AI vs Netflix: which uses more water and energy?
An hour of streaming uses about 0.08 kWh across devices, network and data center; at Mid water terms, roughly 0.4 liters. That equals about 55 ChatGPT prompts. A nightly episode habit out-consumes a daily chat habit; your TV is the bigger pipe.
One AI prompt (Mid)
7 mL
one hour of Netflix
400 mL
Prompt: one typical ChatGPT exchange at the Mid scenario (cooling plus the water behind the electricity). Item: its full water footprint, as published by the source.
One hour of Netflix uses as much water as
≈ 57.1
AI prompts at Mid
1 dot = 1 prompt
Streaming had its own scare cycle in 2019 ('one hour = two fridges!') before careful work by the IEA and Carbon Trust landed near 0.08 kWh per viewing hour, most of it in the viewing device. AI is living the same cycle with the boundaries drawn even less consistently, which is precisely what our scenario toggle is for.
Per hour of attention, chat is light: you'd need to fire off a prompt every minute of the show to match the show. Video generation flips the comparison: a single 5-second AI clip at our mid value (150 Wh) costs the same energy as roughly two hours of streaming, and the only public measurement of a current-quality open model (~940 Wh, MIT Technology Review 2025) costs more than an entire evening. The interesting line isn't AI vs Netflix; it's text vs video, in both directions.
Your number
What does your AI actually use?
Pick your model, set your usage, get the number, with sources.
Sources
Cited on this page, in order of appearance.
- 1Jegham, N., Abdelatti, M., Elmoubarki, L., & Hendawi, A. (2025). How Hungry is AI? Benchmarking Energy, Water, and Carbon Footprint of LLM Inference. arXiv preprint arXiv:2505.09598 (v6, Nov 2025). (accessed 2026-06) preprint
- 2Altman, S. (2025). The Gentle Singularity. blog.samaltman.com: 0.34 Wh and 0.000085 gal of water per average ChatGPT query. (accessed 2026-06) official disclosure
- 3Siddik, M. A. B., Shehabi, A., & Marston, L. (2021). The environmental footprint of data centers in the United States. Environmental Research Letters 16(6): watershed-scale direct + indirect water footprint. (accessed 2026-06) peer-reviewed
- 4O’Donnell, J., & Crownhart, C. (MIT Technology Review; measurements by S. Luccioni) (2025). We did the math on AI's energy footprint. MIT Technology Review, May 2025: ~3.4 MJ (~940 Wh) per current-quality 5-second CogVideoX clip; ~30 Wh for an older low-quality model. (accessed 2026-06) news
- 5US Energy Information Administration (2024). Electricity use in homes. US EIA: average US household electricity consumption ~29 kWh/day. (accessed 2026-06) government