Using it, not all that energy intensive
According to this article, this is not considered true anymore:
As conversations with experts and AI companies made clear, inference, not training, represents an increasing majority of AI’s energy demands and will continue to do so in the near future. It’s now estimated that 80–90% of computing power for AI is used for inference.
I think there's a reason why OpenAI, Microsoft, Google and Facebook hold the energy consumption and water usage numbers so close to their chest.