Laudato Si and AI

AI requires enormous quantities of electricity, water, and rare-earth minerals. Pope Francis's 2015 encyclical offers the framework Catholic teaching now has to apply to one of the most resource-intensive technologies humans have ever built.

The data center that hosts the AI model answering your question this morning consumes more electricity in a day than a small town. The cooling system that keeps the chips from melting evaporates fresh water at a rate comparable to a municipal water utility. The rare-earth minerals that went into the hardware were extracted, often in conditions that would not pass Catholic moral scrutiny, in mines in central Africa, in Latin America, and in southeast Asia. The carbon emissions from the electricity that runs the whole system are higher than the carbon emissions of the airlines you fly on.

None of this shows up in the marketing copy for AI products. The standard framing of AI as an immaterial, weightless software service obscures the substantial industrial infrastructure required to make it work. Catholic teaching has a framework for this kind of obscuring, and it is recent enough to still be working through its implications. The framework is Laudato Si, Pope Francis's 2015 encyclical on the care of our common home, and the integral ecology it articulates is exactly what the AI moment needs.

What Laudato Si actually says

Laudato Si was released in May 2015 and is the most substantial Catholic teaching document on the environment ever produced. It is also a document that many readers know by reputation without having engaged its actual argument. A brief summary is useful before applying it to AI.

The encyclical's title means "Praise be to you," from the opening words of Saint Francis of Assisi's Canticle of the Creatures. Pope Francis took the saint's name and used the encyclical to position care for creation as a substantial Catholic theological theme rather than a secular concern that the Church engages from outside. The document is addressed to "every person living on this planet," and its argument is meant to be accessible to non-Catholic readers without losing its Catholic theological grounding.

Three substantive claims do most of the analytical work.

The earth and the poor cry out together. This is the encyclical's central move. Pope Francis insists that environmental harm and harm to the poor are not two separate problems that happen to be correlated. They are aspects of a single crisis driven by a common logic, what the encyclical calls "the technocratic paradigm." The paradigm treats both the natural world and vulnerable human beings as instruments to be exploited rather than realities to be respected. The earth crying out and the poor crying out are the same cry.

Integral ecology. The encyclical introduces the framework of integral ecology, which insists that ecological, social, economic, and cultural questions cannot be addressed in isolation. A policy that protects an ecosystem while displacing poor communities is not, in this framework, ecologically good. A technology that addresses climate change while concentrating power in the hands of a few is not, in this framework, ecologically defensible. The framework is demanding: it refuses the easy moves by which environmental thinking sometimes evades social questions, and by which social thinking sometimes evades environmental questions.

The critique of the technocratic paradigm. Pope Francis devotes substantial sections of the encyclical to a critique of what he calls the technocratic paradigm, the assumption that human problems are essentially technical problems to be solved by the application of technology, and that nature is essentially raw material to be transformed by human intervention. The encyclical does not reject technology. It rejects the totalizing version of the technocratic paradigm that treats every problem as a technical problem and every limit as something to be overcome.

These three claims sit at the heart of Laudato Si and they apply with full force to AI.

The actual environmental footprint of AI in 2026

Before applying the framework, the facts. The numbers below are drawn from peer-reviewed research and from the International Energy Agency's 2025 reporting.

Electricity. Global data centers consumed about 415 terawatt-hours of electricity in 2024, roughly 1.5 percent of total global electricity consumption. The International Energy Agency projects this to grow to about 945 terawatt-hours by 2030, with AI identified as the main driver. In the United States alone, data centers accounted for over 4 percent of national electricity consumption in 2024, up from about 1.9 percent in 2018, and projections reach as high as 12 percent by 2028. The carbon footprint of AI systems specifically has been estimated at 32 to 80 million tons of CO2 emissions in 2025, comparable to the annual emissions of a large city.

Water. Data centers require substantial cooling. The cooling is typically done with water, much of which evaporates and does not return to the local watershed. Global water consumption by AI systems in 2025 has been estimated at 312 to 765 billion liters annually. Training a single large language model in a typical data center can directly evaporate hundreds of thousands of liters of fresh water. Texas, where many of the largest new AI data centers are being built, projects data center water use of 49 billion gallons in 2025 rising to as much as 399 billion gallons by 2030. The water use is geographically concentrated in regions that often already face water stress.

