Powering AI Could Soon Use as Much Water as 10 Million Americans – Gizmodo

Environmental Impact of Artificial Intelligence and Alignment with Sustainable Development Goals (SDGs)

A recent study published in Nature Sustainability quantifies the significant environmental footprint of the burgeoning Artificial Intelligence (AI) industry in the United States. The findings highlight critical challenges to achieving several Sustainable Development Goals (SDGs), including those related to climate action, clean water, sustainable energy, and responsible production. However, the report also outlines a clear pathway for mitigating these impacts through strategic, coordinated action.

Projected Resource Consumption and Climate Impact (2024-2030)

The analysis projects substantial carbon emissions and water consumption from AI server deployment, directly impacting global and national sustainability targets.

Carbon Emissions and SDG 13: Climate Action

The projected carbon footprint of the U.S. AI industry presents a significant obstacle to SDG 13 (Climate Action). Key projections include:

• An annual output of 24 to 44 million metric tons of carbon dioxide equivalent.
• This emission level is comparable to adding 5 to 10 million new gasoline-powered cars to the road each year.

Water Consumption and SDG 6: Clean Water and Sanitation

The industry’s demand for water poses a direct threat to SDG 6 (Clean Water and Sanitation), particularly in water-stressed regions. Projections indicate:

• Annual water consumption between 731 million and 1,125 million cubic meters.
• This volume is equivalent to the annual household water usage of 6 to 10 million Americans.

Socio-Economic Implications and Community Sustainability

The proliferation of AI data centers has tangible local impacts that challenge the objectives of SDG 11 (Sustainable Cities and Communities).

Impact on Local Communities

Communities hosting these facilities, particularly low-income areas and communities of color, face disproportionate consequences, including:

• Skyrocketing electricity bills.
• Increased local air pollution.
• Significant strain on regional power grid infrastructure.

Mitigation Strategies for a Sustainable AI Industry

The report concludes that the environmental impacts of AI are manageable, with the potential for a 70-85% reduction through the implementation of targeted strategies. These strategies align with multiple SDGs and provide a framework for responsible industrial growth.

Strategic Siting and Infrastructure Development (SDG 9 & SDG 11)

Optimizing the location of new data centers is the most significant factor in reducing their environmental footprint, supporting SDG 9 (Industry, Innovation, and Infrastructure). The same AI workload can have a two- to five-fold difference in environmental impact depending on its location. Recommendations include:

1. Avoid siting new facilities in water-scarce states such as California, Nevada, and Arizona.
2. Prioritize development in regions with low water stress and abundant renewable energy resources.
3. Ideal locations identified include the Midwest and “wind belt” states (Texas, Montana, Nebraska, South Dakota) and New York, which benefits from a clean electricity mix of nuclear, hydro, and renewable power.

Advancing Clean Energy and Efficiency (SDG 7 & SDG 12)

Technological and policy measures are essential for aligning the AI industry with SDG 7 (Affordable and Clean Energy) and SDG 12 (Responsible Consumption and Production). Key actions are:

1. Clean Power Procurement: Mandate and incentivize the use of renewable energy sources to power data centers, contributing to grid decarbonization.
2. Efficient Cooling Technology: Invest in and deploy advanced cooling techniques to drastically reduce both water and electricity consumption.
3. Integrated Planning: Ensure that location, power source, and cooling technology are considered collectively to transform AI infrastructure into a component of a sustainable future.

