<\/p>\n
Additionally, the report highlights that the \u00a0water consumption\u00a0 required for cooling servers stands at \u00a00.26 milliliters\u00a0 per consultation\u2014equivalent to just five drops of water. The carbon footprint of the entire inference process is estimated at only \u00a00.03 grams of CO2 equivalent\u00a0.<\/p>\n
<\/p>\n
Miscalculations Unveiled.<\/strong> Just a year prior, third-party analyses suggested that a single query made through the Google search engine could consume about \u00a03 watt-hours\u00a0 (Wh), ten times more than a traditional search. This inflated estimate led to projections suggesting that deploying AI in search could use enough energy to charge \u00a0seven electric cars per second\u00a0.<\/p>\n <\/p>\n With Google’s official data now available, we see that previous estimates were off by a staggering factor of \u00a012.5\u00a0. Enhanced software techniques\u2014such as \u00a0speculative decoding\u00a0\u2014and more efficient model architectures like the \u00a0Mixture-of-Experts\u00a0 approach have completely changed the landscape of energy consumption.<\/p>\n <\/p>\n Inference versus Training.<\/strong> The figures provided are the most precise to date, focusing solely on \u00a0Gemini’s consumption\u00a0 during the inference stage, which relates to user interactions. The much more \u00a0energy-intensive process\u00a0 of training the expansive language models that power these systems remains undisclosed. However, Google argues that the widespread adoption of generative AI, even within its search engine, has drawn focus toward inference.<\/p>\n <\/p>\n In the realm of user interaction, tech companies are making \u00a0significant efficiency strides\u00a0. Google reports that in the last year, energy consumption has reduced \u00a0by 33 times\u00a0, while the carbon footprint of each Gemini consultation has decreased \u00a0by 44 times\u00a0. This decrease can be attributed not only to more efficient models but also to improvements in their AI hardware\u2014specifically, \u00a0TPUs\u00a0 and \u00a0GPUs\u00a0 that Google manufactures internally.<\/p>\n <\/p>\n Comparative Analysis.<\/strong> Google is not the only entity shedding light on energy consumption; \u00a0Sam Altman\u00a0, CEO of OpenAI, also provided insights on the energy used by ChatGPT. In a June 2025 publication<\/a>, Altman stated that a typical ChatGPT query consumes approximately \u00a00.34 Wh\u00a0 of energy and about \u00a00.3 ml of water\u00a0.<\/p>\n <\/p>\n Although this figure is slightly higher than Gemini’s energy consumption, a direct comparison is challenging due to Altman’s lack of methodological detail. We still do not know if his calculations account for all variables considered by Google, such as electricity used for cooling systems and idle machinery that needs to be prepared for quick surges in demand.<\/p>\n <\/p>\n Both companies draw analogies with \u00a0television energy consumption\u00a0: “An hour of Netflix uses \u00a0100 times\u00a0 more electricity than ChatGPT,” says a presentation from OpenAI. It also claims that the total impact of AI on U.S. carbon emissions is approximately \u00a00.5%\u00a0.<\/p>\n <\/p>\n Images | Google<\/p>\n In conclusion, understanding the energy consumption associated with generative AI is crucial as its use continues to expand across various industries. While Google’s data points to significant efficiency improvements, the broader implications of energy use and environmental impact remain areas of keen interest and scrutiny.<\/p>\n![\"Graph](\"https:\/\/teknomers.com\/en\/wp-content\/uploads\/2025\/08\/Google-has-finally-disclosed-the-energy-and-water-consumption-of.png\")
\n ![\"Graph](\"https:\/\/teknomers.com\/en\/wp-content\/uploads\/2025\/08\/Google-has-finally-disclosed-the-energy-and-water-consumption-of.png\"\/)
\n The amount of prompts processed per kWh by different AI models<\/span>\n <\/div>\n<\/p><\/div>\n<\/div>\n