Google’s AI buildout drove 37% increase in electricity use in 2025 (arstechnica.com)

Google's electricity consumption surged by 37% in 2025, marking the most significant increase in its history, largely driven by the rapid expansion of AI-related infrastructure and operations. This rise aligns with a broader trend, with total usage up over 250% since 2019, influenced by the growth of Google Cloud and YouTube streaming. Despite this stark increase, Google has managed to reduce its operational carbon emissions by 2%, indicating a potential decoupling of emissions from energy consumption as it continues to invest in clean energy sources. Significantly, Google's data centers consumed over 42 million megawatt-hours of electricity in 2025, comparable to the electricity needs of entire nations, showcasing the immense energy demands of modern AI systems. As Google emphasizes the need for increased clean energy investments and collaborations with local stakeholders, this situation highlights the pressing challenge of scaling sustainable energy solutions amidst the escalating needs of AI and data infrastructure. The implications for the AI/ML community are clear: as demand for advanced AI technologies grows, so does the necessity for innovative approaches to energy consumption and environmental impact.