“AI Is Depleting Not Only Our Jobs But Also Our Resources”

By GlobalTimesAI.com
📅 August 2, 2025

1. Sam Altman and the Real Story of AI’s Water Consumption

According to OpenAI CEO Sam Altman, one ChatGPT query uses roughly 0.000085 gallons, or one-fifteenth of a teaspoon, of fresh water. Although it may not seem like much, this adds up to tens of thousands of gallons of fresh water used daily when multiplied by billions of queries per day, bringing attention to an environmental cost that is frequently disregarded.

Additionally, according to press reports, OpenAI’s early AI development in Iowa resulted in a 34% increase in water usage between 2021 and 2022 because massive amounts of water were taken from the Raccoon and Des Moines rivers to cool the supercomputers used to train GPT-4.

---

2Why AI Requires So Much Power and Water

Large amounts of heat are produced by high-performance computing devices (GPUs, TPUs), which are essential to AI-powered data centers. Because of this, they need strong water-based cooling systems. A typical 100 MW data center uses roughly 2 million liters (500,000 gallons) of water per day, which is equal to the average usage of 6,500 homes in the United States, according to the International Energy Agency.

According to additional research, by 2027, AI data centers will use between 4.2 and 6.6 billion m³ of freshwater annually, which is more than half of the UK’s water consumption or Denmark’s. The environmental review emphasizes that in order to ensure sustainability, AI’s water footprint must be taken into account in addition to its carbon footprint.

---

3 Voices from the Industry & Expert Opinions

Prof. Shan‑Wen Ren, UC Riverside

Some AI queries use as much electricity as a mile of driving a Tesla Model 3, according to a Renan expert cited in Vox. They also need roughly 500 mL of water for cooling. He bemoaned the lack of transparency, saying that we are working in a “black box” with no precise information on how much water or energy AI uses.

Abhijit Dubey, CEO of NTT DATA

Dubey stressed that in order to preserve chip performance and avoid bacterial contamination, liquid cooling—in particular, freshwater immersion cooling—is crucial. A significant amount of treated fresh water is used in this process.

Larena Xalim Palas, Founder of Ethical Tech Society

Palas cautioned that the majority of AI data center cooling uses freshwater, which is frequently taken from municipal or irrigation supplies, depleting resources that are needed by agriculture and communities.

---

4. International Demonstrations & Local Issues

In drought-prone areas, communities in Spain, India, Chile, Uruguay, and parts of the United States (particularly Texas, Arizona, and the United Kingdom) have expressed concern about AI data centers depleting local freshwater supplies.

A planned hyperscale AI center in Lincolnshire, UK, drew opposition from Anglian Water, which warned that it might overwhelm local systems and deplete clean water more quickly than reservoirs could replenish, according to The Times. Data centers in central Texas are now responsible for 49 billion gallons of water use annually, which is more than 6% of the state’s water use. This is creating serious tension in the midst of the drought in San Antonio.

---

5. The Carbon Burden and Electricity

A study that compiled data on more than 2,100 data centers in the United States found that the industry currently uses more than 4% of the country’s electricity and produced more than 100 million tons of CO₂ in 2023 alone, which is more than many entire nations.

Training and operating large language models escalate this footprint. For example, training GPT‑3 is estimated to have consumed 700,000 L of fresh water and 490 metric tons of CO₂, equivalent to powering 98 U.S. homes for a year. Inference workloads across billions of prompts add vastly more over time.

---

6. Tech Giants’ Water Commitments & Criticism

The promises made by Microsoft, Google, Amazon, OpenAI, and Meta frequently rely primarily on offsetting rather than lowering actual withdrawals in areas that are under stress.

Policy experts stress that without transparency, these commitments lack accountability, even though Google’s sustainability head acknowledged that water can help reduce carbon emissions.

---

7. Alternatives & Future Direction

In order to significantly lessen dependency on freshwater resources, UNICEF officials like Thomas Dawin have suggested locating data centers offshore or utilizing ocean cooling. Similarly, building centers

in cooler climates—such as Northern Sweden or Finland, where operators use seawater cooling—can provide sustainable alternatives.

---

8. Summary: The AI Environmental Toll

Issue	Key Insight
Water use	Millions of liters per center daily; over billions annually
Electricity	Large energy footprint; ~4% of U.S. electricity consumption
Carbon emissions	Model training alone emits hundreds of tons of CO₂
Freshwater stress	Centers built in drought-prone areas worsen local shortages
Expert warnings	Industry executives and ethicists call for reform
Alternatives	Seawater cooling, reuse systems, regulation needed

AI’s growth is real, but so is its demand for precious water and energy. As demand surges, this resource-intensive model raises urgent questions: At what cost does digital progress come? Can tech scale without draining humanity’s most vital resources?

Final Thought

The astounding advancements in AI come at an environmental cost.

The quest for computing power is crossing paths with freshwater scarcity, especially in drought-prone regions. While leaders like Altman and Ren offer transparency, the spotlight now moves to industry-wide accountability and innovation.

Solutions like seawater-cooled facilities or offshore cooling show promise—but only if backed by regulatory frameworks and ethical stewardship.

As AI becomes pervasive, balancing innovation with sustainability is no longer optional—it’s essential.

---

🔖Tags: #AIEnergyUse #WaterFootprint #DataCenterSustainability #EthicalTech #OpenAI #NTTDATA #ClimateImpact

Disclaimer:
The content of this article is based on publicly available information from various reputable sources, including statements from industry leaders, official research papers, and verified news outlets. All images used are AI-generated and intended solely for illustrative purposes. References and data have been taken from credible sources to ensure accuracy, but readers are encouraged to verify critical details independently for their own understanding.