Google Says AI Uses Just "5 Drops" of Water Per Prompt — Experts Call It Misleading
Google recently made headlines with a bold environmental claim: a median Gemini AI text prompt uses only "0.26 milliliters (or about five drops) of water" and electricity equivalent to watching nine seconds of television. The tech giant's new study positions its AI assistant as remarkably eco-friendly, but experts are pushing back hard, accusing Google of "hiding critical information" and presenting a dangerously incomplete picture.
Google's Environmental Claims
In its latest sustainability report, Google presented what appears to be encouraging news about AI's environmental impact. Google claims it achieved a 44x reduction in total emissions per prompt between May 2024 and May 2025, positioning Gemini as an increasingly efficient AI system.
Google's Key Statistics:
- Water Usage: 0.26 milliliters (5 drops) per prompt
- Energy Consumption: 0.24 watt-hours per prompt
- Carbon Emissions: 0.03 grams of CO2 per prompt
- Energy Comparison: Less than watching 9 seconds of TV
The median Gemini prompt used 33 times more energy in May 2024 than it did in May 2025, according to Google, suggesting dramatic efficiency improvements through model optimizations and software enhancements.
The Expert Pushback
Hidden Water Costs
The most vocal critic has been Shaolei Ren, associate professor at the University of California, Riverside, whose work Google cites in its paper, who argues that the study "hides the critical information" and "spreads the wrong message to the world".
The Core Issue: Google's water figure only reflects cooling systems inside its facilities, excluding indirect water usage—the water consumed at power plants that supply electricity to data centers.
Ren explains that his study included both direct and indirect water use, while Google's report did not, with AI queries potentially using up to 50 milliliters of water per prompt, far higher than Google's 0.26 milliliters.
Methodological Concerns
What Google Counts:
- Direct cooling water at Google data centers
- Immediate facility operations
- On-site water consumption
What Google Omits:
- Water used at power plants generating electricity
- Indirect cooling for the electrical grid
- Manufacturing and infrastructure water costs
- Training phase water consumption (only covers inference)
The research only considered the water consumed by data centers for cooling systems, ignoring the larger picture of electricity consumption and its associated water usage.
The Broader Context
Industry Water Usage Reality
The debate comes amid growing concerns about AI's environmental impact. One study by the non-profit Electric Power Research Institute estimated that a prompt issued to OpenAI's ChatGPT uses 2.9 watt-hours of energy while a traditional search uses about 0.3 watt-hours.
Global Scale Concerns: The global AI demand may require 4.2 – 6.6 billion cubic meters of water withdrawal in 2027, which is more than the total annual water withdrawal of 4 – 6 Denmark or half of the United Kingdom.
Training vs. Inference
Google's study focuses exclusively on inference (answering prompts) but excludes training costs. Training the GPT-3 language model in Microsoft's state-of-the-art U.S. data centers can directly evaporate 700,000 liters of clean freshwater, but such information has been kept a secret.
Why This Matters
Corporate Accountability
Amid a fierce debate about the environmental toll of artificial intelligence, Google released a new study that says its Gemini AI assistant only uses a minimal amount of water and energy for each text prompt. But experts say that the tech giant's claims are misleading.
The controversy highlights a broader issue: as AI companies face increasing scrutiny over environmental impact, how they measure and report usage becomes crucial for public policy and corporate accountability.
Misleading Comparisons
Google's claims are misleading because they draw a false equivalence between onsite and total water consumption, according to experts. The "5 drops" figure creates a compelling narrative that may not reflect the true environmental cost.
Public Understanding
Experts are pushing back on Google's claim that a Gemini prompt uses "5 drops of water," saying it paints the wrong picture by leaving out the bigger, hidden water costs.
The Methodology Gap
Google's Approach:
- Measures only direct facility water usage
- Focuses on operational efficiency improvements
- Excludes upstream electricity generation costs
- Covers inference only, not model training
Comprehensive Approach (Critics' View):
- Includes both direct and indirect water consumption
- Accounts for entire supply chain impact
- Considers electricity generation water requirements
- Includes training and infrastructure costs
Industry Response and Future Implications
The controversy underscores the need for standardized environmental impact measurement across the AI industry. Without consistent methodologies, companies can present dramatically different environmental footprints for similar services.
Key Questions Moving Forward:
- Should AI companies be required to report total lifecycle environmental costs?
- How can the public distinguish between direct and indirect environmental impacts?
- What role should regulatory bodies play in standardizing AI environmental reporting?
Bottom Line
While Google's efficiency improvements for Gemini are likely real and commendable, the "5 drops of water" claim appears to tell only part of the story. The technology giant unveiled a new methodology to measure the environmental impact of its AI models, including energy and water consumption as well as carbon emissions, but critics argue this methodology is incomplete.
The debate highlights a crucial challenge for the AI industry: balancing technological advancement with environmental responsibility requires transparent, comprehensive impact assessment. As AI becomes increasingly central to daily life, the public deserves complete information about its true environmental costs—not just the most favorable metrics companies can present.
For consumers and policymakers making decisions about AI adoption, understanding the difference between direct facility usage and total environmental impact could be crucial for informed decision-making in our increasingly AI-driven world.
The controversy continues as the AI industry grapples with environmental accountability standards. Google has not yet responded to the specific criticisms about its methodology's scope and completeness.
