One of the pitches for investing heavily in AI—especially resource-intensive versions such as large language models (LLMs)—is the argument that these powerful technologies have the potential to help solve sustainability and climate challenges. They can help optimize renewable energy systems, predict weather, and even discover new materials. At the heart of this argument, however, is a paradox: while it’s true that AI has the potential to advance sustainability, its environmental impact could negate much of its benefits.