Energy use is a central constraint, with estimates of about 22.15 billion kWh annually for running ChatGPT globally and roughly 18.9 watts consumed per prompt.

AI is increasingly embedded in workplaces, boosting productivity while raising concerns about overwork, job displacement, and erosion of human skills due to pervasive AI augmentation.

Usage is shifting toward the Global South, led by India with around 13.2 billion prompts per month and the United States at 11.9 billion, signaling rising AI adoption across emerging economies.

AI is framed as a permanent layer of digital infrastructure, underscoring the need for scalable, sustainable energy and economic models to sustain growth while mitigating environmental and social impacts.

ChatGPT has reached scale with roughly 900 million weekly active users and about 1.17 trillion prompts processed annually, marking its move from novelty to essential infrastructure.

Europe, especially the UK, shows high usage intensity with 37.1 million monthly visits and 2.44 billion prompts per month, equating to about 35 prompts per person per month.

There is a tension between performance growth and energy efficiency as larger models and broader adoption raise demand while the industry pursues more efficient architectures and data-centre optimizations.

The UK’s ChatGPT energy footprint exceeds 46,000 MWh per month or over 550 GWh per year, comparable to weeks of continuous operation at a large power plant.

If powered entirely by the U.S. grid, current ChatGPT operations could cost more than $8 million per day in energy expenses, highlighting the financial heft of energy use.