AI Impact on Latin America Environment

I have been asked to contribute some reflections on the environmental problems associated with artificial intelligence (AI) technologies. I would like to focus them, essentially, on the Latin American (LATAM) context, from which I write and conduct my professional work.

The advances in AI have undoubtedly made a global impact today, one that does not escape the power structures that have been in place in the West for centuries. A recent study by Lehdonvirta, Wú, and Hawkins¹ shows that GPU cloud computing capacity is essentially divided into three regions: “Compute North,” “Compute South,” and “Compute Desert.” The first includes countries that host AI computing relevant for development and research, where we find the Global North. The second includes countries where AI computing is more relevant for deployment and encompasses the Global South. Finally, the third groups countries that host no public AI computing clouds. LATAM is in the “Compute South,” with GPU-enabled cloud regions in Brazil (5), Mexico (2), Chile (2), Argentina (1), and Peru (1). However, as the study indicates, none of these cloud-based GPUs are more powerful than the NVIDIA Tesla V100, meaning they do not surpass the more advanced chips available in the U.S. or China.

In the mid-1990s, people talked about a “knowledge economy” that did not need a physical component. However, it is now widely agreed that physical infrastructure like chip factories, data centers, and power grids is essential for a country’s competitiveness, regulatory independence, and political power. This leads to a paradox: while Big Tech companies from the Global North dominate the AI power landscape, one of its preferred locations for establishing physical facilities is the Global South, particularly LATAM. The inequity replicates and reinforces the region’s historical role as a resource provider, raising critical questions about sovereignty and environmental justice.

This geopolitical paradox has severe local consequences. Big Tech AI companies are rapidly building data centers in the region, drawn by factors such as affordable energy and strategic locations, with Brazil and Chile emerging as top destinations². However, this expansion is increasingly concentrated in highly water-stressed areas, including central Chile, northern Mexico, and Uruguay, which recently faced a historic drought. The core environmental conflict lies in the massive non-negotiable resource consumption of these facilities. Data centers require immense amounts of electricity for computation and withdraw millions of liters of water daily for cooling systems. In the common evaporative cooling process, the majority of this water—approximately 80%— is consumed through evaporation and not returned to the local watershed³. The remainder is discharged as wastewater. This massive withdrawal, regardless of the consumption rate, places them in direct competition with local communities for a vital and scarce resource⁴.

This vital resource conflict is quantified by astonishing numbers: a typical 1-megawatt data center can consume over 25.5 million liters of water annually just for cooling⁵. The scale of consumption in LATAM is particularly acute, as seen in Uruguay, where Google’s initial plan for a facility in Canelones was projected to use approximately 7.6 million liters of potable water per day during a historic drought⁶. This volume is equivalent to the domestic water use of about 55,000 people. Such disparity underscores how the “Compute South” bears a disproportionate burden of the AI industry’s physical footprint, directly competing with human needs for a scarce resource.

The rapid expansion of data centers in LATAM is actively encouraged by national governments, eager to attract billions in foreign investment, implement incentives based on reduction of taxes and other advantages for foreign investments. For instance, Brazil has a “powershoring” strategy, promoting its green energy matrix to attract investment⁷. Argentina has a recent “RIGI” law (Spanish abbreviation of Incentive Regime for Large Investments), that offers foreign investors 30-year tax exemptions and protection from legal disputes⁸.

Moreover, the resulting social and environmental tensions are exacerbated by an extensive lack of transparency and regulatory rollbacks that prevent communities from understanding or contesting these projects. In Uruguay, the government initially refused to disclose how much water a new Google data center would use during a severe drought, citing corporate secrecy, forcing citizens to go to court to invoke the regional Escazú Agreement on environmental transparency to obtain the information⁹. Recently, in Chile, the government has deregulated environmental assessments for data centers by administratively changing the thresholds for evaluation, effectively exempting new projects from scrutiny and leaving communities without information about their impacts¹⁰. This dynamic reinforces the region’s role as a supplier of resources (land, water, energy) for foreign capital, with local communities bearing the social and environmental costs.

