

U.S. documents indicate the goal of promoting the adoption of U.S. AI systems, hardware, and standards worldwide. They aim to export the entire AI technology stack. This means imposing a comprehensive model, organized in layers that support one another. This stack cannot function without a foundation: the territories where the critical minerals of the digital age (lithium, copper, niobium, rare earth elements), energy, and water are found.

The other material layer consists of chips and hardware—the components and servers where these tools are trained and where AI models are run. According to the U.S. standard, this refers to Nvidia’s famous GPUs (which not all countries can acquire). The software runs on top of the hardware. In this case, CUDA—a proprietary programming platform that ties AI training to Nvidia’s ecosystem—runs on Nvidia hardware. Next come the data centers and cloud platforms (from Amazon, Microsoft, and Google), which contain a vast number of servers where the extracted data is stored and processed.

There are also data centers that train AI models, and these are the ones that consume the most water and energy. Above these layers are the AI models (such as large language models) and the applications that reach users. All of this, let’s remember, depends on gigantic amounts of data—digital records of nearly everything that is alive and everything humanity has produced.

Therefore, the United States export of its AI dependency package depends on the extraction of data from the entire population and the natural resources of Latin American countries—which are responsible for sustaining the foundation of the chain—and in return, they provide us with predefined AI models. This is the model of the Pax Silica, the agreement that guarantees what the White House needs to counter China’s rise, while maintaining Latin America’s subjugation as a supplier of minerals and territories.

Paths to Sovereignty

In Brazil, the creation of conditions for sovereignty began through mining-related policies, and not necessarily in the digital sphere.

Brazil has the world’s second-largest reserve of rare earth elements, but accounts for a tiny fraction of processing. The technology for processing and refining is concentrated almost entirely in China, which also holds the largest reserve of rare earth elements protected by a sovereign policy. The United States depends on Brazilian rare earths and other critical minerals, such as lithium from Bolivia and Argentina, to compete in the AI race.

Brazil’s National Policy on Critical Minerals, approved in September 2026, provides for investments to process and refine rare earths domestically and establishes a council—linked to the Office of the President—with the power to veto transfers of control over mining companies. This is a response to the acquisition of the Serra Verde mining company in Goiás by USA Rare Earth, with the backing of the U.S. government. Shortly before that, in August, the Brazilian government had announced the development of a sovereign cloud and AI.

The Lula administration’s advances in the area of artificial intelligence sovereignty stem from intensifying geopolitical dynamics. Clearly, there are disputes within the government itself over which path to follow; for a time, a passive and even covert view of sovereignty regarding this technology prevailed, advocating the notion that simply hosting data on national territory would be sufficient for our digital sovereignty. It is worth recalling that there is a law in the U.S. (the Cloud Act) stipulating that data processed in the clouds of U.S. companies falls under that country’s jurisdiction.

We start from a position of deep dependence. The Brazilian government has spent billions on digital licenses and services—primarily with Microsoft, Google, and AWS—rather than investing in developing its own capabilities. And that is why the announcements made in August of this year are so important: a 2.3 billion reais investment package in sovereign AI infrastructure, supercomputers, and the development of AI models in Portuguese, in addition to an agreement to establish a 100 percent national cloud. The Brazilian government has finally stated that the development of these technologies must be based on models in which the code belongs to the nation itself.

Clearly, these are important steps in a long process fraught with limitations and contradictions. Precisely for this reason, they must be integrated into a strategic project for development, industrialization, and popular sovereignty.

Given the pace at which AI is advancing, it does not seem feasible to start from scratch. The race to develop models is becoming increasingly evident due to the progress of China’s open-source models, which also extend to the hardware sector. It is essential to establish the terms of partnerships in AI development, situating each one within the framework of the strategic project to achieve a viable path toward sovereignty. In the case of the supercomputer, the government announced partnerships with the Chinese companies iFlytek and Huawei (which have been sanctioned by the U.S.), establishing terms for technology transfer, the training of 3,000 professionals, and the capacity to lead model development. Obviously, this measure runs counter to U.S. objectives.

On the path toward building digital sovereignty, Brazil asserts that we need AI models in Portuguese, reflecting the diversity of accents and forms of expression. We already have some, trained using ChatGPT, for example. The imposition of foreign AI leads to homogenization—a cognitive standardization that shapes what questions are asked, how answers are structured, and how images are generated when reconstructing memory and recognition. This also encompasses the debate over what constitutes valid knowledge, training data, the human labor crystallized within AI, and the continuous extraction of data that becomes a means of behavior modulation.

An AI trained in Portuguese is a step forward. But it is necessary to channel the sovereign development of AI within the framework of a democratic and popular project for the country. To achieve sovereignty in the realm of AI (and in the digital age), data must be treated as a public good. If we consider data as a factor of production and recognize its central importance to the new quality of the productive forces, ensuring data sovereignty and control is tied to the struggle over the direction of our future as a people—and over our ways of producing and reproducing life.

And in that regard, we still have a long way to go. We need to build the capacity to manage data from local communities in a way that serves its people: data related to the healthcare system, biodiversity, the fight against deforestation, or preparedness for the effects of El Niño; agricultural data that contributes to food sovereignty; as well as data on cities and urban mobility.

There is also data collected by different government sectors that we still lack the capacity to process and integrate. For example, there has never been so much discussion about violence against women. One technology used as a solution is an electronic ankle monitor for the abuser and an alert system for women, so they know when the abuser is approaching. This keeps women in a constant state of alert and ultimately places the responsibility for protecting themselves squarely on their shoulders. However, the Brazilian government still lacks the capacity to use AI to integrate health and social assistance data in order to develop predictive capabilities that could guide, for example, more preventive policies against violence. In every sector, we can think of examples of what is needed and what can be done, as well as what must be regulated and combated.

A sovereign AI is only possible with a democratic and sovereign state. And it must be subject to popular control over what to develop, with what data, and to address what needs of the population: from education to health, from the production of healthy food to ensuring a life free from violence, from work with rights to the right to rest and leisure.

Popular movements have argued that the development of AI, as social infrastructure, must be subordinate to a project that guarantees rights and a dignified life—not to increased surveillance and corporate power. On the other hand, fascism—intrinsically misogynistic and racist—gains strength through the AI of the Empire’s major tech corporations.

That is the struggle currently being waged in Brazil. Precisely because this is such a unique historical period, the United States will use all its tools to try to subjugate us, just as it does with countries in the region currently governed by the far right or with countries devastated by war. Our historic task is to turn the defense of digital sovereignty into a popular struggle, with concrete advances; to develop the capacity to break free from dependence; and to forge sovereignty at this very moment when the foundations of the new technological era are still up for grabs.

Source: Globetrotter

Tica Moreno

Tica Moreno is a member of the national coordinating committee of the World March of Women in Brazil.

Leave a Reply

Your email address will not be published.

Latest from Blog