Stories
AI's Hidden Environmental Cost: Will Climate Constraints Reshape Consumer Pricing?
C
CNBC TV18•05-10-2026, 08:28
AI's Hidden Environmental Cost: Will Climate Constraints Reshape Consumer Pricing?
- •AI's current flat-rate subscription model masks significant environmental costs, including electricity and water consumption for data centers.
- •Data centers consumed 415 terawatt-hours globally in 2024, projected to double by 2030, with AI as a primary driver.
- •While a single AI prompt's resource demand is small, billions of interactions scale up to material infrastructure demands.
- •Environmental factors like heatwaves and droughts could link AI costs to when, where, and how computing power is consumed.
- •Future AI pricing may shift from flat subscriptions to a dynamic model reflecting resource intensity, potentially introducing an "environmental compute tariff".
Key Questions Answered
According to the International Energy Agency, data centres consumed approximately 415 terawatt-hours of electricity globally in 2024, which accounts for around 1.5% of worldwide electricity consumption.
A 2025 Google study measuring production AI infrastructure found that the median Gemini Apps text prompt consumed roughly 0.24 watt-hours of electricity and approximately 0.26 millilitres of water.
Microsoft reported that its average water-use effectiveness decreased from 2.3 litres per kilowatt-hour in its earliest generation of data centres to 0.27 litres per kilowatt-hour in 2025, alongside introducing closed-loop, direct-to-chip cooling designs.