This massive deployment of data centers in LATAM is a contemporary expression of a historical power dynamic, colonialism, now understood through the lens of technology. This techno-colonialism framework reveals how digital innovation and AI practices can entrench power asymmetries between the Global North and South, effectively engendering new forms of violence and inequity¹¹. Just as historical colonial powers exploited raw materials, today’s digital empires extract critical resources, namely land for infrastructure, cheap green energy, and vast quantities of water, to fuel their computational sovereignty—the power to control and benefit from the infrastructure of the digital age. The region’s position in the “Compute South,” where it hosts deployment infrastructure but lacks the advanced chips for development and research, reinforces this subordinate role. Thus, the incentive regimes offered by national governments often function less as neutral economic policy and more as mechanisms of this new dependency, trading long-term sovereignty and ecological security for short-term foreign capital.

Now, to fully comprehend this AI challenge, I think a process-relational worldview¹² offers a crucial lens. This perspective understands social, economic, and ecological reality as a dynamic web of interconnected events and relationships, in which entities—human and non-human—are defined by their participation in complex networks. Within this framework, AI systems are non-human participants. Their algorithms, data centers, and supply chains are deeply embedded within social¹³ and ecological¹⁴ systems, actively reshaping the very networks they join¹⁵

The critical task, therefore, is to examine how these technological artifacts, as relational actors, recalibrate existing networks. They introduce new forms of dependency, redirect flows of capital and resources, and re-engineer social priorities. Consequently, the insertion of AI systems as a powerful node forces LATAM to confront fundamental questions about local reconfiguration—between communities and their water, between national policy and foreign capital, and between present needs and speculative futures: Who benefits (and how)? Who is harmed (and how)? Is it possible to establish balanced negotiations? Is it truly possible to implement technology and innovation within a relationally fair framework?

In this landscape of asymmetrical power, Christianity offers essential conceptual resources to reframe the development and governance of technology not as a barrier to progress, but as a source of wisdom for a more just and sustainable model of technological innovation. Central to this is the biblical principle of stewardship (Genesis 2:15), which frames creation not as a commodity for exploitation but as a sacred trust to be cultivated with care. This stewardship is inherently directed toward the “common good”: a theological principle demanding that technological development prioritizes human dignity, community welfare, and ecological integrity over private profit or unchecked expansion¹⁶.

Furthermore, a Christian anthropology grounded in the belief that humans are relational beings made in the image of a relational God must insist that technology foster authentic human connection and relational flourishing. From this integrated perspective, the church’s advocacy transforms its role. Beyond defending vulnerable communities and their ecosystems by providing moral clarity, practical solidarity, and a platform for marginalized voices, i.e., reframe local resistance, its role also opens to include championing models of “Open Science”¹⁷ —transparent, collaborative, and publicly accountable research— and local digital sovereignty that allow communities to participate in technological development according to their own contexts and needs, rather than being passive recipients of externally designed systems¹⁸. By advocating for these principles, the church and its advocacy for local resistance are not obstacles to development. Instead the church can be an agent of change to articulate an alternative vision: one where AI serves to deepen rather than disrupt humanity’s fundamental relationships with God, neighbor, and the natural world.

From a deeper philosophical perspective, the process-relational analysis of AI insertion in Latin America converges on a Christian theological imperative that is both critical and hopeful. This faith offers not a retreat from technology but a relational wisdom to guide it. This wisdom flows from a foundational belief in a relational God who created a good, interconnected world entrusted to humans. The Christian church’s role is to advocate for technological development that heals rather than severs connections, aligning with the biblical vision of universal flourishing. Moreover, the process-relational insight that we are fundamentally interconnected beings reinforces the Christian call to love our neighbor, where “neighbor” now includes the integrity of local ecosystems affected by server farms and the dignity of communities marginalized by opaque AI systems. Therefore, a Christianity up to date with the AI era can embrace a hopeful, participatory stance.

