The high energy costs of modern AI models are well documented, but new research shows that the impact is probably worse than previously thought. This may be an angle from which telcos can pitch the AI-RAN offering, where they provide resources for AI inferencing in the mobile network, since hyperscalers will want to offload this high-energy burden. This could be the moment for AI-RAN, if telcos are willing to foot the electricity bill. The environmental impact of the tech industry changed drastically with the advent of generative AI. Data centers in the US doubled their electricity consumption from 2017 to 2023, as they came to house hardware for GenAI models. We have become accustomed to snappy anecdotes that quantify the…