IT & AI News 3 min read

Concerns Grow Over AI’s Water Consumption Amid Data Center Expansion

Victoria Sterling

Key Takeaways

  • Protests against data centers are rising in water-scarce regions.
  • Data centers use water for cooling, but their consumption is less than agriculture.
  • Advancements in cooling technology can reduce water usage.
  • Strategic site selection and renewable energy can mitigate water stress.

Protests Emerge Over Water Use

In states like Texas, New Mexico, and Arizona, protests are intensifying against the expansion of data centers. Signs reading “Protect our water” and “Water for people not AI” reflect growing concerns about the environmental impact of these facilities, particularly regarding water consumption and pollution.

The Role of Water in Data Centers

Data centers often rely on water to cool their systems, similar to how sweat cools the human body. The processors in these centers can reach temperatures as high as 176° Fahrenheit, necessitating effective cooling methods. While energy consumption in data centers is well-documented, the specifics of water use remain murky. Online claims vary widely, with some suggesting that chatbots consume significant amounts of water per query, while others argue the issue is overstated. Inconsistent data from tech companies complicates the picture.

Comparative Water Consumption

Experts note that while data centers do consume water, their usage is minor compared to sectors like agriculture. In 2023, it was estimated that cooling systems in data centers used 66 billion liters of water, accounting for less than 1% of total national consumption. However, with the rapid growth of AI data centers, this figure could rise, particularly in drought-prone areas.

Strategies for Reducing Water Strain

Engineers are exploring various strategies to alleviate the pressure on local water resources. Fengqi You, an energy systems expert, highlights the potential for achieving net-zero water usage through improved practices and technologies. Recent advancements in AI have already led to reduced water demands in data centers.

Renewable Energy and Site Selection

Shifts toward renewable energy sources and careful site selection are critical. Data centers should ideally be located in regions with abundant water and renewable energy, such as parts of Montana and Nebraska, rather than in water-scarce areas like Arizona. This strategic approach could reduce future water consumption by up to 86%.

Innovations in Cooling Technology

New cooling technologies are making data centers more efficient. Traditional cooling methods, which rely on fans and evaporative cooling, consume considerable water. In contrast, many modern AI data centers are adopting liquid cooling systems that directly cool processors without significant water loss. This closed-loop system minimizes evaporation and waste.

Future Projections and Sustainability Goals

As the number of data centers grows, water consumption for cooling and electricity generation is expected to rise. A 2025 study predicts that U.S. data centers could consume between 731 billion to 1,125 billion liters of water annually. However, improvements in energy efficiency and a shift to renewable sources could significantly mitigate this impact.

Conclusion

The future of AI’s water consumption will depend on decisions made by tech companies, engineers, and policymakers regarding site selection, energy sources, and cooling technologies. The choices made now will determine the sustainability of AI’s growth and its impact on local water resources.