Analysis of Sustainable Development Goals in the Article

1. Which SDGs are addressed or connected to the issues highlighted in the article?

1. SDG 6: Clean Water and Sanitation
   • The article directly addresses this goal by quantifying the massive water consumption of AI data centers. It states that AI is expected to “consume roughly 193 billion to 297 billion gallons (731 million to 1,125 million cubic meters) of water per year.” It also highlights the issue of building data centers in “water-scarce states, such as California, Nevada, and Arizona,” which directly relates to the sustainable management of water resources.
2. SDG 7: Affordable and Clean Energy
   • This goal is central to the article’s discussion on the energy demands of AI. The article links AI’s electricity demand to carbon emissions and discusses solutions like “clean-power procurement,” “grid decarbonization,” and siting facilities in regions with “plentiful renewable energy resources” and a “clean electricity mix,” such as the Midwest, “wind belt” states, and New York.
3. SDG 9: Industry, Innovation and Infrastructure
   • The article focuses on the infrastructure of the AI industry—data centers. It explores how this infrastructure’s rapid growth has significant environmental consequences and advocates for making it more sustainable through “better facility location,” “efficient cooling,” and the adoption of clean technologies. This aligns with the goal of building resilient and sustainable infrastructure.
4. SDG 12: Responsible Consumption and Production
   • The core theme of the article is the unsustainable resource consumption pattern of the burgeoning AI industry. By quantifying the “resource consumption” (water and energy) and waste (carbon emissions), and proposing ways to “cut those impacts by roughly 70–85%,” the article directly engages with the principles of achieving sustainable management and efficient use of natural resources.
5. SDG 13: Climate Action
   • This goal is explicitly addressed through the quantification of AI’s carbon footprint. The study projects that AI server deployment could produce “26 million to 48 million tons of carbon dioxide equivalent per year,” which is a direct contribution to global climate change. The proposed mitigation strategies, such as using renewable energy and smart siting, are direct climate action measures.

2. What specific targets under those SDGs can be identified based on the article’s content?

1. Target 6.4: Substantially increase water-use efficiency and ensure sustainable withdrawals.
   • The article’s focus on the massive water consumption of AI for cooling (“193 billion to 297 billion gallons of water per year”) and the problem of building data centers in “water-scarce states” directly relates to the need for sustainable water withdrawals. The proposed solution of using “efficient cooling” technology points toward increasing water-use efficiency.
2. Target 7.2: Increase substantially the share of renewable energy.
   • This target is identified through the article’s proposed solutions. It advocates for “clean-power procurement” and siting new facilities in regions with “plentiful renewable energy resources,” specifically mentioning “wind belt” states and New York’s mix of “nuclear power, hydropower, and renewable energy.”
3. Target 9.4: Upgrade infrastructure and retrofit industries to make them sustainable.
   • The article is fundamentally about making AI infrastructure (data centers) sustainable. It calls for upgrading this infrastructure through “smart siting,” “grid decarbonization,” and “advanced cooling techniques” to increase resource-use efficiency and adopt cleaner industrial processes.
4. Target 12.2: Achieve the sustainable management and efficient use of natural resources.
   • The entire analysis in the article, which quantifies the water and energy consumption of AI, is aimed at understanding and managing the industry’s use of natural resources. The statement that “better facility location, clean-power procurement, and efficient cooling can cut those impacts by roughly 70–85%” is a direct call for more sustainable management and efficient use of these resources.
5. Target 13.2: Integrate climate change measures into national policies, strategies and planning.
   • The article provides the data and strategic rationale for integrating climate change measures into the planning of the AI industry. By projecting carbon emissions (“26 million to 48 million tons of carbon dioxide equivalent per year”) and advocating for mitigation strategies like “grid decarbonization,” it provides a basis for industry-level and potentially national-level strategies to combat climate change.

3. Are there any indicators mentioned or implied in the article that can be used to measure progress towards the identified targets?

1. Indicator for Target 6.4 (Water Stress):
   • The article explicitly mentions “water-scarce states” and advocates for siting in “regions with low water stress.” The amount of water consumed, “193 billion to 297 billion gallons per year,” serves as a direct quantitative indicator of the industry’s pressure on freshwater resources. Progress could be measured by a reduction in this consumption figure or a shift in data center locations to less water-stressed areas.
2. Indicator for Target 7.2 (Renewable Energy Share):
   • The article implies this indicator by discussing the “power grid carbon intensity” of different states and highlighting New York for its “clean electricity mix.” The percentage of energy for data centers sourced from renewables would be a direct indicator to measure progress toward “clean-power procurement.”
3. Indicator for Target 9.4 (CO2 Emissions):
   • The article provides a direct, quantifiable indicator of the industry’s environmental impact: “26 million to 48 million tons of carbon dioxide equivalent per year.” This figure can be used as a baseline to measure progress in making the AI infrastructure more sustainable. A reduction in this number would indicate successful retrofitting and upgrading.
4. Indicator for Target 12.2 (Natural Resource Consumption):
   • The article provides specific figures for the consumption of natural resources, namely water (“193 billion to 297 billion gallons”) and the energy that leads to carbon emissions. These figures serve as direct indicators of “Domestic Material Consumption” for the AI sector, allowing for tracking the efficiency of resource use over time.

4. Table of SDGs, Targets, and Indicators

Source: gizmodo.com