Shorter AI prompts may significantly reduce growing energy demand: Report
A new UNU report warns that AI's growing energy footprint is significant, but using shorter prompts and lighter models could meaningfully reduce electricity consumption at scale

With the rapid rise in AI adoption, one of the biggest questions emerging today is how much computing power and energy will be required to sustain its future expansion. A recent report by the United Nations University Institute for Water, Environment and Health (UNU-INWEH) paints an alarming picture of AI's environmental footprint.
According to the report, ChatGPT alone processes around 2.5 billion prompts every single day. At a conservative estimate of 0.42 Wh per prompt, that adds up to roughly 282 GWh of electricity consumption annually. The report notes that this amount of energy is enough to meet the yearly electricity needs of nearly 3 million people in Sub-Saharan Africa.
How the Length of Your Prompt Impacts Energy Use
The report further highlights that the length of your prompt directly affects how much energy AI chatbots consume. It introduces the concept of a "concise mode"—the idea that shorter prompts use less energy because AI inference costs scale with the number of tokens processed.
According to the findings, reducing token usage by just 30% during everyday interactions could lower per-query energy consumption by roughly 25%. This could save between 87 and 98 GWh of electricity annually, equivalent to the residential electricity consumption of up to 756,000 people.
The same principle applies to the type of AI task being performed. A typical ChatGPT-style text query uses about 200 times more energy than a basic spam-filtering operation. Generating a single AI image requires approximately 2.9 Wh of energy, making it around 60 times more demanding than a short text response. Video generation is even more resource-intensive, with complex clips consuming more than 415 Wh each.
To think that AI-generated slop videos—already cluttering video-streaming platforms—are also consuming such large amounts of electricity is a difficult reality to ignore.
What Can You Do About It?
You don't need to stop using AI if it helps with your daily tasks or work. However, it may be worth reconsidering using powerful AI systems to generate endless memes, low-value content, or "brainrot" videos.
The next time you're asking ChatGPT something simple, try to keep your prompt concise. Skip unnecessary pleasantries, get straight to the point, and choose a lighter AI model when a task doesn't require significant computing power. Small habits, when adopted at scale, can add up to a surprisingly meaningful reduction in energy consumption.

Google burns cash for first time as AI spending pushes 2026 capex to $205 billion
Tesla sells more cars but makes less money: Why Musk’s AI pivot is squeezing profits
OpenAI launches Presence to bring AI agents into customer support and enterprise workflows
Samsung Galaxy Fold 8 Ultra, Fold 8, Flip 8 launched: Here is how much it costs in India with discounts
Florida pastor sues OpenAI, says ChatGPT's medical advice delayed emergency treatment: Report
