The energy and water cost of AI
Understand what is known and unknown about AI's electricity, water and carbon footprint
A taste of a lesson
I read that every AI question uses a bottle of water. Is that true?
That kind of figure comes from early estimates with specific assumptions, and it is shaky as a general rule. Water per query depends on the data centre's cooling design, the local climate, the power source, the model, and how long the answer is. Some providers have since published much smaller per prompt estimates for typical text, though methods are hard to compare. The more solid concern is total and local: big facilities in water stressed areas. Exercise: for any figure you see, note its source, date and what it includes.
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- Explain where energy goes in training, inference, cooling and hardware manufacturing
- Describe how water use and carbon emissions depend on location, design and grid mix
- Evaluate per query and total footprint claims critically
- Identify the actions with most influence for individuals, organisations and policymakers
Lesson plan
- 1 Where the energy goes Map the energy use of AI from training to everyday answers. Start
- 2 Water and cooling Understand why and how data centres use water. Start
- 3 Carbon depends on where and when See how grid mix and reporting methods change carbon figures. Start
- 4 Reading footprint claims Judge per query and total figures critically. Start
- 5 What actually helps Identify effective actions at personal, organisational and policy levels. Start
Try asking
About this tutor
For anyone concerned about AI's environmental impact, from individual users to sustainability staff and policy readers. You learn where energy goes in AI: training runs, everyday inference, data centre cooling and hardware manufacturing. You learn why water is used, why location and time of day change carbon emissions, and why per query figures vary so much between estimates. You practise reading environmental claims critically, including both alarming and reassuring ones, and you consider what individuals, organisations and policymakers can actually influence. The tutor is honest that transparency is limited and many numbers are estimates.
Reviews
4.5
2 ratingsSample
- Ingeborg L.Sample
Finally someone separating estimates from facts. The market based versus location based point helped me read our cloud provider's sustainability report.
- Kenneth O.Sample
Balanced. I came in feeling guilty about every prompt and left understanding where the real levers are. Wanted a few more concrete numbers.
About the teacher
I explain the kinds of AI models, what they cost to run and how to run one yourself
9 tutors 328 lessons taught Sample
I teach the practical side of modern models: reasoning models, multimodal models, open and closed weights, running a model on your own computer, and the money, energy and hardware behind every answer. I like starting with something you can see or measure, such as the memory a model needs or the number of tokens a task uses, then explaining the...
See Benedict's profile and tutorsMore like this
Other tutors on the same or nearby topics.