The Environmental Impact of Google Search's AI Overviews
Over the past two years, Google’s AI Overviews have become part of everyday browsing. Where blue links once dominated, many searches now surface an AI-generated summary above website results.
Many users welcome the convenience, valuing concise answers that reduce time spent clicking through pages. The shift also embeds more computation into one of the internet’s most-used services.
As AI becomes more visible at the top of search results, the debate is widening. Environmental impacts and the health of the open web are now central concerns.
Shifting to AI summaries
AI Overviews aim to deliver instant answers by distilling content from across the web. The result is a faster experience for users who want clarity in a single glance.
Ahrefs estimates that 55% of searches now surface an AI Overview. That level of adoption means AI assistance is no longer the exception within Google’s results.
This change alters not only how people find information but also what gets measured behind the scenes. Every AI-generated summary adds computing steps to a familiar query.
The convenience is clear. The environmental implications are less visible to users, yet they scale with every interaction.
Prompting energy use and water
In August 2025, Google published its first comprehensive estimate of the resources used by a prompt in its Gemini applications. The figures offer a useful baseline.
According to Google, a median Gemini text prompt consumes 0.24 Wh of electricity. It produces 0.03 g of CO₂ and uses 0.26 ml of water for data centre cooling.
Google indicates that generating an AI Overview in Search uses similar energy to a standard Gemini prompt. On a per-query basis, those numbers can appear small.
Taken in isolation, a single summary looks negligible. The picture shifts as these interactions multiply across global search volumes.
Search traffic drives higher power consumption
Google says it processes five trillion searches each year. Even modest increases per query become significant when expanded to that scale.
The trend is already visible in Google’s sustainability reporting. The company’s greenhouse gas emissions were 48% higher in 2023 than in 2019.
Google attributes much of the rise to expanding data centre infrastructure and the growing intensity of AI and Gen AI workloads. It acknowledges that further decarbonisation may become harder as AI is embedded more deeply across products.
Researchers warn that rising AI demand is driving a wave of data centre construction. This adds pressure to electricity grids and, in some regions, prolongs reliance on fossil fuel generation.
Publishers, traffic and user control
The discussion is not only about kilowatt-hours and millilitres. It also concerns how information circulates across the open web.
Publishers, website owners and SEO professionals report traffic declines when AI-generated answers appear above traditional results. Ahrefs estimates that AI Overviews can reduce clicks to websites by more than a third.
Users are often unable to disable AI Overviews permanently. Some append “-ai” to queries to reduce the likelihood of a summary appearing.
This dynamic raises a consumer-choice question. If AI-assisted search carries additional environmental costs, should users have greater control over when it appears?
Transparency and mitigation
Google’s disclosure of Gemini’s energy, carbon and water estimates is a meaningful step. Measurement helps make abstract impacts tangible.
Transparency alone does not solve the problem of scale. The footprint of one AI-assisted query may be small, but the cumulative effect of trillions is substantial.
Mitigation pathways are clear in principle. They include efficiency gains in models and hardware, cleaner power for data centres and product design that respects user choice.
As the sustainability adage goes, one cannot manage what it cannot measure. If Google refines its understanding of AI demand and acts on it, real progress can follow.
