US data centres already consume more electricity than the entire nation of Pakistan.
The IEA projects that figure hits 426 TWh by 2030 — a 133% increase in six years. New data centres go up in 18–24 months. New power generation takes 3–5 years. That gap is structural.
One county in Virginia eats 26% of a state's grid
In Virginia — the world's largest data centre market — data centres consume 26% of the state's total electricity. Utilities say capacity is being absorbed faster than generation can be built. Texas, Ireland, and the Netherlands face the same wall.
AI inference is the less obvious driver. Every query, every generated image, every autocomplete requires compute, and usage compounds daily. Meta, Amazon, Alphabet, and Microsoft committed over $320 billion in AI infrastructure spending in 2025 alone — not research budgets, but industrial-scale construction.
$320 billion committed in a single year. That's more than the entire annual GDP of Portugal.
Hyperscalers stopped waiting for the grid
Microsoft restarted a reactor at Three Mile Island. Amazon invested in small modular reactor technology. Google is buying geothermal power at scale. The message is unambiguous: the existing grid cannot keep pace.
Individual chips get more efficient every generation, but total consumption keeps climbing. Cheaper AI gets used for more things — efficiency expands demand rather than containing it.
Microsoft literally restarted Three Mile Island to power its AI. That's not a metaphor — it's a grid strategy.
At current trajectories, data centres reach 6–8% of US electricity by 2030, making computing infrastructure one of the largest single demand categories in the country — ahead of steel, cement, and aluminium manufacturing combined.