A growing share of new artificial intelligence infrastructure in the United States is being planned in some of the country’s driest regions, raising fresh concerns about water use and environmental sustainability as the global race to build AI capacity accelerates.
A recent analysis by The Guardian has found that roughly two thirds of planned AI data centres are expected to be constructed in areas that have experienced some of the most severe drought conditions over the past year. The findings come at a time when large parts of the US are already grappling with record low water levels, prolonged heatwaves, and mounting pressure on natural resources.
The expansion of AI infrastructure is being driven by surging demand for computing power required to train and run advanced machine learning models. Data centres form the backbone of this ecosystem, housing thousands of servers that process vast amounts of data. However, these facilities require significant amounts of electricity and water, particularly for cooling systems that prevent servers from overheating.
Water is typically used in evaporative cooling processes, which help maintain optimal operating temperatures in high performance computing environments. In regions facing water scarcity, this has sparked debate about whether the rapid growth of AI infrastructure is sustainable.
Environmental advocates have warned that building water intensive facilities in drought prone areas could worsen existing shortages and strain already fragile ecosystems. Some critics argue that the AI industry is expanding faster than regulators can assess its environmental impact, particularly in regions where water is already a limited resource.
The issue highlights a growing tension between technological progress and environmental responsibility. On one hand, AI is expected to drive economic growth, innovation, and productivity across multiple sectors. On the other hand, the infrastructure required to support it carries significant environmental costs.
The choice of location for data centres is often influenced by factors such as land availability, electricity prices, tax incentives, and proximity to energy sources. Many of the driest regions in the US, particularly in the southwest, offer large expanses of land and access to renewable energy sources such as solar power, making them attractive to tech companies.
However, these same regions are among the most vulnerable to water stress. States like Arizona, Nevada, and parts of Texas have seen declining water reserves due to prolonged drought conditions, increased demand, and the effects of climate change.
Tech companies have begun responding to these concerns by exploring alternative cooling technologies that use less water. Some firms are investing in air cooled systems or closed loop water recycling systems designed to reduce consumption. Others are experimenting with locating data centres in cooler climates where less cooling is required.
Despite these efforts, experts say the scale of AI expansion means that total resource consumption is likely to continue rising. The rapid deployment of new facilities, combined with increasing computing demands, could offset efficiency gains achieved through improved technology.
The debate also raises broader questions about how infrastructure for emerging technologies should be regulated. Policymakers are under pressure to balance economic growth with environmental protection, particularly as industries like AI become more central to national competitiveness.
In the US, local communities have started pushing back against large scale data centre projects, citing concerns over water use, energy demand, and land use. Some municipalities are calling for stricter environmental assessments and greater transparency from companies seeking to build new facilities.
At the same time, governments are keen to attract investment from tech companies, which often bring jobs, infrastructure development, and economic activity. This creates a complex policy environment where competing priorities must be carefully managed.
Globally, the issue is not limited to the United States. As countries across Europe, Asia, and Africa invest in AI infrastructure, similar concerns are emerging about energy consumption, carbon emissions, and resource use.
The findings from The Guardian analysis serve as a reminder that the AI revolution is not purely digital. It relies heavily on physical infrastructure that consumes real world resources, often in significant quantities.
As the industry continues to grow, the challenge will be finding ways to balance innovation with sustainability. Whether through technological advancements, regulatory frameworks, or changes in location strategy, the long term viability of AI infrastructure will depend on how effectively these competing demands are addressed.
For now, the concentration of new data centres in drought affected areas has intensified scrutiny of the industry’s environmental footprint, adding another layer of complexity to the global push toward an AI driven future.