References:

1. V. Lehdonvirta, B. Wú, and Z. Hawkins, “Compute North vs. Compute South: The Uneven Possibilities of Compute-Based AI Governance around the Globe,” Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society 7, no. 1 (2024).
2. B. Montgomery. Datacenters meet resistance over environmental concerns as AI boom spreads in Latin America. The Guardian, 10 Nov 2025. Available on: https://www.theguardian.com/technology/2025/nov/10/data-centers-latin-america.
3. P. Li, J. Yang, M.A. Islam, and S. Ren. Making AI Less ‘Thirsty’. Association for Computing Machinery 68(7), 2025. https://doi.org/10.1145/3724499.
4. K. Hao, Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI. New York: Penguin Press, 2025.
5. D. Mytton. Data centre water consumption. Nature Partner Journals (npj) Clean Water 4, 11 (2021). Available on: https://doi.org/10.1038/s41545-021-00101-w.
6. N. Ammachchi. Water Scarcity Jeopardizes Data Center Projects in Chile and Uruguay. Nearshore Americas, 16 Nov 2023. Available on: https://nearshoreamericas.com/water-scarcity-jeopardizes-data-center-projects-in-chile-and-uruguay/.
7. M. Osakabe. Despite its advantages, Brazil still struggles to lure green industry. International Valor – Globo, 27 Nov 2025. Available on: https://valorinternational.globo.com/business/news/2025/11/27/despite-its-advantages-brazil-still-struggles-to-lure-green-industry.ghtml.
8. Bruchou & Funes de Rioja. Regulation of the Incentive Regime for Large Investments (“RIGI”), 27 Ago 2024. Available on: https://bruchoufunes.com/en/regulation-of-the-incentive-regime-for-large-investments-rigi/.
9. C. Méndez. Caso Google: Tribunal de Apelaciones reafirmó sentencia y MA deberá dar información sobre volumen de agua que utilizará el data center. La diaria, 1 Mar 2023. Available on: https://ladiaria.com.uy/ambiente/articulo/2023/3/caso-google-tribunal-de-apelaciones-reafirmo-sentencia-y-ma-debera-dar-informacion-sobre-volumen-de-agua-que-utilizara-el-data-center/
10. F. Skoknic and G. Pizarro. Alfombra roja para data centers: sin evaluación ambiental pero con mapa para invertir. LaBot, 12 Sep 2025. Available on: https://www.labot.cl/alfombra-roja-para-data-centers-sin-evaluacion-ambiental-pero-con-mapa-para-invertir/.
11. M. Madianou. Technocolonialism: When Technology for Good is Harmful. Cambridge & Hoboken: Polity Press, 2025.
12. “Process Philosophy”, Stanford Encyclopedia of Philosophy (First published Mon Oct 15, 2012, substantive revision Thu May 26, 2022). Available on: https://plato.stanford.edu/entries/process-philosophy/.
13. T. Correa, F. Luco, M. Humeres, D. Cotoras, A. Davidoff, et al. Reimagining AI in Latin America: situated narratives of users, developers, and decision-makers on understanding and governing AI. Commun. Change 1, 12 (2025). https://doi.org/10.1007/s44382-025-00012-1.
14. A. Lebdioui, A. Melguizo, and V. Muñoz. AI, Biodiversity and Energy in Latin America and the Caribbean: From a Resource Intensive to Symbiotic Tech. United Nations Development Programme in Latin America & the Caribbean (UNDP LAC), Working Paper, 25 Aug 2025. Available on: https://www.undp.org/latin-america/publications/ai-biodiversity-and-energy-latin-america-and-caribbean-resource-intensive-symbiotic-tech.
15. V. Varriale, A. Cammarano, F. Michelino, and M. Caputo. Artificial intelligence in technology networks: A catalyst for achieving the SDGs. Technovation 151, Mar 2026, 103398. https://doi.org/10.1016/j.technovation.2025.103398.
16. Q. McGRath. AI Ethical Framework. In: Laussane Global Analysis: AI Ethics & The Great Comission. November 2025. Available on: https://lausanne.org/global-analysis/ai-ethical-framework.
17. UNESCO Recommendation on Open Science: https://www.unesco.org/en/open-science/about?hub=686.
18. J. Williams. Elinor Ostrom’s 8 rules for managing the commons. The Earthbound Report, 15 Jan 2015. https://earthbound.report/2018/01/15/elinor-ostroms-8-rules-for-managing-the-commons/.

Views and opinions expressed by authors and editors are their own and do not necessarily reflect the view of AI and Faith or any of its leadership.