AI’s Not So Hidden Thirst

by Claire Brady, EdD

“Whether the number is five drops or 50 milliliters per prompt, at scale the impact is massive.”

When most higher education leaders think about artificial intelligence, the conversation centers on teaching, learning, advising, or workforce development. But there’s another angle we can’t afford to ignore: the environmental footprint of AI.

A recent dispute between Google and researchers at UC Riverside highlights just how complicated—and contested—this issue has become.

Google recently published a report claiming that its Gemini apps consume just 0.26 milliliters of water per prompt—about five drops—for the average text generation. That sounds impressive compared to earlier studies suggesting 45–47 milliliters of water were required per page of AI-generated text. But critics argue that Google is comparing “apples to oranges.”

What the data tells us

Datacenters use water in two ways. First, directly, through cooling towers that evaporate water to prevent servers from overheating. Second, indirectly, through the massive water consumption required by energy plants (coal, gas, nuclear) that generate electricity to power datacenters.

Google’s new figure measures only onsite water use. By leaving out the offsite consumption tied to electricity generation, the company minimized the true footprint of its AI workloads.

UC Riverside’s research did include onsite estimates. In fact, its 2023 study put the average onsite use at 2.2 ml per prompt in U.S. datacenters—far lower than the “50 ml” figure Google used for comparison.

In short: yes, AI efficiency is improving, but the environmental impact is still very real. Datacenters are water-intensive by design, and the demand for AI is scaling much faster than sustainability practices.

Why This Matters for Higher Education

Universities are not only consumers of AI technologies but also community leaders, research hubs, and role models for sustainability. As AI adoption expands—from admissions chatbots to research computing—the sector must grapple with its environmental footprint. Campuses already face scrutiny over energy use, water conservation, and climate commitments. The growth of AI adds a new dimension.

Ignoring this reality could create reputational risks (“greenwashing” by overstating efficiency gains) and operational costs (as utilities become more strained during droughts or heatwaves). Just as importantly, students are watching. Gen Z and Gen Alpha are deeply attuned to environmental justice, and they want institutions to walk their sustainability talk.

Action Steps for Campus Leaders

1. Incorporate AI into Sustainability Planning

If your institution has climate action or sustainability goals, explicitly add AI to the conversation. Ask: How much additional server capacity are we drawing on? What is the water and energy cost of those systems?

2. Demand Transparency from Vendors

When contracting with cloud providers or AI vendors, require disclosures on both onsite and offsite water and energy use. Push back against “apples-to-oranges” claims.

3. Educate Campus Stakeholders

Students and faculty often assume digital tools are “clean.” Include AI’s environmental impact in curriculum discussions on ethics, technology, and sustainability. This builds awareness and empowers informed choices.

4. Explore Offsets and Mitigation

If AI adoption is increasing your institution’s digital footprint, pair it with visible investments in water conservation, renewable energy, or sustainability research.

5. Model Responsible Adoption

Leaders should frame AI not only as a teaching and research innovation but also as a stewardship challenge. Model a balanced narrative: AI offers enormous potential, but it is not “free” from ecological costs.

The Leadership Imperative

The truth is still evolving. AI is becoming more efficient, but its growth is outpacing those gains. Whether the number is five drops or 50 milliliters per prompt, at scale the impact is massive.

For higher education leaders, the challenge is clear: don’t treat AI’s environmental footprint as an afterthought. Build it into your decision-making, your sustainability strategies, and your accountability to the students and communities you serve.

AI may help us solve global problems—but only if we remain honest about the resources it consumes along the way.