Challenges

Power bills behind the chatbot

Always-on AI sits on a growing electricity stack. Treat idle inference and vanity jobs as real load, not vibes.

The trap

A fluent answer arrives in seconds. The plant behind it bills in megawatt-hours. Treating every casual prompt as free energy is a craft failure, not a moral panic.

What the evidence shows

The International Energy Agency (2025) estimated global data-center electricity at about 415 TWh in 2024 and projected roughly a doubling by 2030 in its base case (toward about 945 TWh), with AI-related demand an important driver of growth. The Congressional Research Service (2025) reviewed U.S. estimates placing data centers near 4% of U.S. electricity in 2023 in cited analyses, and noted separate estimates that AI may already account for a non-trivial share of data-center energy.

None of that means your one email draft burns a town. It does mean always-on, high-volume, low-value compute is not ethically or operationally free.

What this means for people

Households and firms share grids with data centers. Efficiency and honesty about when not to run a model are part of clear-eyes use.

Practice (15 minutes)

  1. List your top five AI uses this week.
  2. Mark each: necessary, nice, or vanity.
  3. Cut or batch one vanity pattern (extra regenerations, mega-context dumps for curiosity alone).
  4. Prefer the smallest tool that meets the job when the risk is low.

Reflection

Which prompt habit is mostly entertainment dressed as work?

Skeptic check

  • Global and national aggregates are not your firm’s PUE.
  • Outlooks depend on efficiency and demand assumptions (International Energy Agency, 2025).
  • CRS synthesizes multiple studies; open the underlying citations before you cite a single percentage in a board deck (Congressional Research Service, 2025).

See also

References

Congressional Research Service. (2025). Data centers and their energy consumption: Frequently asked questions (CRS Report R48646). https://www.congress.gov/crs-product/R48646

International Energy Agency. (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai