As artificial intelligence becomes one of the most important technological innovations of this generation, Google has further announced a comprehensive methodology for measuring the impact of AI on energy consumption, water use, and carbon emissions, emphasizing its commitment to reducing the environmental burden of AI through continuous innovation and transparency.
Ben Gomes, Google's Chief Technology Officer for Learning and Sustainability, said that AI has been widely used in fields such as medicine, energy, autonomous driving, and quantum computing. In the future, it will help people solve major challenges such as assisted learning, disease diagnosis, traffic network security, and evenPredicting wildfire pathsHowever, realizing the full potential of AI requires a robust energy infrastructure and efficient operations, coupled with the introduction of innovative solutions.
Investing in infrastructure and improving efficiency
Google stated that to meet the massive demands of AI, it is continuously investing in new energy infrastructure, building a smarter and more resilient power grid, and expanding its use of next-generation clean energy. Furthermore, Google is pursuing maximum efficiency in every aspect, from its proprietary hardware design to the software and model operations in its data centers, all with the goal of reducing energy consumption.
One key advancement is the practical effectiveness of AI model efficiency. According to data released by Google, over the past 12 months, the Gemini model has not only consistently delivered higher-quality responses to text prompts, but has also reduced its median energy consumption and carbon footprint to 1/33rd and 1/44th of its original levels, respectively. This translates to a power consumption equivalent to watching TV for less than nine seconds per prompt.
Emission Reduction Results and Environmental Responsibility
Google emphasized that these advances build on years of improving data center efficiency. For example, in 2024, despite a 27% increase in electricity consumption due to overall business and service expansion, data center energy emissions actually decreased by 12% compared to the previous year through efficiency optimization.
Beyond hardware and computational tuning, Google also considers transparency a crucial component of its sustainability efforts. The data methodology released this time aims to address information gaps regarding the energy, water, and carbon impacts of AI inference, providing industry and academia with more benchmarks to follow.
Sustainability commitment continues to deepen
Ben Gomes stated that Google will continue to invest in the technologies and innovations needed for energy transition and AI development, and hopes to drive the development of efficient AI through more accurate data research. This is not only a corporate responsibility, but also a response to a critical moment for global energy, sustainability, and scientific discovery.
Google has now publishedRelated technical reports, detailing data calculation methods and energy innovation strategies, demonstrating its attempt to balance efficiency and environmental responsibility in the AI era.
