SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-training-carbon-factors

Why can equal AI training compute produce different carbon emissions?

Hardware, data-center efficiency, and electricity mix can differ.

Start playing free

  1. Parameter names change emissions
  2. Only model accuracy matters
  3. Infrastructure and location matter ✔

Why: The same amount of computation can run on different processors and facilities. Emissions also depend on when and where electricity is produced, so compute alone is not carbon.

Huh, didn't know that: Sparse models can use far less energy than dense models with similar parameter counts.

The dad joke
The model moved closer to clean power and brightened its carbon report.
Like this card?
Play the game

References

Answer: Infrastructure and location matter. The same amount of computation can run on different processors and facilities. Emissions also depend on when and where electricity is produced, so compute alone is not carbon.

  1. Carbon Emissions and Large Neural Network Training arxiv.org/abs/2104.10350
    Geographic location matters for ML workload scheduling since the fraction of carbon-free energy and resulting CO2e vary
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceWith early stopping, when do you end training?Artificial IntelligenceWhich algorithm makes gradient descent practical for multilayer neural networks?Artificial IntelligenceWhat happens to layers whose training gradients vanish?

That's stuff everyone should know.