Very little on its own. A short text query uses roughly 0.3 Wh and 0.26 mL of water, which amounts to a fraction of a gram of CO2 and the energy of a Google search or two. The impact comes from the aggregate volume of billions of queries and the rapid growth of the data centres that run them. For an organisation, it is this volume that goes into the carbon footprint, through the AI method under Scope 3.
Read the source article: ChatGPT vs Google search energy: is it 10 times more?