Hardware and minerals. The chips, servers, networking equipment, and cooling infrastructure required for AI data centers depend on substantial inputs of copper, aluminum, rare-earth metals, cobalt, lithium, and other materials. Many of these are sourced from mines in central Africa, Latin America, and southeast Asia where extraction conditions raise serious labor and environmental concerns. The Catholic Church has spoken directly to mining ethics in multiple contexts, including a 2019 Vatican-hosted gathering on mining and the common good. The AI hardware supply chain inherits all of those concerns.

Land and community impact. AI data centers are physically large facilities. They require significant land, often agricultural or natural land, and they bring construction, traffic, noise, and ongoing operational impacts to the communities that host them. Northern Virginia, which already hosts the densest concentration of data centers in the world with about 300 facilities, faces ongoing debates about whether more should be permitted. Communities elsewhere are facing similar decisions, often with limited information about what they are agreeing to.

None of these numbers is contested by mainstream sources. The disagreements are about projections and methodology, not about the basic shape of the picture. AI is, by any honest accounting, one of the most resource-intensive technologies humans have ever scaled this quickly.

Applying Laudato Si to AI directly

With the framework and the facts in place, the application becomes straightforward.

The earth and the poor cry out together, in AI as elsewhere. The communities most affected by AI's environmental footprint are not the communities that consume AI services. The data centers are built where land and water are cheap, where labor is non-union, and where regulatory oversight is light. The water that AI evaporates often comes from aquifers and rivers that other communities, often poorer ones, also depend on. The electricity that AI consumes is often generated by power plants located in communities that bear the air-quality costs of generation. The mineral inputs come from extractive industries that have for decades been a source of Catholic concern. The cry of the earth and the cry of the poor, in the AI context, are the same cry, and the Catholic framework refuses to address either without the other.

Integral ecology demands integrated analysis. The standard policy debates about AI tend to compartmentalize: AI's labor effects in one conversation, AI's environmental effects in another, AI's geopolitical effects in a third. Laudato Si's integral ecology insists that these are not separate questions. The same large language model that displaces a creative worker in one country was trained using electricity generated by a coal plant in another country that affects respiratory health in the community around it. The same data center that anchors an AI service strains the water supply of an agricultural community downstream. The integral ecology framework refuses to let any of these be analyzed in isolation, because the harm is happening to the same world.

The technocratic paradigm is exactly what AI represents. Pope Francis's critique of the technocratic paradigm reads, ten years later, as a substantive engagement with what AI is doing now. AI is presented to the public as the solution to a wide range of human problems. The implicit message is that what cannot be addressed by AI will be addressed by the next iteration of AI, or by AI applied at greater scale, or by AI integrated with more domains. The framework is exactly what Pope Francis identified: the assumption that every problem is technical, that every limit is to be overcome, that the application of more technology is always the correct response. Laudato Si does not reject technology. It rejects the totalizing version of this paradigm. Applied to AI, the critique cuts hard.

What Pope Leo XIV brings to the environmental conversation

Pope Leo XIV's environmental sensibility is shaped by his decades in Peru. He served in regions affected by El Niño flooding, by glacial retreat in the Andes, by water scarcity in coastal communities, and by extractive industries that often operated with limited regard for the communities around them. The pastoral feel for environmental questions that Laudato Si articulated theologically, Pope Leo XIV developed by living through it.

His first year as pope has included multiple statements connecting AI to broader concerns about the common good. The first encyclical, Magnifica Humanitas, releasing May 25, 2026, is expected to address AI's environmental costs alongside its other effects, though the specific depth of treatment is not yet known. What can be said in advance is that the encyclical is being written by someone formed in the school of Laudato Si, by someone who served as a missionary in regions that lived the environmental costs of global supply chains, and by someone whose entire pastoral identity treats care of creation and care of the poor as inseparable.

