The price of AI does not include what the surrounding communities pay.
Literal meaning: Costs that are not reflected in the price of a transaction but are borne by a third party — typically the environment, society, or specific communities. In AI: the gap between what companies pay for their infrastructure and what that infrastructure actually costs the world.
Origin: The concept of externalities was formalised by economist Arthur Pigou in the 1920s. Applied to AI, it was brought into sharp focus by Kate Crawford’s Atlas of AI (Yale University Press, 2021), which systematically documented the environmental, labour, and social costs absent from AI’s price tags.
The gap between what AI companies pay for their infrastructure and what that infrastructure actually costs — the difference borne by communities, not shareholders.
The Appeal: Externalisation is not an aberration — it is how markets function when costs are not regulated. Companies that internalise all environmental costs price themselves out of competition. The economic logic is coherent even when the outcome is damaging. This is why the concept requires regulatory response rather than only corporate responsibility.
The Friction: At AI’s scale, externalized costs are enormous in absolute terms. Sacrifice Zones exist because it is cheaper to locate infrastructure near communities with limited political capacity to resist. Greenwashing is partly a response to the reputational risk of visible externalisation: the sustainability pledge manages the optics of costs that have not actually been absorbed. Crawford’s Atlas of AI makes the argument directly: the apparent cheapness of AI services depends on costs being borne by others.
Why This Matters: Externalized costs make visible the structural relationship between price and harm. The AI service that appears free is internalising only a fraction of its actual cost. Once you know to ask who is paying the rest, the market price of a technology becomes one figure among several — and not the most important one.
Related terms: AI Energy Consumption · Carbon Footprint of AI · Sacrifice Zones · Digital Colonialism · Greenwashing · Data Center Water Consumption · Inference Cost · Training Run
Read more:
- Atlas of AI — Crawford, K. (2021). Yale University Press
- The growing energy footprint of artificial intelligence — de Vries, A. (2023). Joule