# Calculating the Cost of Chat: Tool Estimates AI Chatbot Energy Consumption

## Calculating the Cost of Chat: Tool Estimates AI Chatbot Energy Consumption

As AI models become increasingly integrated into our daily lives, questions arise about their broader impact, especially regarding energy consumption. Hugging Face engineer Julien Delavande tackled this concern head-on by developing a tool that estimates the electricity used by AI chatbot messages.

AI models demand significant power to operate, relying on GPUs and specialized chips to handle complex computational tasks. While quantifying this consumption isn’t straightforward, the growing adoption of AI is projected to substantially increase electricity demands in the coming years.

This demand has spurred some companies towards potentially environmentally damaging energy strategies. Delavande’s tool serves as a reminder of this energy footprint, encouraging AI users to consider the environmental implications of their interactions.

“Even small energy savings can scale up across millions of queries — model choice or output length can lead to major environmental impact,” stated Delavande and his team.

[A tweet from Julien Delavande’s account is embedded here, highlighting the tool’s real-time energy consumption display for chatbot messages.]

The tool is designed to integrate with Chat UI, an open-source front-end compatible with models like Meta’s Llama 3.3 70B and Google’s Gemma 3. It estimates the energy consumption of messages in real-time, presenting the data in Watt-hours or Joules. For context, the tool also compares energy usage to common household appliances.

For instance, according to the tool, generating a typical email with Llama 3.3 70B consumes approximately 0.1841 Watt-hours, equivalent to running a microwave for a mere 0.12 seconds or using a toaster for 0.02 seconds.

While acknowledging that the estimates are not perfectly precise, Delavande emphasizes their value in raising awareness. These figures highlight that even seemingly small interactions with chatbots contribute to an overall energy cost.

“With projects like the AI energy score and broader research on AI’s energy footprint, we’re pushing for transparency in the open source community. One day, energy usage could be as visible as nutrition labels on food!” Delavande and his co-creators envision a future where energy consumption is a standard metric for AI models, promoting more sustainable practices within the field.