In the rapidly evolving world of artificial intelligence, where computational demands are skyrocketing, Google has taken a bold step toward transparency by disclosing the energy footprint of its Gemini AI model. According to a recent technical report from the company, a single median text prompt to Gemini consumes just 0.24 watt-hours of electricity—roughly equivalent to the power used for watching television for less than nine seconds. This revelation, detailed in a paper released in August 2025, also pegs the associated carbon emissions at 0.03 grams of CO2 equivalent and water usage at 0.26 milliliters, or about five drops.

The data stems from Google’s comprehensive analysis of its inference-serving stack, factoring in everything from active accelerator power to data center overheads. As MIT Technology Review reported, this marks the first time a major tech firm has provided such granular insights into per-prompt energy use for a widely adopted AI product. The move comes amid growing scrutiny over AI’s environmental impact, with researchers long frustrated by the opacity of Big Tech’s operations.

Unpacking the Efficiency Gains

Google attributes these modest figures to significant optimizations over the past year, including a 33-fold reduction in energy consumption and a 44-fold drop in carbon footprint per prompt. Partha Ranganathan, a Google vice president and engineering fellow, emphasized the company’s “environmental responsibility” in handling AI’s computing demands, as noted in coverage by Axios. These improvements stem from better software efficiency and a shift toward cleaner energy sources in Google’s data centers.

However, the report focuses solely on text-based prompts during the inference phase—excluding more resource-intensive tasks like image or video generation. For scale, Datacenter Dynamics calculated that one million such queries would consume 240 kilowatt-hours, emit 30 kilograms of CO2, and use 260 liters of water. While individually negligible, the aggregate effect across billions of daily AI interactions worldwide raises broader concerns.

Skepticism from Experts

Not everyone is convinced by Google’s numbers. Critics argue that the estimates underplay the full environmental toll by omitting upstream costs, such as the massive energy required for training AI models or manufacturing hardware. Shaolei Ren, an associate professor at the University of California, told TechRadar that the company is “hiding the critical information,” potentially misleading the public on AI’s true sustainability challenges.

This sentiment echoes in ZDNet‘s analysis, which highlights how Google’s figures are lower than many independent estimates, yet the industry-wide energy surge from AI remains opaque. For context, a separate study from Epoch AI suggested ChatGPT uses about 10 times less energy per query than commonly thought, but such comparisons underscore the need for standardized reporting.

Implications for the AI Industry

Google’s disclosure could pressure rivals like OpenAI and Meta to follow suit, fostering a more accountable ecosystem. As BGR pointed out, with Gemini boasting around 47 million users, the per-prompt efficiency might seem trivial, but multiplied across global usage, it equates to substantial resource demands—comparable to powering small cities.

Looking ahead, the push for greener AI involves not just efficiency tweaks but systemic changes, such as renewable energy integration and advanced cooling technologies to curb water use. Industry insiders note that while Google’s report is a commendable start, true progress will require holistic lifecycle assessments, including training and deployment phases.

Balancing Innovation and Sustainability

The debate highlights a core tension: AI’s potential for breakthroughs in fields like medicine and climate modeling versus its hidden costs. RCR Wireless News described the per-prompt impact as “substantially lower than many public estimates,” yet experts warn against complacency.

Ultimately, as AI adoption accelerates, stakeholders must demand greater transparency to ensure technological advancement doesn’t come at an unsustainable environmental price. Google’s initiative, while imperfect, sets a precedent that could reshape how the industry measures and mitigates its footprint.