My topic is the environmental impact of artificial intelligence (AI) data centers, specifically their consumption of water and their expansion onto land that could otherwise be used for agriculture. AI may seem like something that exists entirely online, but it depends on large physical facilities filled with servers. These facilities require enormous amounts of electricity, cooling, water, and land. My goal is to make viewers more aware of the physical environmental costs that are often hidden behind the convenience of AI. One of the most important environmental concerns is water consumption. Data centers generate significant amounts of heat, requiring cooling systems to prevent servers from overheating. A 2021 study by Siddik, Shehabi, and Marston estimated that U.S. data centers had an operational water footprint of approximately 513 billion liters in 2018. Their analysis also found that about three-fourths of this footprint came indirectly from water used to generate the electricity consumed by data centers.¹ This demonstrates that the environmental impact is not limited to the water used directly inside the buildings. The expansion of AI could make this problem substantially larger. In an article written by Tianqi Xiao and colleagues, they modeled the future environmental impacts of AI servers in the United States. They estimated that between 2024 and 2030, AI servers could produce an annual water footprint of approximately 731 million to 1.125 billion cubic meters, depending on how quickly AI infrastructure expands.² Their research also showed that where data centers are located can greatly affect their environmental footprint because different regions have different water and electricity systems. Water use also depends on the cooling technology used. A 2025 life-cycle assessment published in Nature examined different cooling technologies and found that water consumption is connected not only to the cooling system itself but also to electricity generation, server design, and other parts of the data-center infrastructure.³ This means that reducing the environmental impact of AI requires more than simply using less water inside individual facilities. Another major concern is land use. Data centers require large buildings as well as substations, roads, transmission infrastructure, cooling equipment, and other supporting structures. Recent research examining data-center locations in the United States found that land-use structure was one of the strongest factors influencing where data centers are built.⁴ As AI infrastructure expands, companies are increasingly looking for large areas of land that can accommodate these facilities. This creates a particular concern for rural communities and agriculture. When agricultural land is converted into a permanent industrial facility, it can no longer be used for farming in the same way. Recent research examining the geography of U.S. data-center expansion found that planned facilities tend to be located in much less densely populated areas than areas without data centers, meaning that future development is increasingly occurring in rural communities.⁵ This is important because rural land is not simply space. It can provide food production, wildlife habitat, and economic resources for local communities. The current situation shows that AI development is happening faster than many communities have been able to evaluate its environmental consequences. I believe people should understand that AI has a physical footprint. Every time AI infrastructure expands, there are decisions being made about where water comes from, how electricity is generated, and what happens to the land where the facility is built. There are possible solutions. Researchers have demonstrated that improvements in cooling technologies, server efficiency, electricity sources, and the geographic placement of data centers can significantly reduce their environmental impacts.² ³ Communities can also consider land-use policies that prioritize previously developed or industrial land instead of productive agricultural areas. The goal should not necessarily be to stop AI development, but to make sure technological growth does not happen without considering the environmental resources it consumes. I chose this topic because I wanted to represent an environmental issue that is difficult to see in everyday life. My quilt focuses on the contrast between technology and the natural landscape. I want viewers to think about what is physically required to create something that seems as simple as using AI on a computer or phone. I plan to incorporate images of actual data centers into my design. These images emphasize that the issue is real and physical rather than simply digital. Showing the large industrial buildings alongside representations of farmland will help communicate the transformation of the landscape. For my colors, I intentionally used muted and dark tones instead of bright colors. A healthy landscape is often associated with vibrant greens and blues, but I wanted my landscape to appear as though some of its liveliness had been extracted. The muted colors symbolize the depletion of natural resources and the loss of vitality that can occur when land and water are redirected toward industrial development. I also chose a rough, uneven fabric for the land. Instead of making the surface smooth, I wanted the texture to suggest dryness, disturbance, and damaged soil. The physical texture allows the viewer to experience the land as something that has been altered rather than simply seeing it as an image. Finally, I incorporated a found yarn-like material to represent the steam, emissions, and pollution associated with data-center infrastructure. Its irregular appearance suggests something spreading into the surrounding environment. Together, these materials create a visual transition from a living landscape toward an industrial one, showing the environmental costs that can exist behind the growth of AI.

1. Md Abu Bakar Siddik, Arman Shehabi, and Landon Marston, “The Environmental Footprint of Data Centers in the United States,” Environmental Research Letters 16, no. 6 (2021): 064017.

2. Tianqi Xiao et al., “Environmental Impact and Net-Zero Pathways for Sustainable Artificial Intelligence Servers in the USA,” Nature Sustainability 8 (2025): 1541–1553.

3. Christian Belady et al., “Using Life Cycle Assessment to Drive Innovation for Sustainable Cool Clouds,” Nature 641 (2025): 331–338.

4. Seung Jun Choi, Kijin Seong, and Junfeng Jiao, “Investigating Data Center Site Planning of Major ICT Companies in the U.S.: A Spatial and Methodological Framework,” Cities (2026), article 107509.

5. Danae Hernandez-Cortes, Kyle Meng, and Paige Weber, “The Environmental Costs and Geography of U.S. Data Center Expansion,” UC Berkeley Energy Institute Working Paper 363 (2026).