“When computation becomes cognition, and energy flows like thought, the age of Artificial Superintelligence will no longer be a race—it will be a revelation.”

As the Financial Times Asia front page of August 7, 2025, highlights, the world’s frenzied build-out of AI capacity—led by hyperscalers like Microsoft, Amazon, Meta, and Nvidia—has run into a planet-sized problem. We are depleting the very resources that make AI possible:

· Gigawatts of uninterrupted power for GPUs and TPUs

· Billions of gallons of purified water for chip cooling

· Thousands of acres of land for mega-data centres

· Rare materials for chip fabrication and superconductor systems

Global Operating Units (GOUs)—once the invisible engines of cloud economies—are now front and centre in geopolitical calculations. They consume between 50 and 100 MW per site, with projections suggesting global AI compute demand will exceed 500 GW by 2030. That’s equivalent to half of India’s current total power generation capacity. Water demand for cooling? Between 5 and 15 million litres per day per facility. Land? Upwards of 200 acres per hyperscale campus.

This trajectory is unsustainable.

Rewriting the Equation: Why Supply Cannot Keep Up

1. Electricity: Green power remains intermittent. Fossil power contradicts decarbonisation.........

© Free Press Journal