Written by : Adrienne Calzada
Edited by: Gabrielle Adams
Artificial intelligence (AI) is widely described as the next technological frontier, promising efficiency, economic growth, and innovation. However, tools such as Large Language Models (LLMs) and generative AI consume significant amounts of energy and water, with global AI demand in 2027 being projected to account for 4.2–6.6 trillion litres of water. The environmental footprint of these tools is immense, raising carbon emissions, electronic waste and straining local water resources. Yet, as tech companies and world leaders continue to push for AI under the banner of human progress, affected communities are increasingly resisting AI expansion amid resource depletion, whilst companies are raking in billions in profits, prompting a reevaluation of whether the benefits of AI truly outweigh their costs.
Instant output is extremely energy-intensive, with some data centers consuming as much electricity as small cities, mounting pressure on national grids and the communities from which they extract resources. Indeed, AI systems consist of data center networks in which supercomputers operate at extreme capacity; these facilities are what allow AI programs to operate uninterruptedly as they support model training and real-time processing. A typical data center uses approximately 300,000 gallons of water each day, and a large one can use an estimated 5 million gallons of water daily. To put that into context, that is equivalent to water usage in a town of 50,000 residents. Adding to this, where renewable capacity is limited, the demand reinforces dependence on fossil fuels, further embedding carbon-intensive infrastructure into local economies. Moreover, data centers require intensive air cooling to maintain stable temperatures in these networks that, otherwise, generate significant heat. This process requires large and continuous water inputs, taken from local supplies, and scarcely returned into the provisions. In water-stressed regions, this demand competes directly with those of drinking water and agricultural production, rendering AI centers immense material and environmental burdens to the communities they establish in.
The expansion of AI infrastructure is dictated by corporate influence rather than its host communities. These facilities are often located in rural areas, characterized by aging populations, economic marginalization, and lower levels of political power and education. These rural host communities are often forced to negotiate with billion-dollar technology corporations backed by legal teams, financial leverage, and political lobbying networks that far exceed local capacity. These conditions create structural power imbalances between the multinational technology firms and local communities, making it easier for corporations to secure land and water-intensive projects with disregard for local resistance. Despite this asymmetry, local communities push back as their livelihoods are threatened; protests have emerged in India, Uruguay, Chile, and Spain, where residents are arguing that digital infrastructure is being prioritized over local needs.
Spain’s Aragón region offers a powerful example where proposed data center projects have triggered fears of water scarcity. Activist Aurora Gómez leads a campaign known as “Tu Nube Seca Mi Río” (“Your cloud is drying my river”), where her movement seeks to bring attention to the ecosocial impact of data centers. For example, the group is contesting Amazon’s promises to use AI as a tool for water efficiency in agricultural production, which they say are “a deliberate strategy of obfuscation” designed to distract from the company’s intention to increase its water consumption in Aragon. Gómez’s campaign highlights a growing distrust of corporate sustainability narratives and reflects broader tensions between technological expansion and local environmental security.
The efforts made by corporations in establishing their data centers have been described as a form of data-colonialism; profits and resources are extracted from host communities by global firms while the former absorb environmental and material costs without meaningful participation or decision-making ability. The establishment of AI data centers thus reproduces historical patterns of environmental injustice and subjugation, concentrating ecological burdens in communities that bear disproportionate environmental costs relative to the benefits they receive. As such, environmental risk continues to be pushed onto vulnerable populations under the guise of innovation, while tech giants continue to line their pockets with profit. If AI technologies are to constitute genuine progress rather than reflect vested profit interests, technological governance must actively disrupt this predatory cycle rather than enable it under the pretense of progress.
AI has the potential to improve human life in unprecedented ways, from optimizing energy systems to accelerating climate research. However, these benefits can only be realized if AI infrastructure is deployed democratically, carefully and equitably. Scaling data center capacity without strict transparency around energy and water consumption risks institutionalizing a new form of digital extractivism. As regions such as the European Union move to triple data center capacity by 2030 through the AI Continent Action Plan, techno-ecological frameworks become essential for ensuring that technological expansion remains compatible with environmental limits and the local communities in which they are established. Local communities must also be treated as stakeholders and active participants. AI cannot continue to advance at the expense of human well-being; technology companies must thus foreground the material consequences of their infrastructure instead of obscuring them behind narratives of advancement. Progress and sustainability are not mutually exclusive, but they do require regulation, transparency and social responsibility.