TV (1 year) = 516,667 AI prompts
TV (1 year) equals roughly 516,667 AI prompts on the electricity side: 155,000 Wh, while an average text prompt weighs just 0.3 Wh. The math, the source and detailed equivalents are right below, with public, verifiable numbers.
Figures updated on June 30, 2026 · Source : ADEME : consommation des appareils ménagers
The electricity of a prompt covers inference: the work the model does in the data center while answering, cooling included. It does not include model training or server manufacturing. The everyday action, meanwhile, is measured at the plug, so the comparison is one use against another use.
Why 516,667 prompts?
The math fits in three lines, using public, sourced values:
- 1.TV (1 year): 155,000 Wh (ADEME : consommation des appareils ménagers)
- 2.1 text prompt: 0.3 Wh (published range: 0.24 to 0.4 Wh)
- 3.155,000 Wh ÷ 0.3 Wh ≈ 516,667 prompts
In other words, “TV (1 year)” weighs as much as 516,667 questions asked to a generative AI. It is an order of magnitude, not a to-the-decimal measurement: sources publish ranges, and we use a median value, explained in our methodology.
What about a generated image or a reasoning prompt?
Not every AI use weighs the same. A generated image costs about 9 Wh and a “reasoning” prompt about 20 Wh, versus 0.3 Wh for a simple text prompt. So “TV (1 year)” is also roughly 17,222 times one generated image, or 7,750 times one reasoning prompt.
Convert your own usageWhere do these numbers come from?
The value we use for “TV (1 year)” comes from “ADEME : consommation des appareils ménagers”, referenced like all our data on the Sources page. No number on this page is made up: every value can be traced back to a public document.
On the prompt side, published values range from 0.24 to 0.4 Wh: Google and OpenAI have shared their own figures, and the research institute Epoch AI produced an independent estimate. We use 0.3 Wh.