I found a blogpost that cites a Business Insider article that implies this claim as formulated is way off:

Reported energy use implies that ChatGPT consumes about as much energy as 20,000 American homes. An average US coal plant generates enough energy for 80,000 American homes every day. This means that even if OpenAI decided to power every one of its billion ChatGPT queries per day entirely on coal, all those queries together would only need one quarter of a single coal plant. ChatGPT is not the reason new coal plants are being opened to power AI data centers.

It goes on to argue that while it's true that AI related electricity use is booming, it's not because of LLM chatbots:

AI energy use is going to be a massive problem over the next 5 years. Projections say that by 2030 US data centers could use 9% of the country’s energy (they currently use 4%, mostly due to the internet rather than AI). Globally, data centers might rise from using 1% of the global energy grid to 21% of the grid by 2030. ...

97% of the total energy used by AI as of late 2024 is not being used by ChatGPT or similar apps, it’s being used for other services. What are those services? The actual data on which services are using how much energy is fuzzy, but the activities using the most energy are roughly in this order:

* Recommender Systems - Content recommendation engines and personalization models used by streaming platforms, e-commerce sites, social media feeds, and online advertising networks.

* Enterprise Analytics & Predictive AI - AI used in business and enterprise settings for data analytics, forecasting, and decision support.

* Search & Ad Targeting - The machine learning algorithms behind web search engines and online advertising networks.

* Computer vision - AI tasks involving image and video analysis – often referred to as computer vision. It includes models for image classification, object detection, facial recognition, video content analysis, medical image diagnostics, and content moderation (automatically flagging inappropriate images/videos). Examples are the face recognition algorithms used in photo tagging and surveillance, the object detection in self-driving car systems (though inference for autonomous vehicles largely runs on-board, not in data centers, the training of those models is data-center based), and the vision models that power services like Google Lens or Amazon’s image-based product search.

* Voice and Audio AI - AI systems that process spoken language or audio signals. The most prominent examples are voice assistants and speech recognition systems – such as Amazon’s Alexa, Google Assistant, Apple’s Siri, and voice-to-text dictation services.