In a significant move towards transparency, Google has finally revealed how much energy its AI prompts consume. As concerns about the environmental impact of artificial intelligence grow, the tech giant shared a detailed technical report on August 21, 2025, estimating the energy, water, and carbon emissions associated with each AI query.
The Energy Footprint of an AI Prompt
According to Google’s findings, each text prompt given to its Gemini AI apps uses approximately 0.24 watt-hours of electricity. For context, that’s about the same amount of energy used to watch television for less than nine seconds! This report offers a much-needed glimpse into the resource demands of AI.
Jeff Dean, Google’s chief scientist, told the MIT Technology Review that the company aimed for comprehensive reporting. The breakdown shows that Google’s custom AI chips account for 58% of that energy. Supporting equipment like CPUs and memory use 25%, backup systems use 10%, and data center overhead (cooling, power conversion) makes up the remaining 8%.
Remarkable Efficiency Gains
What’s particularly striking is the dramatic improvement in efficiency. Google’s report indicates a massive drop in energy and carbon footprint per Gemini text prompt – 33x and 44x less, respectively, compared to May 2024. The company attributes these gains to advancements in their AI models and software optimizations.
Dean reassured the public, stating that users shouldn’t have major concerns about Gemini’s energy or water usage, as it’s comparable to everyday activities. “like watching a few seconds of TV or consuming five drops of water,” he noted.
Looking Ahead: Continued Investment in Efficiency
While Google’s report is a significant step, it does have limitations. The total number of Gemini queries isn’t disclosed, making it difficult to calculate the overall energy consumption. Additionally, the figures focus on text prompts, while image and video generation tasks, which require substantially more energy, were not included.
Google remains committed to improving efficiency, stating, “Gemini’s efficiency gains are the result of years of work, but this is just the beginning.” They are heavily investing in reducing power and water requirements per prompt and aim to drive industry-wide progress by sharing their methodology. This commitment is crucial for the responsible development of AI.
What are your thoughts on Google’s transparency regarding AI energy use? Do these figures surprise you? Share your opinions in the comments below!