The encyclical may or may not produce a detailed treatment of AI data center water use. It will almost certainly carry forward the integral ecology framework in ways that make AI's environmental costs morally visible. That visibility is itself part of what the encyclical is meant to do.

What the framework requires

Applied to AI, the integral ecology of Laudato Si produces a set of demands that go beyond what most current policy debate addresses.

Transparency about environmental costs. AI companies should disclose, in detail and on a regular basis, the electricity consumption, water use, carbon emissions, and supply-chain impacts of their AI operations. The data should be granular enough to allow informed evaluation by communities, regulators, and customers. Current disclosure is inadequate and the gap is itself a moral problem. The Catholic principle of truth-telling, applied institutionally, supports this demand.

Honest accounting of where costs fall. The communities bearing the water, air-quality, and land-use costs of AI infrastructure are typically not the communities benefiting most from AI services. This asymmetry is part of the technocratic paradigm Laudato Si identified. The Catholic framework requires that the asymmetry be made visible, addressed through compensation where appropriate, and corrected through structural change where the harm is significant.

Investment in the infrastructure that allows scaling without harm. AI does not have to be environmentally catastrophic. Renewable energy, efficient cooling, water-sensitive siting, and grid decarbonization can all substantially reduce the footprint. The investment in this infrastructure is currently lagging the pace of AI deployment, which is a choice the industry and its regulators are making. The Catholic framework would require that the investment match the pace, and that the deployment slow if necessary to allow the infrastructure to catch up.

Refusal of the framing that environmental concerns are secondary. The standard industry response to environmental questions about AI is that the technology is so beneficial that the environmental costs are an acceptable price. The Catholic framework refuses this framing. The costs are real, they fall on identifiable communities and ecosystems, and they are not subsumed in the benefits even when the benefits are genuine. The integral ecology framework requires that the costs be accounted for honestly, not deflected with appeals to the value of the technology.

Solidarity with the communities affected. Catholic teaching has long supported the right of communities affected by industrial development to participate in decisions affecting them, to demand accountability, and to receive fair compensation for impacts they cannot avoid. AI data center development is currently happening with limited community input in many jurisdictions. The Catholic framework supports the standing of those communities to demand more.

What this asks of readers

The framework is most useful when it produces specific orientations. Here are the readers most affected by what this page sets out.

For Catholic users of AI services. The framework does not require you to stop using AI. It does require you to know what the technology costs the world and to refuse the framing that AI is environmentally neutral. The choice to use AI thoughtfully, with awareness of the costs and skepticism of the marketing, is part of what the Catholic framework asks.

For Catholic technologists. Those who design, build, deploy, and operate AI systems carry particular responsibility under integral ecology. The choices about which data centers to build, how to power them, how to cool them, and where to site them are choices with real consequences for real communities and real ecosystems. Catholic engineers and product managers can press on these choices inside the institutions where they are made.

For Catholics in policy and civil society. The regulatory frameworks governing AI's environmental footprint are still being written in most jurisdictions. The EU AI Act addresses environmental impact in a limited way. National regulations are emerging unevenly. Catholic voices have substantial standing on environmental questions, particularly in countries with strong Catholic civic traditions. The voices should be raised on AI specifically, not only on climate generally.

For Catholic institutions. The dioceses, universities, hospitals, charities, and religious orders that are now considering AI tools should consider the environmental footprint as part of the discernment. Catholic institutions have a particular responsibility to use technology in ways consistent with Catholic teaching, including teaching on the care of creation. Laudato Si applies to institutional decisions as much as to individual ones.

For Catholic communities affected by data center development. Catholic parishes and dioceses in regions facing significant data center development have standing to engage local debates about water use, electricity costs, land use, and community impact. The Catholic social teaching tradition supports community organizing, fair participation in decisions, and demands for accountability from developers. These tools are available and should be used.

The 2026 encyclical on AI is being released eleven years after the 2015 encyclical on the environment. The two documents are not separate concerns. They are aspects of the same Catholic engagement with what technology is doing to the human person and to the world the human person inhabits. The integral ecology of Laudato Si is exactly the framework the AI moment requires. The question is whether Catholic readers will use it.

Further reading