Every question you ask an AI model comes with three bills: the subscription you pay, the emissions in the air, and the water from someone’s river.
This cluster maps the material reality behind digital abstraction. The digital economy has a body — it consumes water, occupies land, and draws power. These costs are real. They are just not visible from the interface. And the body has owners: when a company reopens a nuclear plant to feed its data centres, a decision that used to belong to a parliament is taken in a boardroom, and a country’s power supply becomes part of someone’s supply chain.
The question this cluster asks: whose resources are powering your query?
Terms in this cluster
AI Energy Consumption
Data centres consume as much electricity as mid-sized cities. The demand is accelerating. The grid is not keeping up.
Carbon Footprint of AI
CO2 emissions per query, per training run, per model. The numbers exist. They are rarely disclosed in full.
Training Run
Building a model consumes more energy than most households use in years. It happens once — and then again, and again.
Inference Cost
Every question has a bill. Small per query. Enormous in aggregate. Almost never disclosed.
Greenwashing
The sustainability pledge published the same week as the data centre expansion announcement.
Data Center Water Consumption
Cooling requires water. Enormous amounts. Often drawn from regions where water is already scarce.
Nuclear Renaissance
Tech companies reopening nuclear plants. Not because governments asked — because AI needs the electricity.
Digital Colonialism
Data centres in the Global South consuming local energy without local benefit. Your cloud dries their river.
Sacrifice Zones
Areas bearing disproportionate environmental burdens of digital infrastructure. Someone lives next to the data centre.
Externalized Costs
Environmental burdens transferred from companies to communities. The price of AI does not include what communities pay.