Google Explores the True Environmental Impact of AI Queries: Is Their Energy Use Minimal?

Tech
By 24matins.uk,  published 26 August 2025 at 15h30, updated on 26 August 2025 at 15h30.
Tech

Google has released data highlighting the environmental impact of a single AI-powered search, revealing that its energy consumption is lower than many might expect. This disclosure aims to address concerns about the sustainability of widespread AI use in digital services.

Tl;dr

  • Google discloses Gemini’s energy, water, and carbon usage.
  • Efficiency gains may drive higher total consumption.
  • Collective impact requires user responsibility and transparency.

    • Transparency Unveiled: Google Lifts the Veil on Gemini’s Environmental Costs

    For the first time, Google has published detailed figures showing exactly how much energy, water, and carbon are consumed by each query processed through its flagship AI model, Gemini. On paper, the individual toll appears almost negligible—one prompt uses only 0.24 watt-hours of electricity (equivalent to nine seconds of television viewing), generates about 0.03 grams of CO₂, and draws just five drops of water. In other words, a seemingly modest environmental footprint.

    The Efficiency Paradox: A Double-Edged Sword

    And yet, scratch beneath the surface and a more complex reality emerges. Although Google touts dramatic year-over-year improvements—a 33-fold decrease in energy use and a 44-fold reduction in carbon emissions per prompt—this very progress revives an old economic conundrum. William Stanley Jevons once noted that making a resource cheaper tends to spur its consumption. In the case of AI, as prompts become « less costly », they become more frequent; paradoxically, greater efficiency could accelerate overall demand rather than curb it.

    Cumulative Effects: Small Prompts, Large Impact

    That phenomenon takes on even greater significance at scale. While the exact number of daily prompts to Gemini remains undisclosed, estimates suggest several hundred million queries each day. This sheer volume turns what might seem like an inconsequential impact into substantial pressure on power infrastructure—raising the specter that U.S. demand from AI services could triple by 2028. The implications extend well beyond individual actions, threatening to burden national grids and inflate energy bills across the board.

    So what can individuals actually do? Experts recommend a handful of practical steps:

  • Select services that report their true resource consumption.
  • Consolidate queries to avoid unnecessary server loads.
  • Consider how data centers are powered—renewable or fossil-based sources matter.

    • A Shared Responsibility for an AI-Powered Future

    While Google’s push for transparency represents a meaningful step forward, it also highlights a crucial reality: the aggregate environmental effect is far from trivial. As society embraces ever more widespread use of artificial intelligence, we must confront the « hidden costs »—not just for individuals but on a planetary scale. The collective response will shape whether these innovations ultimately serve as catalysts for sustainability or as drivers of new challenges.